{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999841614267386,
  "eval_steps": 500,
  "global_step": 15784,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 6.4256110191345215,
      "learning_rate": 2.109704641350211e-08,
      "loss": 1.1779,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.354576587677002,
      "learning_rate": 4.219409282700422e-08,
      "loss": 1.214,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.824097633361816,
      "learning_rate": 6.329113924050633e-08,
      "loss": 1.1802,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.9395294189453125,
      "learning_rate": 8.438818565400844e-08,
      "loss": 1.1571,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.668644905090332,
      "learning_rate": 1.0548523206751055e-07,
      "loss": 1.1222,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.907066822052002,
      "learning_rate": 1.2658227848101266e-07,
      "loss": 1.2215,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.276373863220215,
      "learning_rate": 1.4767932489451477e-07,
      "loss": 1.2274,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.175144195556641,
      "learning_rate": 1.6877637130801689e-07,
      "loss": 1.1467,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.861614227294922,
      "learning_rate": 1.89873417721519e-07,
      "loss": 1.168,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.300581932067871,
      "learning_rate": 2.109704641350211e-07,
      "loss": 1.1902,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.374689102172852,
      "learning_rate": 2.3206751054852324e-07,
      "loss": 1.1582,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.39311408996582,
      "learning_rate": 2.5316455696202533e-07,
      "loss": 1.1747,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.60832405090332,
      "learning_rate": 2.7426160337552746e-07,
      "loss": 1.1727,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.393230438232422,
      "learning_rate": 2.9535864978902955e-07,
      "loss": 1.1668,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.010373592376709,
      "learning_rate": 3.164556962025317e-07,
      "loss": 1.1542,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.905627727508545,
      "learning_rate": 3.3755274261603377e-07,
      "loss": 1.2184,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.977834224700928,
      "learning_rate": 3.586497890295359e-07,
      "loss": 1.2015,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.738651275634766,
      "learning_rate": 3.79746835443038e-07,
      "loss": 1.2082,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.820187091827393,
      "learning_rate": 4.0084388185654013e-07,
      "loss": 1.1105,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.201944828033447,
      "learning_rate": 4.219409282700422e-07,
      "loss": 1.1601,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.043580532073975,
      "learning_rate": 4.4303797468354435e-07,
      "loss": 1.1457,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.428764820098877,
      "learning_rate": 4.641350210970465e-07,
      "loss": 1.1577,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.656769275665283,
      "learning_rate": 4.852320675105486e-07,
      "loss": 1.0191,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.203961372375488,
      "learning_rate": 5.063291139240507e-07,
      "loss": 1.1022,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.214546203613281,
      "learning_rate": 5.274261603375528e-07,
      "loss": 1.1161,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.631003379821777,
      "learning_rate": 5.485232067510549e-07,
      "loss": 1.1692,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.013113975524902,
      "learning_rate": 5.69620253164557e-07,
      "loss": 1.123,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.746349334716797,
      "learning_rate": 5.907172995780591e-07,
      "loss": 1.059,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.102616310119629,
      "learning_rate": 6.118143459915613e-07,
      "loss": 1.0429,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.59443998336792,
      "learning_rate": 6.329113924050634e-07,
      "loss": 1.0285,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.017740726470947,
      "learning_rate": 6.540084388185656e-07,
      "loss": 0.9746,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.804844856262207,
      "learning_rate": 6.751054852320675e-07,
      "loss": 1.0295,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.023899078369141,
      "learning_rate": 6.962025316455696e-07,
      "loss": 0.9803,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.57771635055542,
      "learning_rate": 7.172995780590718e-07,
      "loss": 1.0228,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.537170886993408,
      "learning_rate": 7.383966244725739e-07,
      "loss": 0.9385,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.388979196548462,
      "learning_rate": 7.59493670886076e-07,
      "loss": 0.9712,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.173189878463745,
      "learning_rate": 7.805907172995782e-07,
      "loss": 0.9959,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2532761096954346,
      "learning_rate": 8.016877637130803e-07,
      "loss": 0.8836,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.966832160949707,
      "learning_rate": 8.227848101265823e-07,
      "loss": 0.9153,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.814666748046875,
      "learning_rate": 8.438818565400844e-07,
      "loss": 0.8867,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.621326446533203,
      "learning_rate": 8.649789029535865e-07,
      "loss": 0.9165,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1372127532958984,
      "learning_rate": 8.860759493670887e-07,
      "loss": 0.8869,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.641986608505249,
      "learning_rate": 9.071729957805908e-07,
      "loss": 0.831,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5991573333740234,
      "learning_rate": 9.28270042194093e-07,
      "loss": 0.837,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.680292844772339,
      "learning_rate": 9.493670886075951e-07,
      "loss": 0.7899,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4318466186523438,
      "learning_rate": 9.704641350210971e-07,
      "loss": 0.8298,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0200953483581543,
      "learning_rate": 9.915611814345991e-07,
      "loss": 0.7765,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9560511112213135,
      "learning_rate": 1.0126582278481013e-06,
      "loss": 0.8225,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1863861083984375,
      "learning_rate": 1.0337552742616035e-06,
      "loss": 0.8678,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9763541221618652,
      "learning_rate": 1.0548523206751057e-06,
      "loss": 0.8258,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8083155155181885,
      "learning_rate": 1.0759493670886077e-06,
      "loss": 0.7827,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7802972793579102,
      "learning_rate": 1.0970464135021099e-06,
      "loss": 0.8143,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8700745105743408,
      "learning_rate": 1.1181434599156118e-06,
      "loss": 0.85,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.676476240158081,
      "learning_rate": 1.139240506329114e-06,
      "loss": 0.805,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8485150337219238,
      "learning_rate": 1.1603375527426162e-06,
      "loss": 0.7847,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.823886513710022,
      "learning_rate": 1.1814345991561182e-06,
      "loss": 0.7407,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.807379961013794,
      "learning_rate": 1.2025316455696204e-06,
      "loss": 0.7688,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6759995222091675,
      "learning_rate": 1.2236286919831226e-06,
      "loss": 0.8022,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8627663850784302,
      "learning_rate": 1.2447257383966246e-06,
      "loss": 0.8492,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8799936771392822,
      "learning_rate": 1.2658227848101267e-06,
      "loss": 0.7646,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7080084085464478,
      "learning_rate": 1.286919831223629e-06,
      "loss": 0.7187,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6503185033798218,
      "learning_rate": 1.3080168776371311e-06,
      "loss": 0.7496,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.507122278213501,
      "learning_rate": 1.3291139240506329e-06,
      "loss": 0.74,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5223617553710938,
      "learning_rate": 1.350210970464135e-06,
      "loss": 0.7642,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4301729202270508,
      "learning_rate": 1.371308016877637e-06,
      "loss": 0.7188,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5367064476013184,
      "learning_rate": 1.3924050632911392e-06,
      "loss": 0.7353,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5191432237625122,
      "learning_rate": 1.4135021097046414e-06,
      "loss": 0.7382,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6111494302749634,
      "learning_rate": 1.4345991561181436e-06,
      "loss": 0.7471,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3847888708114624,
      "learning_rate": 1.4556962025316456e-06,
      "loss": 0.7138,
      "step": 69
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.604929804801941,
      "learning_rate": 1.4767932489451478e-06,
      "loss": 0.8025,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4901641607284546,
      "learning_rate": 1.49789029535865e-06,
      "loss": 0.7477,
      "step": 71
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4785903692245483,
      "learning_rate": 1.518987341772152e-06,
      "loss": 0.7408,
      "step": 72
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3478227853775024,
      "learning_rate": 1.5400843881856542e-06,
      "loss": 0.6841,
      "step": 73
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3157073259353638,
      "learning_rate": 1.5611814345991563e-06,
      "loss": 0.6798,
      "step": 74
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.49278724193573,
      "learning_rate": 1.5822784810126585e-06,
      "loss": 0.7075,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2032966613769531,
      "learning_rate": 1.6033755274261605e-06,
      "loss": 0.7001,
      "step": 76
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3113423585891724,
      "learning_rate": 1.6244725738396625e-06,
      "loss": 0.684,
      "step": 77
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.49677574634552,
      "learning_rate": 1.6455696202531647e-06,
      "loss": 0.7296,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.48220694065094,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.7797,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3708736896514893,
      "learning_rate": 1.6877637130801689e-06,
      "loss": 0.7261,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4457319974899292,
      "learning_rate": 1.708860759493671e-06,
      "loss": 0.7705,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2591625452041626,
      "learning_rate": 1.729957805907173e-06,
      "loss": 0.6925,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3564741611480713,
      "learning_rate": 1.7510548523206752e-06,
      "loss": 0.7613,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3493813276290894,
      "learning_rate": 1.7721518987341774e-06,
      "loss": 0.6696,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2987971305847168,
      "learning_rate": 1.7932489451476796e-06,
      "loss": 0.6874,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4315829277038574,
      "learning_rate": 1.8143459915611816e-06,
      "loss": 0.7451,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3597711324691772,
      "learning_rate": 1.8354430379746838e-06,
      "loss": 0.6905,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.313363790512085,
      "learning_rate": 1.856540084388186e-06,
      "loss": 0.6986,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.448975920677185,
      "learning_rate": 1.877637130801688e-06,
      "loss": 0.7411,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4621657133102417,
      "learning_rate": 1.8987341772151901e-06,
      "loss": 0.7285,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3639302253723145,
      "learning_rate": 1.919831223628692e-06,
      "loss": 0.6993,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2497588396072388,
      "learning_rate": 1.9409282700421943e-06,
      "loss": 0.7117,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3762421607971191,
      "learning_rate": 1.9620253164556965e-06,
      "loss": 0.7167,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2534737586975098,
      "learning_rate": 1.9831223628691982e-06,
      "loss": 0.6592,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3847312927246094,
      "learning_rate": 2.0042194092827004e-06,
      "loss": 0.7365,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3049088716506958,
      "learning_rate": 2.0253164556962026e-06,
      "loss": 0.6969,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3645730018615723,
      "learning_rate": 2.046413502109705e-06,
      "loss": 0.5954,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2779875993728638,
      "learning_rate": 2.067510548523207e-06,
      "loss": 0.6862,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3105217218399048,
      "learning_rate": 2.088607594936709e-06,
      "loss": 0.6493,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3567004203796387,
      "learning_rate": 2.1097046413502114e-06,
      "loss": 0.7487,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3794139623641968,
      "learning_rate": 2.130801687763713e-06,
      "loss": 0.6479,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2701658010482788,
      "learning_rate": 2.1518987341772153e-06,
      "loss": 0.7033,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2012029886245728,
      "learning_rate": 2.1729957805907175e-06,
      "loss": 0.6878,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.284591555595398,
      "learning_rate": 2.1940928270042197e-06,
      "loss": 0.6745,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3748619556427002,
      "learning_rate": 2.2151898734177215e-06,
      "loss": 0.6969,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.363826870918274,
      "learning_rate": 2.2362869198312237e-06,
      "loss": 0.6566,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.235390305519104,
      "learning_rate": 2.257383966244726e-06,
      "loss": 0.6597,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4955501556396484,
      "learning_rate": 2.278481012658228e-06,
      "loss": 0.7158,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3154562711715698,
      "learning_rate": 2.2995780590717302e-06,
      "loss": 0.6884,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2907527685165405,
      "learning_rate": 2.3206751054852324e-06,
      "loss": 0.6706,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4128575325012207,
      "learning_rate": 2.341772151898734e-06,
      "loss": 0.6852,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2075326442718506,
      "learning_rate": 2.3628691983122364e-06,
      "loss": 0.6213,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3743549585342407,
      "learning_rate": 2.3839662447257386e-06,
      "loss": 0.6632,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2332968711853027,
      "learning_rate": 2.4050632911392408e-06,
      "loss": 0.6708,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2831799983978271,
      "learning_rate": 2.426160337552743e-06,
      "loss": 0.6561,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2498950958251953,
      "learning_rate": 2.447257383966245e-06,
      "loss": 0.6214,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3748935461044312,
      "learning_rate": 2.4683544303797473e-06,
      "loss": 0.7216,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.389044165611267,
      "learning_rate": 2.489451476793249e-06,
      "loss": 0.6933,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2250936031341553,
      "learning_rate": 2.5105485232067513e-06,
      "loss": 0.6555,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3902949094772339,
      "learning_rate": 2.5316455696202535e-06,
      "loss": 0.6963,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1964465379714966,
      "learning_rate": 2.5527426160337553e-06,
      "loss": 0.6963,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2163293361663818,
      "learning_rate": 2.573839662447258e-06,
      "loss": 0.6337,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2851331233978271,
      "learning_rate": 2.5949367088607596e-06,
      "loss": 0.6371,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2649273872375488,
      "learning_rate": 2.6160337552742622e-06,
      "loss": 0.6451,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3221467733383179,
      "learning_rate": 2.637130801687764e-06,
      "loss": 0.7405,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2666411399841309,
      "learning_rate": 2.6582278481012658e-06,
      "loss": 0.6165,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3712652921676636,
      "learning_rate": 2.679324894514768e-06,
      "loss": 0.6694,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2705154418945312,
      "learning_rate": 2.70042194092827e-06,
      "loss": 0.6234,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2879364490509033,
      "learning_rate": 2.7215189873417724e-06,
      "loss": 0.6715,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1963900327682495,
      "learning_rate": 2.742616033755274e-06,
      "loss": 0.6046,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3172811269760132,
      "learning_rate": 2.7637130801687767e-06,
      "loss": 0.5856,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3311567306518555,
      "learning_rate": 2.7848101265822785e-06,
      "loss": 0.7124,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2794066667556763,
      "learning_rate": 2.805907172995781e-06,
      "loss": 0.6716,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3711148500442505,
      "learning_rate": 2.827004219409283e-06,
      "loss": 0.6892,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2461341619491577,
      "learning_rate": 2.848101265822785e-06,
      "loss": 0.616,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2201720476150513,
      "learning_rate": 2.8691983122362873e-06,
      "loss": 0.6604,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.288766860961914,
      "learning_rate": 2.8902953586497895e-06,
      "loss": 0.7103,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.404861330986023,
      "learning_rate": 2.9113924050632912e-06,
      "loss": 0.646,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3452450037002563,
      "learning_rate": 2.932489451476794e-06,
      "loss": 0.65,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.358460545539856,
      "learning_rate": 2.9535864978902956e-06,
      "loss": 0.663,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4406980276107788,
      "learning_rate": 2.9746835443037974e-06,
      "loss": 0.646,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2622942924499512,
      "learning_rate": 2.9957805907173e-06,
      "loss": 0.6549,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2825264930725098,
      "learning_rate": 3.0168776371308017e-06,
      "loss": 0.6701,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2501925230026245,
      "learning_rate": 3.037974683544304e-06,
      "loss": 0.6562,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2910652160644531,
      "learning_rate": 3.059071729957806e-06,
      "loss": 0.702,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4148486852645874,
      "learning_rate": 3.0801687763713083e-06,
      "loss": 0.6481,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3893300294876099,
      "learning_rate": 3.10126582278481e-06,
      "loss": 0.6627,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.348795771598816,
      "learning_rate": 3.1223628691983127e-06,
      "loss": 0.6378,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1906160116195679,
      "learning_rate": 3.1434599156118145e-06,
      "loss": 0.6098,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3667880296707153,
      "learning_rate": 3.164556962025317e-06,
      "loss": 0.6956,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.290480136871338,
      "learning_rate": 3.185654008438819e-06,
      "loss": 0.6724,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2710357904434204,
      "learning_rate": 3.206751054852321e-06,
      "loss": 0.653,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3685747385025024,
      "learning_rate": 3.2278481012658232e-06,
      "loss": 0.6718,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4111366271972656,
      "learning_rate": 3.248945147679325e-06,
      "loss": 0.6845,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.438011646270752,
      "learning_rate": 3.270042194092827e-06,
      "loss": 0.6813,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1767878532409668,
      "learning_rate": 3.2911392405063294e-06,
      "loss": 0.6103,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2970857620239258,
      "learning_rate": 3.3122362869198316e-06,
      "loss": 0.6954,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3606257438659668,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.6547,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2998062372207642,
      "learning_rate": 3.354430379746836e-06,
      "loss": 0.6103,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3717045783996582,
      "learning_rate": 3.3755274261603377e-06,
      "loss": 0.5959,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2737528085708618,
      "learning_rate": 3.39662447257384e-06,
      "loss": 0.7052,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1850390434265137,
      "learning_rate": 3.417721518987342e-06,
      "loss": 0.6558,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4002028703689575,
      "learning_rate": 3.4388185654008443e-06,
      "loss": 0.6718,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2701243162155151,
      "learning_rate": 3.459915611814346e-06,
      "loss": 0.6182,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2225536108016968,
      "learning_rate": 3.4810126582278487e-06,
      "loss": 0.6627,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2968800067901611,
      "learning_rate": 3.5021097046413504e-06,
      "loss": 0.6921,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2677010297775269,
      "learning_rate": 3.523206751054853e-06,
      "loss": 0.6065,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2947274446487427,
      "learning_rate": 3.544303797468355e-06,
      "loss": 0.654,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.283827543258667,
      "learning_rate": 3.5654008438818566e-06,
      "loss": 0.619,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2735494375228882,
      "learning_rate": 3.586497890295359e-06,
      "loss": 0.6638,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.306062936782837,
      "learning_rate": 3.607594936708861e-06,
      "loss": 0.6402,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.271420955657959,
      "learning_rate": 3.628691983122363e-06,
      "loss": 0.7022,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3412617444992065,
      "learning_rate": 3.649789029535865e-06,
      "loss": 0.6584,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2950142621994019,
      "learning_rate": 3.6708860759493675e-06,
      "loss": 0.6655,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3346741199493408,
      "learning_rate": 3.6919831223628693e-06,
      "loss": 0.6208,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2691746950149536,
      "learning_rate": 3.713080168776372e-06,
      "loss": 0.6727,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.215303659439087,
      "learning_rate": 3.7341772151898737e-06,
      "loss": 0.636,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4336482286453247,
      "learning_rate": 3.755274261603376e-06,
      "loss": 0.6203,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.29073166847229,
      "learning_rate": 3.776371308016878e-06,
      "loss": 0.6743,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3532391786575317,
      "learning_rate": 3.7974683544303802e-06,
      "loss": 0.6038,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4867761135101318,
      "learning_rate": 3.818565400843882e-06,
      "loss": 0.6562,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2906378507614136,
      "learning_rate": 3.839662447257384e-06,
      "loss": 0.6336,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2839657068252563,
      "learning_rate": 3.860759493670886e-06,
      "loss": 0.6719,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3496785163879395,
      "learning_rate": 3.8818565400843886e-06,
      "loss": 0.6481,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.404876708984375,
      "learning_rate": 3.902953586497891e-06,
      "loss": 0.5807,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4152852296829224,
      "learning_rate": 3.924050632911393e-06,
      "loss": 0.6433,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3215677738189697,
      "learning_rate": 3.945147679324895e-06,
      "loss": 0.6719,
      "step": 187
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.31669282913208,
      "learning_rate": 3.9662447257383965e-06,
      "loss": 0.6771,
      "step": 188
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2074658870697021,
      "learning_rate": 3.9873417721518995e-06,
      "loss": 0.657,
      "step": 189
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.282547116279602,
      "learning_rate": 4.008438818565401e-06,
      "loss": 0.6981,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3060194253921509,
      "learning_rate": 4.029535864978903e-06,
      "loss": 0.6262,
      "step": 191
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2825993299484253,
      "learning_rate": 4.050632911392405e-06,
      "loss": 0.6416,
      "step": 192
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2881138324737549,
      "learning_rate": 4.0717299578059074e-06,
      "loss": 0.5945,
      "step": 193
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2855603694915771,
      "learning_rate": 4.09282700421941e-06,
      "loss": 0.615,
      "step": 194
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3043166399002075,
      "learning_rate": 4.113924050632912e-06,
      "loss": 0.7045,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2501403093338013,
      "learning_rate": 4.135021097046414e-06,
      "loss": 0.6447,
      "step": 196
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2611316442489624,
      "learning_rate": 4.156118143459915e-06,
      "loss": 0.6681,
      "step": 197
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3795578479766846,
      "learning_rate": 4.177215189873418e-06,
      "loss": 0.6826,
      "step": 198
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3322073221206665,
      "learning_rate": 4.19831223628692e-06,
      "loss": 0.6414,
      "step": 199
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2988042831420898,
      "learning_rate": 4.219409282700423e-06,
      "loss": 0.6194,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2681517601013184,
      "learning_rate": 4.240506329113924e-06,
      "loss": 0.6497,
      "step": 201
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4114967584609985,
      "learning_rate": 4.261603375527426e-06,
      "loss": 0.7306,
      "step": 202
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2016258239746094,
      "learning_rate": 4.2827004219409285e-06,
      "loss": 0.6195,
      "step": 203
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4103429317474365,
      "learning_rate": 4.303797468354431e-06,
      "loss": 0.6501,
      "step": 204
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.246969223022461,
      "learning_rate": 4.324894514767933e-06,
      "loss": 0.6464,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2877082824707031,
      "learning_rate": 4.345991561181435e-06,
      "loss": 0.6357,
      "step": 206
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2532278299331665,
      "learning_rate": 4.367088607594937e-06,
      "loss": 0.6386,
      "step": 207
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.39167058467865,
      "learning_rate": 4.3881856540084394e-06,
      "loss": 0.6713,
      "step": 208
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2851932048797607,
      "learning_rate": 4.409282700421942e-06,
      "loss": 0.5946,
      "step": 209
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2754656076431274,
      "learning_rate": 4.430379746835443e-06,
      "loss": 0.66,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3959016799926758,
      "learning_rate": 4.451476793248945e-06,
      "loss": 0.6753,
      "step": 211
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2273962497711182,
      "learning_rate": 4.472573839662447e-06,
      "loss": 0.5892,
      "step": 212
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2749860286712646,
      "learning_rate": 4.4936708860759495e-06,
      "loss": 0.7182,
      "step": 213
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4585115909576416,
      "learning_rate": 4.514767932489452e-06,
      "loss": 0.6285,
      "step": 214
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.291067361831665,
      "learning_rate": 4.535864978902954e-06,
      "loss": 0.6543,
      "step": 215
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2669910192489624,
      "learning_rate": 4.556962025316456e-06,
      "loss": 0.6242,
      "step": 216
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.303594708442688,
      "learning_rate": 4.578059071729958e-06,
      "loss": 0.6184,
      "step": 217
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3243204355239868,
      "learning_rate": 4.5991561181434605e-06,
      "loss": 0.678,
      "step": 218
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2260215282440186,
      "learning_rate": 4.620253164556963e-06,
      "loss": 0.6327,
      "step": 219
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3388763666152954,
      "learning_rate": 4.641350210970465e-06,
      "loss": 0.5952,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4450039863586426,
      "learning_rate": 4.662447257383967e-06,
      "loss": 0.6568,
      "step": 221
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.388580322265625,
      "learning_rate": 4.683544303797468e-06,
      "loss": 0.617,
      "step": 222
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2090303897857666,
      "learning_rate": 4.7046413502109714e-06,
      "loss": 0.6157,
      "step": 223
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.400683879852295,
      "learning_rate": 4.725738396624473e-06,
      "loss": 0.686,
      "step": 224
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3882019519805908,
      "learning_rate": 4.746835443037975e-06,
      "loss": 0.6476,
      "step": 225
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3863544464111328,
      "learning_rate": 4.767932489451477e-06,
      "loss": 0.6183,
      "step": 226
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2990771532058716,
      "learning_rate": 4.789029535864979e-06,
      "loss": 0.5885,
      "step": 227
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2280645370483398,
      "learning_rate": 4.8101265822784815e-06,
      "loss": 0.5813,
      "step": 228
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3123680353164673,
      "learning_rate": 4.831223628691984e-06,
      "loss": 0.6771,
      "step": 229
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.398052453994751,
      "learning_rate": 4.852320675105486e-06,
      "loss": 0.6691,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.208593726158142,
      "learning_rate": 4.873417721518987e-06,
      "loss": 0.5812,
      "step": 231
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3669441938400269,
      "learning_rate": 4.89451476793249e-06,
      "loss": 0.6813,
      "step": 232
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3295658826828003,
      "learning_rate": 4.915611814345992e-06,
      "loss": 0.6214,
      "step": 233
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.269345760345459,
      "learning_rate": 4.936708860759495e-06,
      "loss": 0.5648,
      "step": 234
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.429956078529358,
      "learning_rate": 4.957805907172996e-06,
      "loss": 0.6428,
      "step": 235
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2624046802520752,
      "learning_rate": 4.978902953586498e-06,
      "loss": 0.6186,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4069180488586426,
      "learning_rate": 5e-06,
      "loss": 0.6369,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2838106155395508,
      "learning_rate": 5.021097046413503e-06,
      "loss": 0.6694,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2982817888259888,
      "learning_rate": 5.042194092827004e-06,
      "loss": 0.6414,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2129889726638794,
      "learning_rate": 5.063291139240507e-06,
      "loss": 0.6026,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3048906326293945,
      "learning_rate": 5.084388185654009e-06,
      "loss": 0.6512,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2372359037399292,
      "learning_rate": 5.1054852320675105e-06,
      "loss": 0.5698,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.251413345336914,
      "learning_rate": 5.126582278481013e-06,
      "loss": 0.6061,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2215440273284912,
      "learning_rate": 5.147679324894516e-06,
      "loss": 0.5464,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.246283769607544,
      "learning_rate": 5.168776371308017e-06,
      "loss": 0.618,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2899631261825562,
      "learning_rate": 5.189873417721519e-06,
      "loss": 0.6387,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3374756574630737,
      "learning_rate": 5.2109704641350215e-06,
      "loss": 0.634,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3475295305252075,
      "learning_rate": 5.2320675105485245e-06,
      "loss": 0.6372,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.252197265625,
      "learning_rate": 5.253164556962026e-06,
      "loss": 0.6681,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2391002178192139,
      "learning_rate": 5.274261603375528e-06,
      "loss": 0.6076,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3592095375061035,
      "learning_rate": 5.295358649789029e-06,
      "loss": 0.6513,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2067376375198364,
      "learning_rate": 5.3164556962025316e-06,
      "loss": 0.6251,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1935391426086426,
      "learning_rate": 5.337552742616035e-06,
      "loss": 0.6499,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2630081176757812,
      "learning_rate": 5.358649789029536e-06,
      "loss": 0.6361,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2552322149276733,
      "learning_rate": 5.379746835443038e-06,
      "loss": 0.6477,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2347304821014404,
      "learning_rate": 5.40084388185654e-06,
      "loss": 0.618,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4220587015151978,
      "learning_rate": 5.421940928270043e-06,
      "loss": 0.6588,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.246965765953064,
      "learning_rate": 5.443037974683545e-06,
      "loss": 0.6084,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2736470699310303,
      "learning_rate": 5.464135021097047e-06,
      "loss": 0.6702,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2836354970932007,
      "learning_rate": 5.485232067510548e-06,
      "loss": 0.6246,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2576249837875366,
      "learning_rate": 5.506329113924051e-06,
      "loss": 0.6357,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3883161544799805,
      "learning_rate": 5.5274261603375535e-06,
      "loss": 0.6021,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.266746163368225,
      "learning_rate": 5.548523206751056e-06,
      "loss": 0.5971,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1999036073684692,
      "learning_rate": 5.569620253164557e-06,
      "loss": 0.6817,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2391798496246338,
      "learning_rate": 5.590717299578059e-06,
      "loss": 0.6384,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2688382863998413,
      "learning_rate": 5.611814345991562e-06,
      "loss": 0.6419,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2850573062896729,
      "learning_rate": 5.6329113924050636e-06,
      "loss": 0.6192,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2011271715164185,
      "learning_rate": 5.654008438818566e-06,
      "loss": 0.5714,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2120412588119507,
      "learning_rate": 5.675105485232067e-06,
      "loss": 0.6382,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2102055549621582,
      "learning_rate": 5.69620253164557e-06,
      "loss": 0.6238,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3209874629974365,
      "learning_rate": 5.717299578059072e-06,
      "loss": 0.6885,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2263984680175781,
      "learning_rate": 5.7383966244725745e-06,
      "loss": 0.6304,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.271883487701416,
      "learning_rate": 5.759493670886076e-06,
      "loss": 0.6758,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3089218139648438,
      "learning_rate": 5.780590717299579e-06,
      "loss": 0.6316,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.329040288925171,
      "learning_rate": 5.801687763713081e-06,
      "loss": 0.6756,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2899181842803955,
      "learning_rate": 5.8227848101265824e-06,
      "loss": 0.6439,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1611582040786743,
      "learning_rate": 5.843881856540085e-06,
      "loss": 0.6241,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1877211332321167,
      "learning_rate": 5.864978902953588e-06,
      "loss": 0.5726,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.23735511302948,
      "learning_rate": 5.886075949367089e-06,
      "loss": 0.6491,
      "step": 279
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3182356357574463,
      "learning_rate": 5.907172995780591e-06,
      "loss": 0.6364,
      "step": 280
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4192547798156738,
      "learning_rate": 5.928270042194093e-06,
      "loss": 0.6907,
      "step": 281
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.256334662437439,
      "learning_rate": 5.949367088607595e-06,
      "loss": 0.6639,
      "step": 282
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2886290550231934,
      "learning_rate": 5.970464135021098e-06,
      "loss": 0.6576,
      "step": 283
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1646943092346191,
      "learning_rate": 5.9915611814346e-06,
      "loss": 0.6297,
      "step": 284
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.215549349784851,
      "learning_rate": 6.012658227848101e-06,
      "loss": 0.6089,
      "step": 285
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2174471616744995,
      "learning_rate": 6.0337552742616035e-06,
      "loss": 0.5496,
      "step": 286
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.255016565322876,
      "learning_rate": 6.0548523206751065e-06,
      "loss": 0.6058,
      "step": 287
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2536252737045288,
      "learning_rate": 6.075949367088608e-06,
      "loss": 0.5952,
      "step": 288
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2640748023986816,
      "learning_rate": 6.09704641350211e-06,
      "loss": 0.618,
      "step": 289
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2663012742996216,
      "learning_rate": 6.118143459915612e-06,
      "loss": 0.6525,
      "step": 290
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2259374856948853,
      "learning_rate": 6.139240506329115e-06,
      "loss": 0.6079,
      "step": 291
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3531912565231323,
      "learning_rate": 6.160337552742617e-06,
      "loss": 0.6583,
      "step": 292
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3680455684661865,
      "learning_rate": 6.181434599156119e-06,
      "loss": 0.6359,
      "step": 293
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4130650758743286,
      "learning_rate": 6.20253164556962e-06,
      "loss": 0.64,
      "step": 294
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.283833622932434,
      "learning_rate": 6.223628691983122e-06,
      "loss": 0.6207,
      "step": 295
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3075100183486938,
      "learning_rate": 6.244725738396625e-06,
      "loss": 0.6618,
      "step": 296
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2828034162521362,
      "learning_rate": 6.265822784810128e-06,
      "loss": 0.5909,
      "step": 297
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2394911050796509,
      "learning_rate": 6.286919831223629e-06,
      "loss": 0.6256,
      "step": 298
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3512240648269653,
      "learning_rate": 6.308016877637131e-06,
      "loss": 0.6058,
      "step": 299
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2114907503128052,
      "learning_rate": 6.329113924050634e-06,
      "loss": 0.6713,
      "step": 300
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.235547423362732,
      "learning_rate": 6.3502109704641355e-06,
      "loss": 0.5995,
      "step": 301
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3060510158538818,
      "learning_rate": 6.371308016877638e-06,
      "loss": 0.624,
      "step": 302
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2200214862823486,
      "learning_rate": 6.392405063291139e-06,
      "loss": 0.6165,
      "step": 303
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.299473524093628,
      "learning_rate": 6.413502109704642e-06,
      "loss": 0.6672,
      "step": 304
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1858652830123901,
      "learning_rate": 6.434599156118144e-06,
      "loss": 0.6199,
      "step": 305
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2412612438201904,
      "learning_rate": 6.4556962025316464e-06,
      "loss": 0.6092,
      "step": 306
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1821410655975342,
      "learning_rate": 6.476793248945148e-06,
      "loss": 0.6029,
      "step": 307
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3090671300888062,
      "learning_rate": 6.49789029535865e-06,
      "loss": 0.6865,
      "step": 308
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2623692750930786,
      "learning_rate": 6.518987341772153e-06,
      "loss": 0.6331,
      "step": 309
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2148505449295044,
      "learning_rate": 6.540084388185654e-06,
      "loss": 0.5844,
      "step": 310
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.206979751586914,
      "learning_rate": 6.5611814345991565e-06,
      "loss": 0.6161,
      "step": 311
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2950704097747803,
      "learning_rate": 6.582278481012659e-06,
      "loss": 0.6101,
      "step": 312
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2553280591964722,
      "learning_rate": 6.603375527426161e-06,
      "loss": 0.6826,
      "step": 313
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.237945795059204,
      "learning_rate": 6.624472573839663e-06,
      "loss": 0.581,
      "step": 314
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.280417799949646,
      "learning_rate": 6.645569620253165e-06,
      "loss": 0.6078,
      "step": 315
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.371483325958252,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.6104,
      "step": 316
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2257425785064697,
      "learning_rate": 6.68776371308017e-06,
      "loss": 0.5975,
      "step": 317
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2869188785552979,
      "learning_rate": 6.708860759493672e-06,
      "loss": 0.6517,
      "step": 318
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.474818229675293,
      "learning_rate": 6.729957805907173e-06,
      "loss": 0.6719,
      "step": 319
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.31566321849823,
      "learning_rate": 6.751054852320675e-06,
      "loss": 0.6426,
      "step": 320
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3478425741195679,
      "learning_rate": 6.772151898734178e-06,
      "loss": 0.6462,
      "step": 321
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1856145858764648,
      "learning_rate": 6.79324894514768e-06,
      "loss": 0.6296,
      "step": 322
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2015944719314575,
      "learning_rate": 6.814345991561182e-06,
      "loss": 0.6179,
      "step": 323
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2188371419906616,
      "learning_rate": 6.835443037974684e-06,
      "loss": 0.6049,
      "step": 324
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.313015103340149,
      "learning_rate": 6.8565400843881855e-06,
      "loss": 0.5847,
      "step": 325
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.424932599067688,
      "learning_rate": 6.8776371308016885e-06,
      "loss": 0.6946,
      "step": 326
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3193198442459106,
      "learning_rate": 6.898734177215191e-06,
      "loss": 0.6464,
      "step": 327
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.247219443321228,
      "learning_rate": 6.919831223628692e-06,
      "loss": 0.6904,
      "step": 328
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3283206224441528,
      "learning_rate": 6.940928270042194e-06,
      "loss": 0.6483,
      "step": 329
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2536122798919678,
      "learning_rate": 6.962025316455697e-06,
      "loss": 0.6327,
      "step": 330
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2731391191482544,
      "learning_rate": 6.9831223628691995e-06,
      "loss": 0.6554,
      "step": 331
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2560125589370728,
      "learning_rate": 7.004219409282701e-06,
      "loss": 0.6123,
      "step": 332
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2068290710449219,
      "learning_rate": 7.025316455696203e-06,
      "loss": 0.6428,
      "step": 333
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1613904237747192,
      "learning_rate": 7.046413502109706e-06,
      "loss": 0.5978,
      "step": 334
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3051648139953613,
      "learning_rate": 7.067510548523207e-06,
      "loss": 0.686,
      "step": 335
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2591476440429688,
      "learning_rate": 7.08860759493671e-06,
      "loss": 0.5891,
      "step": 336
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2479435205459595,
      "learning_rate": 7.109704641350211e-06,
      "loss": 0.6221,
      "step": 337
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.326296091079712,
      "learning_rate": 7.130801687763713e-06,
      "loss": 0.6327,
      "step": 338
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2220957279205322,
      "learning_rate": 7.151898734177216e-06,
      "loss": 0.6364,
      "step": 339
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1175003051757812,
      "learning_rate": 7.172995780590718e-06,
      "loss": 0.5575,
      "step": 340
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2847028970718384,
      "learning_rate": 7.19409282700422e-06,
      "loss": 0.5732,
      "step": 341
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.261705756187439,
      "learning_rate": 7.215189873417722e-06,
      "loss": 0.6702,
      "step": 342
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3592685461044312,
      "learning_rate": 7.236286919831225e-06,
      "loss": 0.6816,
      "step": 343
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1502933502197266,
      "learning_rate": 7.257383966244726e-06,
      "loss": 0.6092,
      "step": 344
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1761858463287354,
      "learning_rate": 7.2784810126582285e-06,
      "loss": 0.6094,
      "step": 345
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.249475121498108,
      "learning_rate": 7.29957805907173e-06,
      "loss": 0.6694,
      "step": 346
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2194502353668213,
      "learning_rate": 7.320675105485233e-06,
      "loss": 0.5863,
      "step": 347
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2675786018371582,
      "learning_rate": 7.341772151898735e-06,
      "loss": 0.6129,
      "step": 348
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1856513023376465,
      "learning_rate": 7.362869198312237e-06,
      "loss": 0.5751,
      "step": 349
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2041898965835571,
      "learning_rate": 7.3839662447257386e-06,
      "loss": 0.6197,
      "step": 350
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4077736139297485,
      "learning_rate": 7.405063291139241e-06,
      "loss": 0.6282,
      "step": 351
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.231197714805603,
      "learning_rate": 7.426160337552744e-06,
      "loss": 0.6035,
      "step": 352
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1258540153503418,
      "learning_rate": 7.447257383966245e-06,
      "loss": 0.613,
      "step": 353
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.286897897720337,
      "learning_rate": 7.468354430379747e-06,
      "loss": 0.5775,
      "step": 354
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.216144323348999,
      "learning_rate": 7.4894514767932495e-06,
      "loss": 0.6519,
      "step": 355
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3355437517166138,
      "learning_rate": 7.510548523206752e-06,
      "loss": 0.6512,
      "step": 356
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.551064372062683,
      "learning_rate": 7.531645569620254e-06,
      "loss": 0.5391,
      "step": 357
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2588145732879639,
      "learning_rate": 7.552742616033756e-06,
      "loss": 0.6467,
      "step": 358
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2731475830078125,
      "learning_rate": 7.5738396624472574e-06,
      "loss": 0.6195,
      "step": 359
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2335354089736938,
      "learning_rate": 7.5949367088607605e-06,
      "loss": 0.5462,
      "step": 360
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2900036573410034,
      "learning_rate": 7.616033755274263e-06,
      "loss": 0.6777,
      "step": 361
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3332929611206055,
      "learning_rate": 7.637130801687764e-06,
      "loss": 0.5707,
      "step": 362
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2227133512496948,
      "learning_rate": 7.658227848101265e-06,
      "loss": 0.6009,
      "step": 363
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2745919227600098,
      "learning_rate": 7.679324894514768e-06,
      "loss": 0.6533,
      "step": 364
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2688459157943726,
      "learning_rate": 7.700421940928271e-06,
      "loss": 0.6577,
      "step": 365
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0965393781661987,
      "learning_rate": 7.721518987341773e-06,
      "loss": 0.5903,
      "step": 366
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2288157939910889,
      "learning_rate": 7.742616033755274e-06,
      "loss": 0.5953,
      "step": 367
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3503426313400269,
      "learning_rate": 7.763713080168777e-06,
      "loss": 0.6585,
      "step": 368
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1747641563415527,
      "learning_rate": 7.78481012658228e-06,
      "loss": 0.6289,
      "step": 369
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2812758684158325,
      "learning_rate": 7.805907172995782e-06,
      "loss": 0.638,
      "step": 370
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3146843910217285,
      "learning_rate": 7.827004219409283e-06,
      "loss": 0.6224,
      "step": 371
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1805860996246338,
      "learning_rate": 7.848101265822786e-06,
      "loss": 0.5742,
      "step": 372
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3260769844055176,
      "learning_rate": 7.869198312236287e-06,
      "loss": 0.5949,
      "step": 373
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2296415567398071,
      "learning_rate": 7.89029535864979e-06,
      "loss": 0.5776,
      "step": 374
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2580851316452026,
      "learning_rate": 7.911392405063292e-06,
      "loss": 0.6392,
      "step": 375
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1601670980453491,
      "learning_rate": 7.932489451476793e-06,
      "loss": 0.5467,
      "step": 376
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3382383584976196,
      "learning_rate": 7.953586497890296e-06,
      "loss": 0.6324,
      "step": 377
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2548341751098633,
      "learning_rate": 7.974683544303799e-06,
      "loss": 0.6006,
      "step": 378
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.270756721496582,
      "learning_rate": 7.9957805907173e-06,
      "loss": 0.6084,
      "step": 379
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4024615287780762,
      "learning_rate": 8.016877637130802e-06,
      "loss": 0.648,
      "step": 380
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0922244787216187,
      "learning_rate": 8.037974683544305e-06,
      "loss": 0.5659,
      "step": 381
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2462538480758667,
      "learning_rate": 8.059071729957806e-06,
      "loss": 0.6239,
      "step": 382
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3170617818832397,
      "learning_rate": 8.080168776371309e-06,
      "loss": 0.6455,
      "step": 383
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2223594188690186,
      "learning_rate": 8.10126582278481e-06,
      "loss": 0.6454,
      "step": 384
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2347493171691895,
      "learning_rate": 8.122362869198312e-06,
      "loss": 0.5671,
      "step": 385
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1949635744094849,
      "learning_rate": 8.143459915611815e-06,
      "loss": 0.635,
      "step": 386
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.260286569595337,
      "learning_rate": 8.164556962025318e-06,
      "loss": 0.6135,
      "step": 387
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3054084777832031,
      "learning_rate": 8.18565400843882e-06,
      "loss": 0.6299,
      "step": 388
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4015529155731201,
      "learning_rate": 8.20675105485232e-06,
      "loss": 0.6197,
      "step": 389
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.281866431236267,
      "learning_rate": 8.227848101265824e-06,
      "loss": 0.6863,
      "step": 390
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.171478509902954,
      "learning_rate": 8.248945147679327e-06,
      "loss": 0.5865,
      "step": 391
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2968735694885254,
      "learning_rate": 8.270042194092828e-06,
      "loss": 0.6147,
      "step": 392
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1304808855056763,
      "learning_rate": 8.29113924050633e-06,
      "loss": 0.5535,
      "step": 393
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1816270351409912,
      "learning_rate": 8.31223628691983e-06,
      "loss": 0.6069,
      "step": 394
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.217431902885437,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.5988,
      "step": 395
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1394872665405273,
      "learning_rate": 8.354430379746837e-06,
      "loss": 0.5513,
      "step": 396
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.251174807548523,
      "learning_rate": 8.375527426160338e-06,
      "loss": 0.6487,
      "step": 397
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2142462730407715,
      "learning_rate": 8.39662447257384e-06,
      "loss": 0.6135,
      "step": 398
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1851967573165894,
      "learning_rate": 8.417721518987342e-06,
      "loss": 0.5548,
      "step": 399
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1517951488494873,
      "learning_rate": 8.438818565400846e-06,
      "loss": 0.5771,
      "step": 400
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4722862243652344,
      "learning_rate": 8.459915611814347e-06,
      "loss": 0.6222,
      "step": 401
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2671726942062378,
      "learning_rate": 8.481012658227848e-06,
      "loss": 0.592,
      "step": 402
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3089793920516968,
      "learning_rate": 8.502109704641351e-06,
      "loss": 0.6425,
      "step": 403
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.251240849494934,
      "learning_rate": 8.523206751054853e-06,
      "loss": 0.585,
      "step": 404
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3644077777862549,
      "learning_rate": 8.544303797468356e-06,
      "loss": 0.6152,
      "step": 405
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3811988830566406,
      "learning_rate": 8.565400843881857e-06,
      "loss": 0.647,
      "step": 406
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2296816110610962,
      "learning_rate": 8.586497890295358e-06,
      "loss": 0.6464,
      "step": 407
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3101154565811157,
      "learning_rate": 8.607594936708861e-06,
      "loss": 0.5907,
      "step": 408
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3449515104293823,
      "learning_rate": 8.628691983122364e-06,
      "loss": 0.6398,
      "step": 409
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2597445249557495,
      "learning_rate": 8.649789029535866e-06,
      "loss": 0.6462,
      "step": 410
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.109322428703308,
      "learning_rate": 8.670886075949367e-06,
      "loss": 0.6258,
      "step": 411
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2519370317459106,
      "learning_rate": 8.69198312236287e-06,
      "loss": 0.7079,
      "step": 412
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2595500946044922,
      "learning_rate": 8.713080168776371e-06,
      "loss": 0.6475,
      "step": 413
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.129270076751709,
      "learning_rate": 8.734177215189874e-06,
      "loss": 0.5764,
      "step": 414
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.285945177078247,
      "learning_rate": 8.755274261603376e-06,
      "loss": 0.6278,
      "step": 415
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2584099769592285,
      "learning_rate": 8.776371308016879e-06,
      "loss": 0.6143,
      "step": 416
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2111893892288208,
      "learning_rate": 8.79746835443038e-06,
      "loss": 0.6285,
      "step": 417
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1318415403366089,
      "learning_rate": 8.818565400843883e-06,
      "loss": 0.6495,
      "step": 418
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1128448247909546,
      "learning_rate": 8.839662447257385e-06,
      "loss": 0.5995,
      "step": 419
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1388238668441772,
      "learning_rate": 8.860759493670886e-06,
      "loss": 0.6209,
      "step": 420
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.27338445186615,
      "learning_rate": 8.881856540084389e-06,
      "loss": 0.6759,
      "step": 421
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.169218897819519,
      "learning_rate": 8.90295358649789e-06,
      "loss": 0.5423,
      "step": 422
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2038993835449219,
      "learning_rate": 8.924050632911393e-06,
      "loss": 0.5966,
      "step": 423
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1410913467407227,
      "learning_rate": 8.945147679324895e-06,
      "loss": 0.5852,
      "step": 424
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1868853569030762,
      "learning_rate": 8.966244725738398e-06,
      "loss": 0.5782,
      "step": 425
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2283443212509155,
      "learning_rate": 8.987341772151899e-06,
      "loss": 0.5799,
      "step": 426
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3854525089263916,
      "learning_rate": 9.008438818565402e-06,
      "loss": 0.6789,
      "step": 427
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2753512859344482,
      "learning_rate": 9.029535864978903e-06,
      "loss": 0.5774,
      "step": 428
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3836292028427124,
      "learning_rate": 9.050632911392407e-06,
      "loss": 0.6171,
      "step": 429
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2557188272476196,
      "learning_rate": 9.071729957805908e-06,
      "loss": 0.6662,
      "step": 430
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2048022747039795,
      "learning_rate": 9.09282700421941e-06,
      "loss": 0.6439,
      "step": 431
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.355612874031067,
      "learning_rate": 9.113924050632912e-06,
      "loss": 0.6367,
      "step": 432
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.238109827041626,
      "learning_rate": 9.135021097046414e-06,
      "loss": 0.5903,
      "step": 433
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.22252357006073,
      "learning_rate": 9.156118143459917e-06,
      "loss": 0.6198,
      "step": 434
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4150843620300293,
      "learning_rate": 9.177215189873418e-06,
      "loss": 0.6179,
      "step": 435
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2307286262512207,
      "learning_rate": 9.198312236286921e-06,
      "loss": 0.5735,
      "step": 436
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.193342685699463,
      "learning_rate": 9.219409282700422e-06,
      "loss": 0.6004,
      "step": 437
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.158647894859314,
      "learning_rate": 9.240506329113925e-06,
      "loss": 0.6049,
      "step": 438
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.204749584197998,
      "learning_rate": 9.261603375527427e-06,
      "loss": 0.664,
      "step": 439
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.111306071281433,
      "learning_rate": 9.28270042194093e-06,
      "loss": 0.6079,
      "step": 440
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.299516201019287,
      "learning_rate": 9.303797468354431e-06,
      "loss": 0.6563,
      "step": 441
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3435544967651367,
      "learning_rate": 9.324894514767934e-06,
      "loss": 0.6232,
      "step": 442
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2585128545761108,
      "learning_rate": 9.345991561181435e-06,
      "loss": 0.6193,
      "step": 443
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.248583197593689,
      "learning_rate": 9.367088607594937e-06,
      "loss": 0.6057,
      "step": 444
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1512912511825562,
      "learning_rate": 9.38818565400844e-06,
      "loss": 0.6147,
      "step": 445
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2142647504806519,
      "learning_rate": 9.409282700421943e-06,
      "loss": 0.5901,
      "step": 446
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3141642808914185,
      "learning_rate": 9.430379746835444e-06,
      "loss": 0.6572,
      "step": 447
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.09758722782135,
      "learning_rate": 9.451476793248946e-06,
      "loss": 0.5999,
      "step": 448
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1424005031585693,
      "learning_rate": 9.472573839662449e-06,
      "loss": 0.5772,
      "step": 449
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.171758770942688,
      "learning_rate": 9.49367088607595e-06,
      "loss": 0.6537,
      "step": 450
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0351412296295166,
      "learning_rate": 9.514767932489453e-06,
      "loss": 0.5663,
      "step": 451
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1341289281845093,
      "learning_rate": 9.535864978902954e-06,
      "loss": 0.6281,
      "step": 452
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.115142822265625,
      "learning_rate": 9.556962025316456e-06,
      "loss": 0.5919,
      "step": 453
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2643224000930786,
      "learning_rate": 9.578059071729959e-06,
      "loss": 0.6949,
      "step": 454
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2069917917251587,
      "learning_rate": 9.599156118143462e-06,
      "loss": 0.6319,
      "step": 455
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1984665393829346,
      "learning_rate": 9.620253164556963e-06,
      "loss": 0.6516,
      "step": 456
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4079281091690063,
      "learning_rate": 9.641350210970464e-06,
      "loss": 0.6593,
      "step": 457
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1656159162521362,
      "learning_rate": 9.662447257383967e-06,
      "loss": 0.5894,
      "step": 458
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1576460599899292,
      "learning_rate": 9.68354430379747e-06,
      "loss": 0.5557,
      "step": 459
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2340763807296753,
      "learning_rate": 9.704641350210972e-06,
      "loss": 0.6803,
      "step": 460
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1936951875686646,
      "learning_rate": 9.725738396624473e-06,
      "loss": 0.5841,
      "step": 461
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1302146911621094,
      "learning_rate": 9.746835443037975e-06,
      "loss": 0.5765,
      "step": 462
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3100109100341797,
      "learning_rate": 9.767932489451478e-06,
      "loss": 0.6448,
      "step": 463
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.223763346672058,
      "learning_rate": 9.78902953586498e-06,
      "loss": 0.6135,
      "step": 464
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2290325164794922,
      "learning_rate": 9.810126582278482e-06,
      "loss": 0.6517,
      "step": 465
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3180433511734009,
      "learning_rate": 9.831223628691983e-06,
      "loss": 0.6086,
      "step": 466
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0901906490325928,
      "learning_rate": 9.852320675105486e-06,
      "loss": 0.5328,
      "step": 467
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3592183589935303,
      "learning_rate": 9.87341772151899e-06,
      "loss": 0.6652,
      "step": 468
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3109347820281982,
      "learning_rate": 9.89451476793249e-06,
      "loss": 0.6091,
      "step": 469
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2154662609100342,
      "learning_rate": 9.915611814345992e-06,
      "loss": 0.6194,
      "step": 470
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1480965614318848,
      "learning_rate": 9.936708860759493e-06,
      "loss": 0.6695,
      "step": 471
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3060542345046997,
      "learning_rate": 9.957805907172996e-06,
      "loss": 0.6688,
      "step": 472
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5538403987884521,
      "learning_rate": 9.9789029535865e-06,
      "loss": 0.6385,
      "step": 473
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.187518835067749,
      "learning_rate": 1e-05,
      "loss": 0.6299,
      "step": 474
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.238229751586914,
      "learning_rate": 9.999999894733699e-06,
      "loss": 0.6207,
      "step": 475
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2438807487487793,
      "learning_rate": 9.999999578934793e-06,
      "loss": 0.6678,
      "step": 476
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2306510210037231,
      "learning_rate": 9.9999990526033e-06,
      "loss": 0.5857,
      "step": 477
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2313143014907837,
      "learning_rate": 9.99999831573924e-06,
      "loss": 0.5777,
      "step": 478
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2812029123306274,
      "learning_rate": 9.999997368342644e-06,
      "loss": 0.6478,
      "step": 479
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1557166576385498,
      "learning_rate": 9.999996210413553e-06,
      "loss": 0.6285,
      "step": 480
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1735140085220337,
      "learning_rate": 9.999994841952016e-06,
      "loss": 0.6189,
      "step": 481
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2090628147125244,
      "learning_rate": 9.99999326295809e-06,
      "loss": 0.672,
      "step": 482
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2201446294784546,
      "learning_rate": 9.99999147343184e-06,
      "loss": 0.5718,
      "step": 483
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.196315884590149,
      "learning_rate": 9.999989473373344e-06,
      "loss": 0.5772,
      "step": 484
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.229958176612854,
      "learning_rate": 9.999987262782684e-06,
      "loss": 0.6194,
      "step": 485
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2189887762069702,
      "learning_rate": 9.999984841659955e-06,
      "loss": 0.6379,
      "step": 486
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4202560186386108,
      "learning_rate": 9.999982210005258e-06,
      "loss": 0.6053,
      "step": 487
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1163641214370728,
      "learning_rate": 9.999979367818704e-06,
      "loss": 0.5955,
      "step": 488
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1054067611694336,
      "learning_rate": 9.999976315100412e-06,
      "loss": 0.568,
      "step": 489
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2214596271514893,
      "learning_rate": 9.99997305185051e-06,
      "loss": 0.5809,
      "step": 490
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1755683422088623,
      "learning_rate": 9.999969578069137e-06,
      "loss": 0.6626,
      "step": 491
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.170556664466858,
      "learning_rate": 9.99996589375644e-06,
      "loss": 0.5958,
      "step": 492
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.284775733947754,
      "learning_rate": 9.999961998912573e-06,
      "loss": 0.6206,
      "step": 493
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.216280221939087,
      "learning_rate": 9.999957893537697e-06,
      "loss": 0.6323,
      "step": 494
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2194132804870605,
      "learning_rate": 9.999953577631991e-06,
      "loss": 0.6386,
      "step": 495
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2110742330551147,
      "learning_rate": 9.999949051195631e-06,
      "loss": 0.6262,
      "step": 496
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1123095750808716,
      "learning_rate": 9.999944314228811e-06,
      "loss": 0.59,
      "step": 497
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.190177083015442,
      "learning_rate": 9.99993936673173e-06,
      "loss": 0.5809,
      "step": 498
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1872690916061401,
      "learning_rate": 9.999934208704595e-06,
      "loss": 0.5754,
      "step": 499
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.129944920539856,
      "learning_rate": 9.999928840147624e-06,
      "loss": 0.6183,
      "step": 500
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1830451488494873,
      "learning_rate": 9.999923261061043e-06,
      "loss": 0.6128,
      "step": 501
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.171960711479187,
      "learning_rate": 9.999917471445086e-06,
      "loss": 0.5835,
      "step": 502
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1681584119796753,
      "learning_rate": 9.999911471299998e-06,
      "loss": 0.617,
      "step": 503
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2297478914260864,
      "learning_rate": 9.999905260626033e-06,
      "loss": 0.6365,
      "step": 504
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1890262365341187,
      "learning_rate": 9.99989883942345e-06,
      "loss": 0.6715,
      "step": 505
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2099827527999878,
      "learning_rate": 9.999892207692521e-06,
      "loss": 0.5958,
      "step": 506
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4914437532424927,
      "learning_rate": 9.999885365433523e-06,
      "loss": 0.6158,
      "step": 507
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2526476383209229,
      "learning_rate": 9.999878312646748e-06,
      "loss": 0.6795,
      "step": 508
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2908756732940674,
      "learning_rate": 9.999871049332488e-06,
      "loss": 0.6971,
      "step": 509
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.206207036972046,
      "learning_rate": 9.999863575491053e-06,
      "loss": 0.6511,
      "step": 510
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1960170269012451,
      "learning_rate": 9.999855891122754e-06,
      "loss": 0.6687,
      "step": 511
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.252910852432251,
      "learning_rate": 9.999847996227918e-06,
      "loss": 0.6606,
      "step": 512
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2176874876022339,
      "learning_rate": 9.999839890806877e-06,
      "loss": 0.6424,
      "step": 513
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1943899393081665,
      "learning_rate": 9.99983157485997e-06,
      "loss": 0.632,
      "step": 514
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1325303316116333,
      "learning_rate": 9.99982304838755e-06,
      "loss": 0.6118,
      "step": 515
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2069593667984009,
      "learning_rate": 9.999814311389973e-06,
      "loss": 0.5657,
      "step": 516
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1481478214263916,
      "learning_rate": 9.99980536386761e-06,
      "loss": 0.6379,
      "step": 517
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1180585622787476,
      "learning_rate": 9.999796205820835e-06,
      "loss": 0.6024,
      "step": 518
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2709500789642334,
      "learning_rate": 9.999786837250034e-06,
      "loss": 0.6019,
      "step": 519
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2685295343399048,
      "learning_rate": 9.999777258155604e-06,
      "loss": 0.6268,
      "step": 520
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2882782220840454,
      "learning_rate": 9.999767468537947e-06,
      "loss": 0.6204,
      "step": 521
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2946155071258545,
      "learning_rate": 9.999757468397473e-06,
      "loss": 0.6244,
      "step": 522
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1825463771820068,
      "learning_rate": 9.999747257734605e-06,
      "loss": 0.5921,
      "step": 523
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3448220491409302,
      "learning_rate": 9.999736836549773e-06,
      "loss": 0.6155,
      "step": 524
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0695064067840576,
      "learning_rate": 9.999726204843417e-06,
      "loss": 0.573,
      "step": 525
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0890889167785645,
      "learning_rate": 9.999715362615983e-06,
      "loss": 0.5706,
      "step": 526
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.129686951637268,
      "learning_rate": 9.999704309867926e-06,
      "loss": 0.6327,
      "step": 527
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.162773847579956,
      "learning_rate": 9.999693046599715e-06,
      "loss": 0.5607,
      "step": 528
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2680752277374268,
      "learning_rate": 9.99968157281182e-06,
      "loss": 0.5889,
      "step": 529
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2788618803024292,
      "learning_rate": 9.999669888504731e-06,
      "loss": 0.606,
      "step": 530
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1362082958221436,
      "learning_rate": 9.999657993678932e-06,
      "loss": 0.6007,
      "step": 531
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.184064269065857,
      "learning_rate": 9.999645888334927e-06,
      "loss": 0.6283,
      "step": 532
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1241337060928345,
      "learning_rate": 9.999633572473228e-06,
      "loss": 0.6264,
      "step": 533
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2259488105773926,
      "learning_rate": 9.999621046094353e-06,
      "loss": 0.6546,
      "step": 534
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1768872737884521,
      "learning_rate": 9.999608309198827e-06,
      "loss": 0.5647,
      "step": 535
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1303322315216064,
      "learning_rate": 9.999595361787187e-06,
      "loss": 0.5865,
      "step": 536
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2701512575149536,
      "learning_rate": 9.999582203859977e-06,
      "loss": 0.6254,
      "step": 537
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2194970846176147,
      "learning_rate": 9.999568835417755e-06,
      "loss": 0.6115,
      "step": 538
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1903170347213745,
      "learning_rate": 9.99955525646108e-06,
      "loss": 0.6354,
      "step": 539
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1576377153396606,
      "learning_rate": 9.999541466990526e-06,
      "loss": 0.6241,
      "step": 540
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1156654357910156,
      "learning_rate": 9.999527467006674e-06,
      "loss": 0.5825,
      "step": 541
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1373095512390137,
      "learning_rate": 9.999513256510112e-06,
      "loss": 0.59,
      "step": 542
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4597004652023315,
      "learning_rate": 9.999498835501438e-06,
      "loss": 0.641,
      "step": 543
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1645809412002563,
      "learning_rate": 9.99948420398126e-06,
      "loss": 0.6794,
      "step": 544
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2074896097183228,
      "learning_rate": 9.999469361950195e-06,
      "loss": 0.6506,
      "step": 545
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1769201755523682,
      "learning_rate": 9.999454309408868e-06,
      "loss": 0.6291,
      "step": 546
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1868668794631958,
      "learning_rate": 9.999439046357908e-06,
      "loss": 0.6222,
      "step": 547
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2111363410949707,
      "learning_rate": 9.999423572797964e-06,
      "loss": 0.5993,
      "step": 548
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.208024024963379,
      "learning_rate": 9.999407888729686e-06,
      "loss": 0.618,
      "step": 549
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2100365161895752,
      "learning_rate": 9.999391994153734e-06,
      "loss": 0.5905,
      "step": 550
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2433186769485474,
      "learning_rate": 9.999375889070773e-06,
      "loss": 0.6088,
      "step": 551
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0811570882797241,
      "learning_rate": 9.99935957348149e-06,
      "loss": 0.6018,
      "step": 552
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0921618938446045,
      "learning_rate": 9.999343047386562e-06,
      "loss": 0.5659,
      "step": 553
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.173858880996704,
      "learning_rate": 9.999326310786692e-06,
      "loss": 0.5816,
      "step": 554
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1979855298995972,
      "learning_rate": 9.999309363682582e-06,
      "loss": 0.5685,
      "step": 555
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1871007680892944,
      "learning_rate": 9.999292206074946e-06,
      "loss": 0.5614,
      "step": 556
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2109037637710571,
      "learning_rate": 9.999274837964507e-06,
      "loss": 0.6184,
      "step": 557
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3156616687774658,
      "learning_rate": 9.999257259351995e-06,
      "loss": 0.6321,
      "step": 558
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1706651449203491,
      "learning_rate": 9.999239470238151e-06,
      "loss": 0.5902,
      "step": 559
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1345263719558716,
      "learning_rate": 9.999221470623726e-06,
      "loss": 0.6015,
      "step": 560
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.209594488143921,
      "learning_rate": 9.999203260509473e-06,
      "loss": 0.5741,
      "step": 561
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.112541913986206,
      "learning_rate": 9.999184839896163e-06,
      "loss": 0.6346,
      "step": 562
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1392542123794556,
      "learning_rate": 9.99916620878457e-06,
      "loss": 0.5987,
      "step": 563
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2119113206863403,
      "learning_rate": 9.99914736717548e-06,
      "loss": 0.6329,
      "step": 564
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1497561931610107,
      "learning_rate": 9.999128315069684e-06,
      "loss": 0.6207,
      "step": 565
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2501511573791504,
      "learning_rate": 9.999109052467986e-06,
      "loss": 0.621,
      "step": 566
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1397020816802979,
      "learning_rate": 9.999089579371195e-06,
      "loss": 0.594,
      "step": 567
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1550371646881104,
      "learning_rate": 9.999069895780133e-06,
      "loss": 0.5853,
      "step": 568
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1494985818862915,
      "learning_rate": 9.99905000169563e-06,
      "loss": 0.5681,
      "step": 569
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1133198738098145,
      "learning_rate": 9.99902989711852e-06,
      "loss": 0.5421,
      "step": 570
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1223570108413696,
      "learning_rate": 9.99900958204965e-06,
      "loss": 0.596,
      "step": 571
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1596499681472778,
      "learning_rate": 9.99898905648988e-06,
      "loss": 0.5999,
      "step": 572
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1666220426559448,
      "learning_rate": 9.998968320440068e-06,
      "loss": 0.599,
      "step": 573
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1381292343139648,
      "learning_rate": 9.998947373901092e-06,
      "loss": 0.5885,
      "step": 574
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.002712607383728,
      "learning_rate": 9.998926216873833e-06,
      "loss": 0.5451,
      "step": 575
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.16362726688385,
      "learning_rate": 9.998904849359179e-06,
      "loss": 0.6408,
      "step": 576
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1892794370651245,
      "learning_rate": 9.998883271358033e-06,
      "loss": 0.6021,
      "step": 577
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.097153902053833,
      "learning_rate": 9.998861482871303e-06,
      "loss": 0.5795,
      "step": 578
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1804133653640747,
      "learning_rate": 9.998839483899904e-06,
      "loss": 0.6161,
      "step": 579
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2267212867736816,
      "learning_rate": 9.998817274444765e-06,
      "loss": 0.6092,
      "step": 580
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1391352415084839,
      "learning_rate": 9.998794854506819e-06,
      "loss": 0.5594,
      "step": 581
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1507866382598877,
      "learning_rate": 9.998772224087011e-06,
      "loss": 0.6018,
      "step": 582
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.114465355873108,
      "learning_rate": 9.998749383186296e-06,
      "loss": 0.5713,
      "step": 583
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.209734320640564,
      "learning_rate": 9.998726331805632e-06,
      "loss": 0.5776,
      "step": 584
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1595596075057983,
      "learning_rate": 9.998703069945995e-06,
      "loss": 0.5707,
      "step": 585
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2344090938568115,
      "learning_rate": 9.998679597608357e-06,
      "loss": 0.6657,
      "step": 586
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.140102744102478,
      "learning_rate": 9.998655914793711e-06,
      "loss": 0.5602,
      "step": 587
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.185093641281128,
      "learning_rate": 9.998632021503055e-06,
      "loss": 0.5627,
      "step": 588
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0849400758743286,
      "learning_rate": 9.998607917737393e-06,
      "loss": 0.5439,
      "step": 589
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2265455722808838,
      "learning_rate": 9.99858360349774e-06,
      "loss": 0.5755,
      "step": 590
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.20005202293396,
      "learning_rate": 9.99855907878512e-06,
      "loss": 0.5961,
      "step": 591
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.128761887550354,
      "learning_rate": 9.998534343600567e-06,
      "loss": 0.5522,
      "step": 592
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.067212700843811,
      "learning_rate": 9.99850939794512e-06,
      "loss": 0.6102,
      "step": 593
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1241345405578613,
      "learning_rate": 9.998484241819833e-06,
      "loss": 0.5949,
      "step": 594
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2213162183761597,
      "learning_rate": 9.99845887522576e-06,
      "loss": 0.5527,
      "step": 595
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2188106775283813,
      "learning_rate": 9.998433298163974e-06,
      "loss": 0.6375,
      "step": 596
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0957785844802856,
      "learning_rate": 9.99840751063555e-06,
      "loss": 0.6003,
      "step": 597
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1837674379348755,
      "learning_rate": 9.998381512641574e-06,
      "loss": 0.5664,
      "step": 598
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1946494579315186,
      "learning_rate": 9.99835530418314e-06,
      "loss": 0.632,
      "step": 599
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1648926734924316,
      "learning_rate": 9.998328885261352e-06,
      "loss": 0.6483,
      "step": 600
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1611378192901611,
      "learning_rate": 9.998302255877323e-06,
      "loss": 0.6712,
      "step": 601
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1499131917953491,
      "learning_rate": 9.998275416032176e-06,
      "loss": 0.5657,
      "step": 602
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2018954753875732,
      "learning_rate": 9.998248365727037e-06,
      "loss": 0.5774,
      "step": 603
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.237338662147522,
      "learning_rate": 9.998221104963047e-06,
      "loss": 0.6025,
      "step": 604
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0457485914230347,
      "learning_rate": 9.998193633741353e-06,
      "loss": 0.5635,
      "step": 605
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1704199314117432,
      "learning_rate": 9.998165952063113e-06,
      "loss": 0.5775,
      "step": 606
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0876002311706543,
      "learning_rate": 9.998138059929493e-06,
      "loss": 0.5717,
      "step": 607
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2615928649902344,
      "learning_rate": 9.998109957341665e-06,
      "loss": 0.6188,
      "step": 608
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1969835758209229,
      "learning_rate": 9.998081644300815e-06,
      "loss": 0.6108,
      "step": 609
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.231022834777832,
      "learning_rate": 9.998053120808133e-06,
      "loss": 0.5429,
      "step": 610
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2075358629226685,
      "learning_rate": 9.998024386864821e-06,
      "loss": 0.5969,
      "step": 611
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1285358667373657,
      "learning_rate": 9.99799544247209e-06,
      "loss": 0.5597,
      "step": 612
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1488717794418335,
      "learning_rate": 9.997966287631157e-06,
      "loss": 0.575,
      "step": 613
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1601454019546509,
      "learning_rate": 9.997936922343253e-06,
      "loss": 0.5865,
      "step": 614
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2259072065353394,
      "learning_rate": 9.997907346609608e-06,
      "loss": 0.6072,
      "step": 615
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1978145837783813,
      "learning_rate": 9.997877560431472e-06,
      "loss": 0.613,
      "step": 616
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1502573490142822,
      "learning_rate": 9.9978475638101e-06,
      "loss": 0.6282,
      "step": 617
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1207703351974487,
      "learning_rate": 9.997817356746751e-06,
      "loss": 0.5915,
      "step": 618
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.049208641052246,
      "learning_rate": 9.9977869392427e-06,
      "loss": 0.5555,
      "step": 619
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2138818502426147,
      "learning_rate": 9.997756311299229e-06,
      "loss": 0.6416,
      "step": 620
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.236164927482605,
      "learning_rate": 9.997725472917623e-06,
      "loss": 0.6295,
      "step": 621
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1690360307693481,
      "learning_rate": 9.997694424099184e-06,
      "loss": 0.5542,
      "step": 622
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0889374017715454,
      "learning_rate": 9.99766316484522e-06,
      "loss": 0.5348,
      "step": 623
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1452760696411133,
      "learning_rate": 9.997631695157043e-06,
      "loss": 0.5982,
      "step": 624
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1351194381713867,
      "learning_rate": 9.997600015035982e-06,
      "loss": 0.6595,
      "step": 625
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.218604326248169,
      "learning_rate": 9.99756812448337e-06,
      "loss": 0.5595,
      "step": 626
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.171190857887268,
      "learning_rate": 9.99753602350055e-06,
      "loss": 0.6254,
      "step": 627
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0584040880203247,
      "learning_rate": 9.997503712088873e-06,
      "loss": 0.5831,
      "step": 628
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1942036151885986,
      "learning_rate": 9.9974711902497e-06,
      "loss": 0.6383,
      "step": 629
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.178575873374939,
      "learning_rate": 9.997438457984398e-06,
      "loss": 0.6528,
      "step": 630
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.159193992614746,
      "learning_rate": 9.997405515294349e-06,
      "loss": 0.5805,
      "step": 631
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1333950757980347,
      "learning_rate": 9.99737236218094e-06,
      "loss": 0.6203,
      "step": 632
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2014952898025513,
      "learning_rate": 9.997338998645562e-06,
      "loss": 0.6107,
      "step": 633
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1212387084960938,
      "learning_rate": 9.997305424689626e-06,
      "loss": 0.6099,
      "step": 634
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1277556419372559,
      "learning_rate": 9.997271640314542e-06,
      "loss": 0.6042,
      "step": 635
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.149415135383606,
      "learning_rate": 9.997237645521733e-06,
      "loss": 0.614,
      "step": 636
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1585355997085571,
      "learning_rate": 9.997203440312632e-06,
      "loss": 0.5867,
      "step": 637
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.165755271911621,
      "learning_rate": 9.997169024688678e-06,
      "loss": 0.57,
      "step": 638
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1480127573013306,
      "learning_rate": 9.997134398651318e-06,
      "loss": 0.5741,
      "step": 639
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1376844644546509,
      "learning_rate": 9.997099562202015e-06,
      "loss": 0.6307,
      "step": 640
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2017743587493896,
      "learning_rate": 9.997064515342232e-06,
      "loss": 0.5886,
      "step": 641
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2070648670196533,
      "learning_rate": 9.997029258073445e-06,
      "loss": 0.6148,
      "step": 642
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1960395574569702,
      "learning_rate": 9.99699379039714e-06,
      "loss": 0.6255,
      "step": 643
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.232494831085205,
      "learning_rate": 9.996958112314811e-06,
      "loss": 0.5936,
      "step": 644
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1607786417007446,
      "learning_rate": 9.996922223827958e-06,
      "loss": 0.5937,
      "step": 645
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5826513767242432,
      "learning_rate": 9.996886124938092e-06,
      "loss": 0.6248,
      "step": 646
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1372714042663574,
      "learning_rate": 9.996849815646736e-06,
      "loss": 0.5862,
      "step": 647
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1420936584472656,
      "learning_rate": 9.996813295955417e-06,
      "loss": 0.6175,
      "step": 648
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1022953987121582,
      "learning_rate": 9.996776565865671e-06,
      "loss": 0.5774,
      "step": 649
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1705585718154907,
      "learning_rate": 9.996739625379049e-06,
      "loss": 0.595,
      "step": 650
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2827461957931519,
      "learning_rate": 9.9967024744971e-06,
      "loss": 0.5825,
      "step": 651
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2297942638397217,
      "learning_rate": 9.996665113221396e-06,
      "loss": 0.6021,
      "step": 652
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0891423225402832,
      "learning_rate": 9.996627541553504e-06,
      "loss": 0.5903,
      "step": 653
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1050660610198975,
      "learning_rate": 9.996589759495008e-06,
      "loss": 0.5593,
      "step": 654
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.292536973953247,
      "learning_rate": 9.9965517670475e-06,
      "loss": 0.6653,
      "step": 655
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2416012287139893,
      "learning_rate": 9.996513564212577e-06,
      "loss": 0.5872,
      "step": 656
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0946153402328491,
      "learning_rate": 9.996475150991852e-06,
      "loss": 0.5968,
      "step": 657
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.259873628616333,
      "learning_rate": 9.99643652738694e-06,
      "loss": 0.6341,
      "step": 658
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1964812278747559,
      "learning_rate": 9.996397693399465e-06,
      "loss": 0.6192,
      "step": 659
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1455782651901245,
      "learning_rate": 9.996358649031066e-06,
      "loss": 0.6043,
      "step": 660
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1354379653930664,
      "learning_rate": 9.996319394283384e-06,
      "loss": 0.5798,
      "step": 661
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1652181148529053,
      "learning_rate": 9.996279929158074e-06,
      "loss": 0.6829,
      "step": 662
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.167168378829956,
      "learning_rate": 9.996240253656796e-06,
      "loss": 0.5979,
      "step": 663
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1037861108779907,
      "learning_rate": 9.996200367781224e-06,
      "loss": 0.5988,
      "step": 664
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1305545568466187,
      "learning_rate": 9.996160271533033e-06,
      "loss": 0.6001,
      "step": 665
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2025097608566284,
      "learning_rate": 9.996119964913914e-06,
      "loss": 0.5899,
      "step": 666
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2237567901611328,
      "learning_rate": 9.996079447925563e-06,
      "loss": 0.6244,
      "step": 667
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1735132932662964,
      "learning_rate": 9.996038720569688e-06,
      "loss": 0.5966,
      "step": 668
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0479474067687988,
      "learning_rate": 9.995997782848e-06,
      "loss": 0.6146,
      "step": 669
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.071723222732544,
      "learning_rate": 9.995956634762227e-06,
      "loss": 0.5648,
      "step": 670
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1734130382537842,
      "learning_rate": 9.995915276314099e-06,
      "loss": 0.605,
      "step": 671
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.209862232208252,
      "learning_rate": 9.995873707505358e-06,
      "loss": 0.5932,
      "step": 672
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1562873125076294,
      "learning_rate": 9.995831928337756e-06,
      "loss": 0.5997,
      "step": 673
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2242332696914673,
      "learning_rate": 9.99578993881305e-06,
      "loss": 0.6525,
      "step": 674
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2193008661270142,
      "learning_rate": 9.995747738933009e-06,
      "loss": 0.5475,
      "step": 675
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.106239914894104,
      "learning_rate": 9.995705328699408e-06,
      "loss": 0.6182,
      "step": 676
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2187772989273071,
      "learning_rate": 9.995662708114036e-06,
      "loss": 0.6567,
      "step": 677
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.192084789276123,
      "learning_rate": 9.995619877178685e-06,
      "loss": 0.636,
      "step": 678
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1521915197372437,
      "learning_rate": 9.99557683589516e-06,
      "loss": 0.6575,
      "step": 679
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2543455362319946,
      "learning_rate": 9.995533584265273e-06,
      "loss": 0.6198,
      "step": 680
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.362525224685669,
      "learning_rate": 9.995490122290845e-06,
      "loss": 0.6835,
      "step": 681
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1788954734802246,
      "learning_rate": 9.995446449973705e-06,
      "loss": 0.6033,
      "step": 682
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.065611720085144,
      "learning_rate": 9.995402567315695e-06,
      "loss": 0.5724,
      "step": 683
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1058067083358765,
      "learning_rate": 9.99535847431866e-06,
      "loss": 0.6266,
      "step": 684
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1485073566436768,
      "learning_rate": 9.995314170984457e-06,
      "loss": 0.5928,
      "step": 685
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.17409348487854,
      "learning_rate": 9.99526965731495e-06,
      "loss": 0.5886,
      "step": 686
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1527072191238403,
      "learning_rate": 9.995224933312016e-06,
      "loss": 0.6152,
      "step": 687
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2115527391433716,
      "learning_rate": 9.995179998977537e-06,
      "loss": 0.5918,
      "step": 688
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0342247486114502,
      "learning_rate": 9.995134854313407e-06,
      "loss": 0.5787,
      "step": 689
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2421724796295166,
      "learning_rate": 9.995089499321521e-06,
      "loss": 0.6279,
      "step": 690
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2693426609039307,
      "learning_rate": 9.995043934003796e-06,
      "loss": 0.5894,
      "step": 691
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1222983598709106,
      "learning_rate": 9.994998158362148e-06,
      "loss": 0.6051,
      "step": 692
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3039426803588867,
      "learning_rate": 9.994952172398502e-06,
      "loss": 0.6284,
      "step": 693
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.194825530052185,
      "learning_rate": 9.994905976114799e-06,
      "loss": 0.6795,
      "step": 694
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4020371437072754,
      "learning_rate": 9.994859569512978e-06,
      "loss": 0.5985,
      "step": 695
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.410729169845581,
      "learning_rate": 9.994812952594998e-06,
      "loss": 0.6006,
      "step": 696
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3058116436004639,
      "learning_rate": 9.994766125362821e-06,
      "loss": 0.5779,
      "step": 697
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1895015239715576,
      "learning_rate": 9.994719087818416e-06,
      "loss": 0.6208,
      "step": 698
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1635770797729492,
      "learning_rate": 9.994671839963766e-06,
      "loss": 0.586,
      "step": 699
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1304101943969727,
      "learning_rate": 9.994624381800861e-06,
      "loss": 0.5658,
      "step": 700
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0639395713806152,
      "learning_rate": 9.994576713331699e-06,
      "loss": 0.5744,
      "step": 701
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1864066123962402,
      "learning_rate": 9.994528834558285e-06,
      "loss": 0.6339,
      "step": 702
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1535203456878662,
      "learning_rate": 9.994480745482636e-06,
      "loss": 0.6583,
      "step": 703
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.063400149345398,
      "learning_rate": 9.99443244610678e-06,
      "loss": 0.5663,
      "step": 704
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0646471977233887,
      "learning_rate": 9.994383936432745e-06,
      "loss": 0.5485,
      "step": 705
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2647340297698975,
      "learning_rate": 9.994335216462579e-06,
      "loss": 0.5776,
      "step": 706
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1197046041488647,
      "learning_rate": 9.99428628619833e-06,
      "loss": 0.5988,
      "step": 707
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2443758249282837,
      "learning_rate": 9.994237145642058e-06,
      "loss": 0.6535,
      "step": 708
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0635206699371338,
      "learning_rate": 9.994187794795835e-06,
      "loss": 0.5706,
      "step": 709
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1187067031860352,
      "learning_rate": 9.994138233661737e-06,
      "loss": 0.6307,
      "step": 710
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2295868396759033,
      "learning_rate": 9.994088462241851e-06,
      "loss": 0.6248,
      "step": 711
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2285339832305908,
      "learning_rate": 9.994038480538274e-06,
      "loss": 0.5898,
      "step": 712
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2357820272445679,
      "learning_rate": 9.993988288553109e-06,
      "loss": 0.6122,
      "step": 713
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1227229833602905,
      "learning_rate": 9.993937886288471e-06,
      "loss": 0.5517,
      "step": 714
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.25047767162323,
      "learning_rate": 9.99388727374648e-06,
      "loss": 0.6434,
      "step": 715
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0810105800628662,
      "learning_rate": 9.993836450929268e-06,
      "loss": 0.5894,
      "step": 716
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1047571897506714,
      "learning_rate": 9.993785417838978e-06,
      "loss": 0.5227,
      "step": 717
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.249957799911499,
      "learning_rate": 9.993734174477752e-06,
      "loss": 0.5827,
      "step": 718
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2992061376571655,
      "learning_rate": 9.993682720847755e-06,
      "loss": 0.5683,
      "step": 719
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4556405544281006,
      "learning_rate": 9.99363105695115e-06,
      "loss": 0.6054,
      "step": 720
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3055137395858765,
      "learning_rate": 9.993579182790111e-06,
      "loss": 0.6127,
      "step": 721
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1818618774414062,
      "learning_rate": 9.993527098366826e-06,
      "loss": 0.554,
      "step": 722
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.219436526298523,
      "learning_rate": 9.993474803683486e-06,
      "loss": 0.6292,
      "step": 723
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.216051697731018,
      "learning_rate": 9.993422298742293e-06,
      "loss": 0.576,
      "step": 724
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2530943155288696,
      "learning_rate": 9.993369583545456e-06,
      "loss": 0.5972,
      "step": 725
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2716203927993774,
      "learning_rate": 9.993316658095198e-06,
      "loss": 0.6217,
      "step": 726
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1261982917785645,
      "learning_rate": 9.993263522393745e-06,
      "loss": 0.6376,
      "step": 727
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1981035470962524,
      "learning_rate": 9.993210176443338e-06,
      "loss": 0.611,
      "step": 728
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3648395538330078,
      "learning_rate": 9.993156620246219e-06,
      "loss": 0.6541,
      "step": 729
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.226815938949585,
      "learning_rate": 9.993102853804643e-06,
      "loss": 0.6128,
      "step": 730
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1845619678497314,
      "learning_rate": 9.993048877120876e-06,
      "loss": 0.5807,
      "step": 731
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.221136450767517,
      "learning_rate": 9.992994690197192e-06,
      "loss": 0.6016,
      "step": 732
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1222420930862427,
      "learning_rate": 9.992940293035871e-06,
      "loss": 0.6127,
      "step": 733
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.106492042541504,
      "learning_rate": 9.992885685639203e-06,
      "loss": 0.5583,
      "step": 734
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1257810592651367,
      "learning_rate": 9.992830868009487e-06,
      "loss": 0.6227,
      "step": 735
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1145000457763672,
      "learning_rate": 9.992775840149031e-06,
      "loss": 0.5552,
      "step": 736
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.205623745918274,
      "learning_rate": 9.992720602060155e-06,
      "loss": 0.6107,
      "step": 737
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2029021978378296,
      "learning_rate": 9.992665153745182e-06,
      "loss": 0.6487,
      "step": 738
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.222591519355774,
      "learning_rate": 9.992609495206448e-06,
      "loss": 0.648,
      "step": 739
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.11492121219635,
      "learning_rate": 9.992553626446296e-06,
      "loss": 0.5802,
      "step": 740
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2474178075790405,
      "learning_rate": 9.992497547467079e-06,
      "loss": 0.6533,
      "step": 741
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1707470417022705,
      "learning_rate": 9.992441258271157e-06,
      "loss": 0.6291,
      "step": 742
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2481380701065063,
      "learning_rate": 9.992384758860902e-06,
      "loss": 0.6109,
      "step": 743
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1987773180007935,
      "learning_rate": 9.99232804923869e-06,
      "loss": 0.605,
      "step": 744
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2029579877853394,
      "learning_rate": 9.992271129406914e-06,
      "loss": 0.6233,
      "step": 745
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.165914535522461,
      "learning_rate": 9.992213999367965e-06,
      "loss": 0.5919,
      "step": 746
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1722065210342407,
      "learning_rate": 9.992156659124253e-06,
      "loss": 0.5918,
      "step": 747
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.137799620628357,
      "learning_rate": 9.99209910867819e-06,
      "loss": 0.6184,
      "step": 748
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.033355951309204,
      "learning_rate": 9.9920413480322e-06,
      "loss": 0.5688,
      "step": 749
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2363529205322266,
      "learning_rate": 9.991983377188715e-06,
      "loss": 0.5838,
      "step": 750
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1456830501556396,
      "learning_rate": 9.991925196150174e-06,
      "loss": 0.5965,
      "step": 751
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.098024606704712,
      "learning_rate": 9.99186680491903e-06,
      "loss": 0.5547,
      "step": 752
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1140855550765991,
      "learning_rate": 9.99180820349774e-06,
      "loss": 0.6416,
      "step": 753
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2580140829086304,
      "learning_rate": 9.991749391888772e-06,
      "loss": 0.6367,
      "step": 754
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2655225992202759,
      "learning_rate": 9.991690370094603e-06,
      "loss": 0.6472,
      "step": 755
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2652753591537476,
      "learning_rate": 9.991631138117715e-06,
      "loss": 0.6705,
      "step": 756
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1617118120193481,
      "learning_rate": 9.991571695960606e-06,
      "loss": 0.6106,
      "step": 757
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9705402851104736,
      "learning_rate": 9.991512043625777e-06,
      "loss": 0.5495,
      "step": 758
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1828492879867554,
      "learning_rate": 9.991452181115739e-06,
      "loss": 0.5876,
      "step": 759
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1426823139190674,
      "learning_rate": 9.991392108433016e-06,
      "loss": 0.6483,
      "step": 760
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.100878119468689,
      "learning_rate": 9.991331825580132e-06,
      "loss": 0.6322,
      "step": 761
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1711697578430176,
      "learning_rate": 9.99127133255963e-06,
      "loss": 0.6244,
      "step": 762
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1956875324249268,
      "learning_rate": 9.991210629374058e-06,
      "loss": 0.5852,
      "step": 763
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.176194429397583,
      "learning_rate": 9.991149716025967e-06,
      "loss": 0.606,
      "step": 764
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2652812004089355,
      "learning_rate": 9.991088592517924e-06,
      "loss": 0.6008,
      "step": 765
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.122599720954895,
      "learning_rate": 9.991027258852505e-06,
      "loss": 0.596,
      "step": 766
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0915685892105103,
      "learning_rate": 9.990965715032289e-06,
      "loss": 0.5382,
      "step": 767
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0834283828735352,
      "learning_rate": 9.99090396105987e-06,
      "loss": 0.5706,
      "step": 768
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1483625173568726,
      "learning_rate": 9.990841996937846e-06,
      "loss": 0.6083,
      "step": 769
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1269500255584717,
      "learning_rate": 9.990779822668827e-06,
      "loss": 0.6419,
      "step": 770
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0941485166549683,
      "learning_rate": 9.990717438255435e-06,
      "loss": 0.5956,
      "step": 771
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1730438470840454,
      "learning_rate": 9.99065484370029e-06,
      "loss": 0.6609,
      "step": 772
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1437172889709473,
      "learning_rate": 9.99059203900603e-06,
      "loss": 0.5855,
      "step": 773
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1737104654312134,
      "learning_rate": 9.990529024175303e-06,
      "loss": 0.514,
      "step": 774
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.113284945487976,
      "learning_rate": 9.990465799210757e-06,
      "loss": 0.5614,
      "step": 775
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1417404413223267,
      "learning_rate": 9.99040236411506e-06,
      "loss": 0.6639,
      "step": 776
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1533658504486084,
      "learning_rate": 9.990338718890878e-06,
      "loss": 0.6042,
      "step": 777
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1291682720184326,
      "learning_rate": 9.990274863540891e-06,
      "loss": 0.5664,
      "step": 778
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1921929121017456,
      "learning_rate": 9.990210798067792e-06,
      "loss": 0.6371,
      "step": 779
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2903766632080078,
      "learning_rate": 9.990146522474273e-06,
      "loss": 0.6215,
      "step": 780
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.139093279838562,
      "learning_rate": 9.990082036763046e-06,
      "loss": 0.6203,
      "step": 781
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0165280103683472,
      "learning_rate": 9.990017340936823e-06,
      "loss": 0.5773,
      "step": 782
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0505095720291138,
      "learning_rate": 9.989952434998328e-06,
      "loss": 0.6327,
      "step": 783
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1323219537734985,
      "learning_rate": 9.989887318950295e-06,
      "loss": 0.6173,
      "step": 784
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1180750131607056,
      "learning_rate": 9.989821992795467e-06,
      "loss": 0.5911,
      "step": 785
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.067559003829956,
      "learning_rate": 9.989756456536593e-06,
      "loss": 0.555,
      "step": 786
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2159489393234253,
      "learning_rate": 9.989690710176433e-06,
      "loss": 0.603,
      "step": 787
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1859744787216187,
      "learning_rate": 9.989624753717752e-06,
      "loss": 0.632,
      "step": 788
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1191562414169312,
      "learning_rate": 9.989558587163332e-06,
      "loss": 0.5882,
      "step": 789
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.043337106704712,
      "learning_rate": 9.989492210515958e-06,
      "loss": 0.5937,
      "step": 790
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1230212450027466,
      "learning_rate": 9.989425623778423e-06,
      "loss": 0.611,
      "step": 791
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.199364423751831,
      "learning_rate": 9.989358826953533e-06,
      "loss": 0.6173,
      "step": 792
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.087085247039795,
      "learning_rate": 9.989291820044099e-06,
      "loss": 0.5921,
      "step": 793
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1019243001937866,
      "learning_rate": 9.989224603052943e-06,
      "loss": 0.5381,
      "step": 794
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1220672130584717,
      "learning_rate": 9.989157175982896e-06,
      "loss": 0.6278,
      "step": 795
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.167831540107727,
      "learning_rate": 9.989089538836795e-06,
      "loss": 0.5562,
      "step": 796
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1575336456298828,
      "learning_rate": 9.98902169161749e-06,
      "loss": 0.6192,
      "step": 797
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1505603790283203,
      "learning_rate": 9.988953634327836e-06,
      "loss": 0.6589,
      "step": 798
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0694489479064941,
      "learning_rate": 9.988885366970701e-06,
      "loss": 0.6314,
      "step": 799
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1580840349197388,
      "learning_rate": 9.988816889548958e-06,
      "loss": 0.6169,
      "step": 800
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1379700899124146,
      "learning_rate": 9.98874820206549e-06,
      "loss": 0.597,
      "step": 801
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1595264673233032,
      "learning_rate": 9.988679304523192e-06,
      "loss": 0.5541,
      "step": 802
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1024327278137207,
      "learning_rate": 9.988610196924962e-06,
      "loss": 0.6035,
      "step": 803
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.074629306793213,
      "learning_rate": 9.98854087927371e-06,
      "loss": 0.5977,
      "step": 804
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0339958667755127,
      "learning_rate": 9.988471351572355e-06,
      "loss": 0.5759,
      "step": 805
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.179978609085083,
      "learning_rate": 9.988401613823825e-06,
      "loss": 0.6151,
      "step": 806
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2351336479187012,
      "learning_rate": 9.988331666031056e-06,
      "loss": 0.6234,
      "step": 807
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1425678730010986,
      "learning_rate": 9.988261508196994e-06,
      "loss": 0.5846,
      "step": 808
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1734411716461182,
      "learning_rate": 9.988191140324595e-06,
      "loss": 0.5806,
      "step": 809
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.999897837638855,
      "learning_rate": 9.988120562416817e-06,
      "loss": 0.5956,
      "step": 810
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0977915525436401,
      "learning_rate": 9.988049774476636e-06,
      "loss": 0.5887,
      "step": 811
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0584194660186768,
      "learning_rate": 9.98797877650703e-06,
      "loss": 0.5919,
      "step": 812
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0896602869033813,
      "learning_rate": 9.987907568510991e-06,
      "loss": 0.5705,
      "step": 813
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2355543375015259,
      "learning_rate": 9.987836150491515e-06,
      "loss": 0.6367,
      "step": 814
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2684205770492554,
      "learning_rate": 9.98776452245161e-06,
      "loss": 0.6017,
      "step": 815
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9456241726875305,
      "learning_rate": 9.987692684394294e-06,
      "loss": 0.5532,
      "step": 816
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.085779070854187,
      "learning_rate": 9.987620636322589e-06,
      "loss": 0.5819,
      "step": 817
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2396453619003296,
      "learning_rate": 9.987548378239529e-06,
      "loss": 0.6707,
      "step": 818
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.173922061920166,
      "learning_rate": 9.987475910148156e-06,
      "loss": 0.6068,
      "step": 819
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.970253050327301,
      "learning_rate": 9.987403232051525e-06,
      "loss": 0.5501,
      "step": 820
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0929089784622192,
      "learning_rate": 9.987330343952692e-06,
      "loss": 0.5669,
      "step": 821
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.085546851158142,
      "learning_rate": 9.987257245854729e-06,
      "loss": 0.5842,
      "step": 822
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1120977401733398,
      "learning_rate": 9.987183937760713e-06,
      "loss": 0.6162,
      "step": 823
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1035264730453491,
      "learning_rate": 9.98711041967373e-06,
      "loss": 0.6099,
      "step": 824
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0656368732452393,
      "learning_rate": 9.987036691596877e-06,
      "loss": 0.5894,
      "step": 825
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1329158544540405,
      "learning_rate": 9.986962753533257e-06,
      "loss": 0.6243,
      "step": 826
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0665289163589478,
      "learning_rate": 9.986888605485983e-06,
      "loss": 0.5545,
      "step": 827
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1567240953445435,
      "learning_rate": 9.986814247458177e-06,
      "loss": 0.583,
      "step": 828
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1955913305282593,
      "learning_rate": 9.986739679452973e-06,
      "loss": 0.6278,
      "step": 829
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0708664655685425,
      "learning_rate": 9.986664901473508e-06,
      "loss": 0.5302,
      "step": 830
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0710406303405762,
      "learning_rate": 9.98658991352293e-06,
      "loss": 0.5877,
      "step": 831
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0962172746658325,
      "learning_rate": 9.986514715604401e-06,
      "loss": 0.5817,
      "step": 832
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1612870693206787,
      "learning_rate": 9.986439307721083e-06,
      "loss": 0.6012,
      "step": 833
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1956734657287598,
      "learning_rate": 9.98636368987615e-06,
      "loss": 0.604,
      "step": 834
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.098853588104248,
      "learning_rate": 9.98628786207279e-06,
      "loss": 0.5551,
      "step": 835
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0404162406921387,
      "learning_rate": 9.986211824314193e-06,
      "loss": 0.5897,
      "step": 836
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.105139136314392,
      "learning_rate": 9.986135576603564e-06,
      "loss": 0.6254,
      "step": 837
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0726534128189087,
      "learning_rate": 9.98605911894411e-06,
      "loss": 0.5993,
      "step": 838
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0231831073760986,
      "learning_rate": 9.985982451339054e-06,
      "loss": 0.5989,
      "step": 839
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1226348876953125,
      "learning_rate": 9.985905573791619e-06,
      "loss": 0.6273,
      "step": 840
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0960124731063843,
      "learning_rate": 9.985828486305046e-06,
      "loss": 0.6176,
      "step": 841
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1282395124435425,
      "learning_rate": 9.98575118888258e-06,
      "loss": 0.5881,
      "step": 842
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2027688026428223,
      "learning_rate": 9.985673681527474e-06,
      "loss": 0.6038,
      "step": 843
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2403696775436401,
      "learning_rate": 9.985595964242996e-06,
      "loss": 0.5738,
      "step": 844
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.21244478225708,
      "learning_rate": 9.985518037032413e-06,
      "loss": 0.6733,
      "step": 845
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1472803354263306,
      "learning_rate": 9.98543989989901e-06,
      "loss": 0.6156,
      "step": 846
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2124862670898438,
      "learning_rate": 9.985361552846076e-06,
      "loss": 0.5946,
      "step": 847
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.091420292854309,
      "learning_rate": 9.98528299587691e-06,
      "loss": 0.5529,
      "step": 848
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.020264744758606,
      "learning_rate": 9.98520422899482e-06,
      "loss": 0.612,
      "step": 849
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.190901756286621,
      "learning_rate": 9.985125252203122e-06,
      "loss": 0.5835,
      "step": 850
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1265788078308105,
      "learning_rate": 9.985046065505141e-06,
      "loss": 0.5982,
      "step": 851
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1043763160705566,
      "learning_rate": 9.984966668904211e-06,
      "loss": 0.557,
      "step": 852
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1575721502304077,
      "learning_rate": 9.984887062403678e-06,
      "loss": 0.6205,
      "step": 853
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1319472789764404,
      "learning_rate": 9.984807246006891e-06,
      "loss": 0.5921,
      "step": 854
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1670777797698975,
      "learning_rate": 9.984727219717212e-06,
      "loss": 0.608,
      "step": 855
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0812734365463257,
      "learning_rate": 9.984646983538009e-06,
      "loss": 0.6118,
      "step": 856
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2277408838272095,
      "learning_rate": 9.984566537472662e-06,
      "loss": 0.6244,
      "step": 857
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.131480097770691,
      "learning_rate": 9.98448588152456e-06,
      "loss": 0.6142,
      "step": 858
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1145899295806885,
      "learning_rate": 9.984405015697097e-06,
      "loss": 0.5584,
      "step": 859
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1443415880203247,
      "learning_rate": 9.984323939993678e-06,
      "loss": 0.5902,
      "step": 860
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.162580966949463,
      "learning_rate": 9.984242654417716e-06,
      "loss": 0.6336,
      "step": 861
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0649712085723877,
      "learning_rate": 9.984161158972636e-06,
      "loss": 0.5705,
      "step": 862
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1186282634735107,
      "learning_rate": 9.984079453661869e-06,
      "loss": 0.6159,
      "step": 863
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.147649884223938,
      "learning_rate": 9.983997538488851e-06,
      "loss": 0.593,
      "step": 864
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0798815488815308,
      "learning_rate": 9.983915413457036e-06,
      "loss": 0.5644,
      "step": 865
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.051112174987793,
      "learning_rate": 9.983833078569883e-06,
      "loss": 0.6032,
      "step": 866
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1565731763839722,
      "learning_rate": 9.983750533830856e-06,
      "loss": 0.6559,
      "step": 867
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2471048831939697,
      "learning_rate": 9.98366777924343e-06,
      "loss": 0.6315,
      "step": 868
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1234080791473389,
      "learning_rate": 9.983584814811092e-06,
      "loss": 0.5931,
      "step": 869
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1437253952026367,
      "learning_rate": 9.983501640537333e-06,
      "loss": 0.6697,
      "step": 870
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1277947425842285,
      "learning_rate": 9.983418256425656e-06,
      "loss": 0.6253,
      "step": 871
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1978877782821655,
      "learning_rate": 9.983334662479572e-06,
      "loss": 0.5755,
      "step": 872
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0907119512557983,
      "learning_rate": 9.983250858702603e-06,
      "loss": 0.5602,
      "step": 873
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0816748142242432,
      "learning_rate": 9.983166845098275e-06,
      "loss": 0.5799,
      "step": 874
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0800241231918335,
      "learning_rate": 9.983082621670126e-06,
      "loss": 0.541,
      "step": 875
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.234094500541687,
      "learning_rate": 9.982998188421702e-06,
      "loss": 0.6351,
      "step": 876
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1599239110946655,
      "learning_rate": 9.98291354535656e-06,
      "loss": 0.6103,
      "step": 877
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1438367366790771,
      "learning_rate": 9.982828692478261e-06,
      "loss": 0.5765,
      "step": 878
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3228342533111572,
      "learning_rate": 9.982743629790382e-06,
      "loss": 0.5715,
      "step": 879
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1042391061782837,
      "learning_rate": 9.982658357296502e-06,
      "loss": 0.5674,
      "step": 880
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.194968819618225,
      "learning_rate": 9.982572875000212e-06,
      "loss": 0.5971,
      "step": 881
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1787186861038208,
      "learning_rate": 9.98248718290511e-06,
      "loss": 0.6431,
      "step": 882
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9943352341651917,
      "learning_rate": 9.982401281014806e-06,
      "loss": 0.5489,
      "step": 883
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1775997877120972,
      "learning_rate": 9.982315169332918e-06,
      "loss": 0.655,
      "step": 884
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1457655429840088,
      "learning_rate": 9.982228847863069e-06,
      "loss": 0.5942,
      "step": 885
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0527290105819702,
      "learning_rate": 9.982142316608897e-06,
      "loss": 0.5787,
      "step": 886
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2122856378555298,
      "learning_rate": 9.982055575574042e-06,
      "loss": 0.6067,
      "step": 887
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1700632572174072,
      "learning_rate": 9.981968624762159e-06,
      "loss": 0.6277,
      "step": 888
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1444690227508545,
      "learning_rate": 9.981881464176908e-06,
      "loss": 0.5792,
      "step": 889
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1081745624542236,
      "learning_rate": 9.981794093821957e-06,
      "loss": 0.5623,
      "step": 890
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0724091529846191,
      "learning_rate": 9.981706513700989e-06,
      "loss": 0.6175,
      "step": 891
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0415534973144531,
      "learning_rate": 9.98161872381769e-06,
      "loss": 0.5494,
      "step": 892
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1281497478485107,
      "learning_rate": 9.981530724175756e-06,
      "loss": 0.5118,
      "step": 893
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.136470079421997,
      "learning_rate": 9.981442514778892e-06,
      "loss": 0.5983,
      "step": 894
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0326050519943237,
      "learning_rate": 9.981354095630816e-06,
      "loss": 0.5643,
      "step": 895
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1196739673614502,
      "learning_rate": 9.981265466735244e-06,
      "loss": 0.5643,
      "step": 896
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0784188508987427,
      "learning_rate": 9.981176628095913e-06,
      "loss": 0.595,
      "step": 897
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1706982851028442,
      "learning_rate": 9.981087579716564e-06,
      "loss": 0.5883,
      "step": 898
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.096197485923767,
      "learning_rate": 9.980998321600944e-06,
      "loss": 0.5545,
      "step": 899
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0704962015151978,
      "learning_rate": 9.98090885375281e-06,
      "loss": 0.6562,
      "step": 900
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2794032096862793,
      "learning_rate": 9.980819176175932e-06,
      "loss": 0.6009,
      "step": 901
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1802647113800049,
      "learning_rate": 9.980729288874088e-06,
      "loss": 0.6218,
      "step": 902
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1883010864257812,
      "learning_rate": 9.98063919185106e-06,
      "loss": 0.5624,
      "step": 903
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2000126838684082,
      "learning_rate": 9.980548885110641e-06,
      "loss": 0.5604,
      "step": 904
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0881659984588623,
      "learning_rate": 9.980458368656635e-06,
      "loss": 0.6595,
      "step": 905
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1219079494476318,
      "learning_rate": 9.980367642492852e-06,
      "loss": 0.6681,
      "step": 906
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2020022869110107,
      "learning_rate": 9.980276706623114e-06,
      "loss": 0.5726,
      "step": 907
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.061338186264038,
      "learning_rate": 9.98018556105125e-06,
      "loss": 0.5565,
      "step": 908
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.218387484550476,
      "learning_rate": 9.980094205781094e-06,
      "loss": 0.5792,
      "step": 909
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1016128063201904,
      "learning_rate": 9.980002640816498e-06,
      "loss": 0.58,
      "step": 910
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0820120573043823,
      "learning_rate": 9.979910866161313e-06,
      "loss": 0.6373,
      "step": 911
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1650376319885254,
      "learning_rate": 9.979818881819408e-06,
      "loss": 0.5965,
      "step": 912
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1960431337356567,
      "learning_rate": 9.979726687794651e-06,
      "loss": 0.6847,
      "step": 913
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.202374815940857,
      "learning_rate": 9.979634284090927e-06,
      "loss": 0.5545,
      "step": 914
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1402519941329956,
      "learning_rate": 9.979541670712125e-06,
      "loss": 0.6338,
      "step": 915
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1344772577285767,
      "learning_rate": 9.979448847662148e-06,
      "loss": 0.5729,
      "step": 916
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.156521201133728,
      "learning_rate": 9.979355814944901e-06,
      "loss": 0.6434,
      "step": 917
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.120430588722229,
      "learning_rate": 9.979262572564303e-06,
      "loss": 0.552,
      "step": 918
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.125915288925171,
      "learning_rate": 9.979169120524279e-06,
      "loss": 0.57,
      "step": 919
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1396992206573486,
      "learning_rate": 9.979075458828765e-06,
      "loss": 0.6027,
      "step": 920
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1143865585327148,
      "learning_rate": 9.978981587481705e-06,
      "loss": 0.6111,
      "step": 921
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3239092826843262,
      "learning_rate": 9.978887506487049e-06,
      "loss": 0.6774,
      "step": 922
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0860577821731567,
      "learning_rate": 9.978793215848763e-06,
      "loss": 0.577,
      "step": 923
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1568961143493652,
      "learning_rate": 9.978698715570814e-06,
      "loss": 0.6195,
      "step": 924
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1322360038757324,
      "learning_rate": 9.97860400565718e-06,
      "loss": 0.5702,
      "step": 925
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.077203392982483,
      "learning_rate": 9.978509086111852e-06,
      "loss": 0.6116,
      "step": 926
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0534157752990723,
      "learning_rate": 9.978413956938824e-06,
      "loss": 0.5759,
      "step": 927
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1379859447479248,
      "learning_rate": 9.978318618142104e-06,
      "loss": 0.5823,
      "step": 928
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1714255809783936,
      "learning_rate": 9.978223069725706e-06,
      "loss": 0.5625,
      "step": 929
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1636358499526978,
      "learning_rate": 9.978127311693653e-06,
      "loss": 0.5302,
      "step": 930
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.116036295890808,
      "learning_rate": 9.978031344049975e-06,
      "loss": 0.5863,
      "step": 931
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1864908933639526,
      "learning_rate": 9.977935166798714e-06,
      "loss": 0.6192,
      "step": 932
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0718437433242798,
      "learning_rate": 9.977838779943921e-06,
      "loss": 0.5689,
      "step": 933
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1931647062301636,
      "learning_rate": 9.977742183489653e-06,
      "loss": 0.6382,
      "step": 934
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1988134384155273,
      "learning_rate": 9.977645377439979e-06,
      "loss": 0.5826,
      "step": 935
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1773120164871216,
      "learning_rate": 9.977548361798975e-06,
      "loss": 0.5932,
      "step": 936
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1375964879989624,
      "learning_rate": 9.977451136570722e-06,
      "loss": 0.6283,
      "step": 937
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1152372360229492,
      "learning_rate": 9.97735370175932e-06,
      "loss": 0.5704,
      "step": 938
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0861738920211792,
      "learning_rate": 9.977256057368865e-06,
      "loss": 0.6062,
      "step": 939
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.097822904586792,
      "learning_rate": 9.977158203403475e-06,
      "loss": 0.6039,
      "step": 940
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0957974195480347,
      "learning_rate": 9.977060139867268e-06,
      "loss": 0.5616,
      "step": 941
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0065202713012695,
      "learning_rate": 9.97696186676437e-06,
      "loss": 0.5236,
      "step": 942
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1139683723449707,
      "learning_rate": 9.976863384098923e-06,
      "loss": 0.6152,
      "step": 943
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1870875358581543,
      "learning_rate": 9.976764691875072e-06,
      "loss": 0.6356,
      "step": 944
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1149030923843384,
      "learning_rate": 9.976665790096971e-06,
      "loss": 0.5808,
      "step": 945
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0395978689193726,
      "learning_rate": 9.976566678768787e-06,
      "loss": 0.5415,
      "step": 946
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.148126244544983,
      "learning_rate": 9.976467357894693e-06,
      "loss": 0.5846,
      "step": 947
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.151173710823059,
      "learning_rate": 9.97636782747887e-06,
      "loss": 0.6315,
      "step": 948
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1681454181671143,
      "learning_rate": 9.976268087525509e-06,
      "loss": 0.6539,
      "step": 949
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.097345232963562,
      "learning_rate": 9.976168138038812e-06,
      "loss": 0.6086,
      "step": 950
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.06402587890625,
      "learning_rate": 9.976067979022983e-06,
      "loss": 0.6146,
      "step": 951
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0805296897888184,
      "learning_rate": 9.975967610482243e-06,
      "loss": 0.5981,
      "step": 952
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0387953519821167,
      "learning_rate": 9.975867032420816e-06,
      "loss": 0.5773,
      "step": 953
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0913708209991455,
      "learning_rate": 9.97576624484294e-06,
      "loss": 0.6334,
      "step": 954
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1604536771774292,
      "learning_rate": 9.975665247752855e-06,
      "loss": 0.6094,
      "step": 955
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.125901460647583,
      "learning_rate": 9.975564041154817e-06,
      "loss": 0.5946,
      "step": 956
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1342413425445557,
      "learning_rate": 9.975462625053085e-06,
      "loss": 0.6266,
      "step": 957
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1175763607025146,
      "learning_rate": 9.97536099945193e-06,
      "loss": 0.5808,
      "step": 958
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0494226217269897,
      "learning_rate": 9.975259164355632e-06,
      "loss": 0.6203,
      "step": 959
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1396679878234863,
      "learning_rate": 9.97515711976848e-06,
      "loss": 0.6191,
      "step": 960
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1368567943572998,
      "learning_rate": 9.975054865694767e-06,
      "loss": 0.6091,
      "step": 961
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1227906942367554,
      "learning_rate": 9.9749524021388e-06,
      "loss": 0.6355,
      "step": 962
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.091934323310852,
      "learning_rate": 9.974849729104894e-06,
      "loss": 0.6362,
      "step": 963
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.120595097541809,
      "learning_rate": 9.974746846597373e-06,
      "loss": 0.6677,
      "step": 964
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.110969066619873,
      "learning_rate": 9.974643754620567e-06,
      "loss": 0.6195,
      "step": 965
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.074960708618164,
      "learning_rate": 9.97454045317882e-06,
      "loss": 0.5969,
      "step": 966
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1078239679336548,
      "learning_rate": 9.974436942276477e-06,
      "loss": 0.5878,
      "step": 967
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0919568538665771,
      "learning_rate": 9.974333221917903e-06,
      "loss": 0.6343,
      "step": 968
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0447242259979248,
      "learning_rate": 9.974229292107458e-06,
      "loss": 0.5797,
      "step": 969
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1111451387405396,
      "learning_rate": 9.974125152849523e-06,
      "loss": 0.627,
      "step": 970
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1359940767288208,
      "learning_rate": 9.974020804148482e-06,
      "loss": 0.5759,
      "step": 971
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1420283317565918,
      "learning_rate": 9.973916246008727e-06,
      "loss": 0.5921,
      "step": 972
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0663940906524658,
      "learning_rate": 9.973811478434662e-06,
      "loss": 0.616,
      "step": 973
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1082817316055298,
      "learning_rate": 9.9737065014307e-06,
      "loss": 0.611,
      "step": 974
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0665665864944458,
      "learning_rate": 9.973601315001258e-06,
      "loss": 0.6119,
      "step": 975
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1018037796020508,
      "learning_rate": 9.973495919150766e-06,
      "loss": 0.6169,
      "step": 976
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0560823678970337,
      "learning_rate": 9.973390313883664e-06,
      "loss": 0.5525,
      "step": 977
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0385541915893555,
      "learning_rate": 9.973284499204396e-06,
      "loss": 0.5556,
      "step": 978
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1537460088729858,
      "learning_rate": 9.973178475117419e-06,
      "loss": 0.6003,
      "step": 979
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0920456647872925,
      "learning_rate": 9.973072241627196e-06,
      "loss": 0.6364,
      "step": 980
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1587941646575928,
      "learning_rate": 9.972965798738202e-06,
      "loss": 0.6213,
      "step": 981
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0415784120559692,
      "learning_rate": 9.972859146454917e-06,
      "loss": 0.5436,
      "step": 982
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1108245849609375,
      "learning_rate": 9.972752284781832e-06,
      "loss": 0.6119,
      "step": 983
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1475266218185425,
      "learning_rate": 9.97264521372345e-06,
      "loss": 0.5819,
      "step": 984
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.013890027999878,
      "learning_rate": 9.972537933284274e-06,
      "loss": 0.5579,
      "step": 985
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0834392309188843,
      "learning_rate": 9.972430443468826e-06,
      "loss": 0.617,
      "step": 986
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0244008302688599,
      "learning_rate": 9.972322744281628e-06,
      "loss": 0.537,
      "step": 987
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0993061065673828,
      "learning_rate": 9.972214835727218e-06,
      "loss": 0.6344,
      "step": 988
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1424047946929932,
      "learning_rate": 9.972106717810137e-06,
      "loss": 0.5968,
      "step": 989
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1044074296951294,
      "learning_rate": 9.97199839053494e-06,
      "loss": 0.6462,
      "step": 990
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1467039585113525,
      "learning_rate": 9.971889853906186e-06,
      "loss": 0.6162,
      "step": 991
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0741792917251587,
      "learning_rate": 9.971781107928447e-06,
      "loss": 0.556,
      "step": 992
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1318819522857666,
      "learning_rate": 9.9716721526063e-06,
      "loss": 0.5789,
      "step": 993
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0687830448150635,
      "learning_rate": 9.971562987944336e-06,
      "loss": 0.6205,
      "step": 994
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0508983135223389,
      "learning_rate": 9.971453613947147e-06,
      "loss": 0.6119,
      "step": 995
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.043202519416809,
      "learning_rate": 9.971344030619342e-06,
      "loss": 0.5733,
      "step": 996
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0849485397338867,
      "learning_rate": 9.971234237965534e-06,
      "loss": 0.5942,
      "step": 997
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1442948579788208,
      "learning_rate": 9.971124235990346e-06,
      "loss": 0.62,
      "step": 998
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1188938617706299,
      "learning_rate": 9.971014024698408e-06,
      "loss": 0.5871,
      "step": 999
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1602274179458618,
      "learning_rate": 9.970903604094365e-06,
      "loss": 0.6137,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1121653318405151,
      "learning_rate": 9.970792974182863e-06,
      "loss": 0.6155,
      "step": 1001
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0772852897644043,
      "learning_rate": 9.97068213496856e-06,
      "loss": 0.5483,
      "step": 1002
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1640626192092896,
      "learning_rate": 9.970571086456124e-06,
      "loss": 0.5862,
      "step": 1003
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1127629280090332,
      "learning_rate": 9.970459828650232e-06,
      "loss": 0.5686,
      "step": 1004
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2149871587753296,
      "learning_rate": 9.970348361555566e-06,
      "loss": 0.5699,
      "step": 1005
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1070051193237305,
      "learning_rate": 9.970236685176821e-06,
      "loss": 0.6147,
      "step": 1006
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0872983932495117,
      "learning_rate": 9.9701247995187e-06,
      "loss": 0.5657,
      "step": 1007
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1832880973815918,
      "learning_rate": 9.970012704585916e-06,
      "loss": 0.6123,
      "step": 1008
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1965460777282715,
      "learning_rate": 9.969900400383183e-06,
      "loss": 0.6197,
      "step": 1009
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1256316900253296,
      "learning_rate": 9.969787886915236e-06,
      "loss": 0.5836,
      "step": 1010
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.096758484840393,
      "learning_rate": 9.969675164186807e-06,
      "loss": 0.6016,
      "step": 1011
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0753967761993408,
      "learning_rate": 9.969562232202647e-06,
      "loss": 0.5918,
      "step": 1012
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.175142765045166,
      "learning_rate": 9.969449090967509e-06,
      "loss": 0.6166,
      "step": 1013
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0600279569625854,
      "learning_rate": 9.969335740486157e-06,
      "loss": 0.5064,
      "step": 1014
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0868827104568481,
      "learning_rate": 9.969222180763363e-06,
      "loss": 0.5997,
      "step": 1015
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.199386477470398,
      "learning_rate": 9.96910841180391e-06,
      "loss": 0.5554,
      "step": 1016
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.175727367401123,
      "learning_rate": 9.968994433612589e-06,
      "loss": 0.6348,
      "step": 1017
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0635806322097778,
      "learning_rate": 9.968880246194198e-06,
      "loss": 0.6146,
      "step": 1018
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1115741729736328,
      "learning_rate": 9.968765849553544e-06,
      "loss": 0.6164,
      "step": 1019
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.145827054977417,
      "learning_rate": 9.968651243695446e-06,
      "loss": 0.5657,
      "step": 1020
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1853764057159424,
      "learning_rate": 9.968536428624729e-06,
      "loss": 0.6122,
      "step": 1021
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1163513660430908,
      "learning_rate": 9.968421404346228e-06,
      "loss": 0.6144,
      "step": 1022
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1911154985427856,
      "learning_rate": 9.968306170864786e-06,
      "loss": 0.6815,
      "step": 1023
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1843432188034058,
      "learning_rate": 9.968190728185251e-06,
      "loss": 0.5854,
      "step": 1024
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0040591955184937,
      "learning_rate": 9.968075076312492e-06,
      "loss": 0.5802,
      "step": 1025
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1693812608718872,
      "learning_rate": 9.96795921525137e-06,
      "loss": 0.6038,
      "step": 1026
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.116028070449829,
      "learning_rate": 9.967843145006771e-06,
      "loss": 0.5952,
      "step": 1027
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.187366247177124,
      "learning_rate": 9.967726865583578e-06,
      "loss": 0.5987,
      "step": 1028
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1540720462799072,
      "learning_rate": 9.967610376986687e-06,
      "loss": 0.5899,
      "step": 1029
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3024755716323853,
      "learning_rate": 9.967493679221006e-06,
      "loss": 0.6473,
      "step": 1030
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0868054628372192,
      "learning_rate": 9.967376772291446e-06,
      "loss": 0.5728,
      "step": 1031
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.077175498008728,
      "learning_rate": 9.96725965620293e-06,
      "loss": 0.5796,
      "step": 1032
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1268564462661743,
      "learning_rate": 9.96714233096039e-06,
      "loss": 0.5815,
      "step": 1033
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1633411645889282,
      "learning_rate": 9.967024796568766e-06,
      "loss": 0.6436,
      "step": 1034
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1669288873672485,
      "learning_rate": 9.966907053033006e-06,
      "loss": 0.6238,
      "step": 1035
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.12864351272583,
      "learning_rate": 9.96678910035807e-06,
      "loss": 0.6005,
      "step": 1036
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0559660196304321,
      "learning_rate": 9.966670938548923e-06,
      "loss": 0.6212,
      "step": 1037
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.215419054031372,
      "learning_rate": 9.96655256761054e-06,
      "loss": 0.58,
      "step": 1038
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1690610647201538,
      "learning_rate": 9.966433987547906e-06,
      "loss": 0.5528,
      "step": 1039
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0882534980773926,
      "learning_rate": 9.966315198366011e-06,
      "loss": 0.6244,
      "step": 1040
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1955314874649048,
      "learning_rate": 9.966196200069863e-06,
      "loss": 0.6262,
      "step": 1041
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2565194368362427,
      "learning_rate": 9.966076992664469e-06,
      "loss": 0.6175,
      "step": 1042
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1895841360092163,
      "learning_rate": 9.965957576154848e-06,
      "loss": 0.5811,
      "step": 1043
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1305845975875854,
      "learning_rate": 9.96583795054603e-06,
      "loss": 0.6107,
      "step": 1044
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1758750677108765,
      "learning_rate": 9.965718115843048e-06,
      "loss": 0.6219,
      "step": 1045
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.077109456062317,
      "learning_rate": 9.965598072050953e-06,
      "loss": 0.6318,
      "step": 1046
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.064382553100586,
      "learning_rate": 9.965477819174796e-06,
      "loss": 0.5972,
      "step": 1047
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0851179361343384,
      "learning_rate": 9.96535735721964e-06,
      "loss": 0.5786,
      "step": 1048
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0553301572799683,
      "learning_rate": 9.965236686190563e-06,
      "loss": 0.6608,
      "step": 1049
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1511121988296509,
      "learning_rate": 9.965115806092638e-06,
      "loss": 0.5651,
      "step": 1050
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1849722862243652,
      "learning_rate": 9.96499471693096e-06,
      "loss": 0.5827,
      "step": 1051
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0399516820907593,
      "learning_rate": 9.964873418710628e-06,
      "loss": 0.5149,
      "step": 1052
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1090704202651978,
      "learning_rate": 9.964751911436748e-06,
      "loss": 0.5881,
      "step": 1053
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.10074782371521,
      "learning_rate": 9.964630195114432e-06,
      "loss": 0.5828,
      "step": 1054
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1069676876068115,
      "learning_rate": 9.964508269748814e-06,
      "loss": 0.6137,
      "step": 1055
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.06825590133667,
      "learning_rate": 9.96438613534502e-06,
      "loss": 0.5847,
      "step": 1056
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0860706567764282,
      "learning_rate": 9.964263791908198e-06,
      "loss": 0.5926,
      "step": 1057
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1424225568771362,
      "learning_rate": 9.964141239443497e-06,
      "loss": 0.557,
      "step": 1058
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1204310655593872,
      "learning_rate": 9.964018477956075e-06,
      "loss": 0.6212,
      "step": 1059
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.011555552482605,
      "learning_rate": 9.963895507451104e-06,
      "loss": 0.5961,
      "step": 1060
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.108712077140808,
      "learning_rate": 9.963772327933764e-06,
      "loss": 0.6081,
      "step": 1061
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1001176834106445,
      "learning_rate": 9.963648939409236e-06,
      "loss": 0.6096,
      "step": 1062
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0480148792266846,
      "learning_rate": 9.96352534188272e-06,
      "loss": 0.5524,
      "step": 1063
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0474708080291748,
      "learning_rate": 9.963401535359418e-06,
      "loss": 0.565,
      "step": 1064
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0563637018203735,
      "learning_rate": 9.963277519844544e-06,
      "loss": 0.5545,
      "step": 1065
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2024121284484863,
      "learning_rate": 9.963153295343319e-06,
      "loss": 0.6513,
      "step": 1066
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0970734357833862,
      "learning_rate": 9.963028861860975e-06,
      "loss": 0.616,
      "step": 1067
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1616771221160889,
      "learning_rate": 9.962904219402752e-06,
      "loss": 0.6491,
      "step": 1068
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1157522201538086,
      "learning_rate": 9.962779367973896e-06,
      "loss": 0.6063,
      "step": 1069
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.102754831314087,
      "learning_rate": 9.962654307579665e-06,
      "loss": 0.5964,
      "step": 1070
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0813874006271362,
      "learning_rate": 9.962529038225324e-06,
      "loss": 0.6397,
      "step": 1071
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0415786504745483,
      "learning_rate": 9.962403559916149e-06,
      "loss": 0.5828,
      "step": 1072
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.076107144355774,
      "learning_rate": 9.962277872657422e-06,
      "loss": 0.6111,
      "step": 1073
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0587031841278076,
      "learning_rate": 9.962151976454439e-06,
      "loss": 0.5755,
      "step": 1074
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2270575761795044,
      "learning_rate": 9.962025871312497e-06,
      "loss": 0.6434,
      "step": 1075
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1191643476486206,
      "learning_rate": 9.961899557236907e-06,
      "loss": 0.62,
      "step": 1076
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0456963777542114,
      "learning_rate": 9.961773034232987e-06,
      "loss": 0.582,
      "step": 1077
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.031314730644226,
      "learning_rate": 9.961646302306066e-06,
      "loss": 0.5834,
      "step": 1078
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.999323308467865,
      "learning_rate": 9.961519361461481e-06,
      "loss": 0.5689,
      "step": 1079
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1052522659301758,
      "learning_rate": 9.961392211704573e-06,
      "loss": 0.602,
      "step": 1080
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1391727924346924,
      "learning_rate": 9.9612648530407e-06,
      "loss": 0.5811,
      "step": 1081
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1579862833023071,
      "learning_rate": 9.961137285475223e-06,
      "loss": 0.5907,
      "step": 1082
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.085680365562439,
      "learning_rate": 9.961009509013512e-06,
      "loss": 0.6132,
      "step": 1083
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0278522968292236,
      "learning_rate": 9.96088152366095e-06,
      "loss": 0.6039,
      "step": 1084
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0606110095977783,
      "learning_rate": 9.960753329422925e-06,
      "loss": 0.5905,
      "step": 1085
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0981287956237793,
      "learning_rate": 9.960624926304834e-06,
      "loss": 0.5811,
      "step": 1086
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0174564123153687,
      "learning_rate": 9.960496314312085e-06,
      "loss": 0.5961,
      "step": 1087
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0941609144210815,
      "learning_rate": 9.96036749345009e-06,
      "loss": 0.5961,
      "step": 1088
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0558466911315918,
      "learning_rate": 9.960238463724278e-06,
      "loss": 0.5721,
      "step": 1089
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0658026933670044,
      "learning_rate": 9.96010922514008e-06,
      "loss": 0.5975,
      "step": 1090
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.031471848487854,
      "learning_rate": 9.959979777702935e-06,
      "loss": 0.5878,
      "step": 1091
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.068110704421997,
      "learning_rate": 9.959850121418298e-06,
      "loss": 0.5674,
      "step": 1092
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0213265419006348,
      "learning_rate": 9.959720256291626e-06,
      "loss": 0.5638,
      "step": 1093
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1408964395523071,
      "learning_rate": 9.959590182328387e-06,
      "loss": 0.621,
      "step": 1094
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.088255763053894,
      "learning_rate": 9.95945989953406e-06,
      "loss": 0.6143,
      "step": 1095
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1432324647903442,
      "learning_rate": 9.959329407914129e-06,
      "loss": 0.584,
      "step": 1096
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1209784746170044,
      "learning_rate": 9.959198707474087e-06,
      "loss": 0.5808,
      "step": 1097
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.054780125617981,
      "learning_rate": 9.959067798219442e-06,
      "loss": 0.6093,
      "step": 1098
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0760698318481445,
      "learning_rate": 9.958936680155702e-06,
      "loss": 0.5989,
      "step": 1099
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.042184591293335,
      "learning_rate": 9.958805353288388e-06,
      "loss": 0.5551,
      "step": 1100
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1320393085479736,
      "learning_rate": 9.958673817623033e-06,
      "loss": 0.5596,
      "step": 1101
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1195068359375,
      "learning_rate": 9.958542073165172e-06,
      "loss": 0.6096,
      "step": 1102
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1104859113693237,
      "learning_rate": 9.958410119920355e-06,
      "loss": 0.5914,
      "step": 1103
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.078932285308838,
      "learning_rate": 9.958277957894137e-06,
      "loss": 0.5668,
      "step": 1104
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0539746284484863,
      "learning_rate": 9.958145587092082e-06,
      "loss": 0.5268,
      "step": 1105
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1207184791564941,
      "learning_rate": 9.958013007519764e-06,
      "loss": 0.6298,
      "step": 1106
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.040810465812683,
      "learning_rate": 9.957880219182767e-06,
      "loss": 0.525,
      "step": 1107
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1117539405822754,
      "learning_rate": 9.957747222086682e-06,
      "loss": 0.6414,
      "step": 1108
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1745656728744507,
      "learning_rate": 9.957614016237106e-06,
      "loss": 0.627,
      "step": 1109
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0769981145858765,
      "learning_rate": 9.957480601639652e-06,
      "loss": 0.6363,
      "step": 1110
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.079752802848816,
      "learning_rate": 9.957346978299935e-06,
      "loss": 0.586,
      "step": 1111
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1357930898666382,
      "learning_rate": 9.957213146223581e-06,
      "loss": 0.607,
      "step": 1112
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0285855531692505,
      "learning_rate": 9.957079105416228e-06,
      "loss": 0.5827,
      "step": 1113
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.126513123512268,
      "learning_rate": 9.956944855883516e-06,
      "loss": 0.6163,
      "step": 1114
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0914958715438843,
      "learning_rate": 9.956810397631103e-06,
      "loss": 0.6203,
      "step": 1115
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1428629159927368,
      "learning_rate": 9.956675730664646e-06,
      "loss": 0.5632,
      "step": 1116
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1023683547973633,
      "learning_rate": 9.956540854989817e-06,
      "loss": 0.5896,
      "step": 1117
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1191551685333252,
      "learning_rate": 9.956405770612295e-06,
      "loss": 0.606,
      "step": 1118
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.077361822128296,
      "learning_rate": 9.956270477537768e-06,
      "loss": 0.6225,
      "step": 1119
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1346423625946045,
      "learning_rate": 9.956134975771934e-06,
      "loss": 0.622,
      "step": 1120
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2580161094665527,
      "learning_rate": 9.955999265320495e-06,
      "loss": 0.6432,
      "step": 1121
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0496320724487305,
      "learning_rate": 9.95586334618917e-06,
      "loss": 0.6011,
      "step": 1122
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.081202745437622,
      "learning_rate": 9.95572721838368e-06,
      "loss": 0.5669,
      "step": 1123
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.075972318649292,
      "learning_rate": 9.955590881909753e-06,
      "loss": 0.5716,
      "step": 1124
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0538547039031982,
      "learning_rate": 9.955454336773136e-06,
      "loss": 0.546,
      "step": 1125
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9653191566467285,
      "learning_rate": 9.955317582979575e-06,
      "loss": 0.5264,
      "step": 1126
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1157087087631226,
      "learning_rate": 9.95518062053483e-06,
      "loss": 0.5734,
      "step": 1127
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1833845376968384,
      "learning_rate": 9.955043449444665e-06,
      "loss": 0.6188,
      "step": 1128
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.036758303642273,
      "learning_rate": 9.95490606971486e-06,
      "loss": 0.5751,
      "step": 1129
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9968557357788086,
      "learning_rate": 9.954768481351196e-06,
      "loss": 0.5704,
      "step": 1130
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.187659740447998,
      "learning_rate": 9.954630684359468e-06,
      "loss": 0.5832,
      "step": 1131
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1751303672790527,
      "learning_rate": 9.954492678745477e-06,
      "loss": 0.601,
      "step": 1132
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.05357825756073,
      "learning_rate": 9.954354464515035e-06,
      "loss": 0.5443,
      "step": 1133
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0725443363189697,
      "learning_rate": 9.95421604167396e-06,
      "loss": 0.5463,
      "step": 1134
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2123327255249023,
      "learning_rate": 9.954077410228084e-06,
      "loss": 0.6595,
      "step": 1135
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.052491545677185,
      "learning_rate": 9.95393857018324e-06,
      "loss": 0.588,
      "step": 1136
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1696809530258179,
      "learning_rate": 9.95379952154528e-06,
      "loss": 0.571,
      "step": 1137
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3228956460952759,
      "learning_rate": 9.953660264320053e-06,
      "loss": 0.6355,
      "step": 1138
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.166395902633667,
      "learning_rate": 9.953520798513425e-06,
      "loss": 0.6844,
      "step": 1139
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0823599100112915,
      "learning_rate": 9.953381124131269e-06,
      "loss": 0.5826,
      "step": 1140
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1569372415542603,
      "learning_rate": 9.953241241179462e-06,
      "loss": 0.5752,
      "step": 1141
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0618703365325928,
      "learning_rate": 9.953101149663902e-06,
      "loss": 0.5845,
      "step": 1142
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0329277515411377,
      "learning_rate": 9.95296084959048e-06,
      "loss": 0.5383,
      "step": 1143
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1169556379318237,
      "learning_rate": 9.952820340965109e-06,
      "loss": 0.588,
      "step": 1144
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0022128820419312,
      "learning_rate": 9.952679623793702e-06,
      "loss": 0.5851,
      "step": 1145
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.05393385887146,
      "learning_rate": 9.952538698082185e-06,
      "loss": 0.601,
      "step": 1146
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0881694555282593,
      "learning_rate": 9.95239756383649e-06,
      "loss": 0.6145,
      "step": 1147
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.06650972366333,
      "learning_rate": 9.952256221062566e-06,
      "loss": 0.5194,
      "step": 1148
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.181684970855713,
      "learning_rate": 9.952114669766357e-06,
      "loss": 0.5987,
      "step": 1149
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0362557172775269,
      "learning_rate": 9.951972909953828e-06,
      "loss": 0.535,
      "step": 1150
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1614176034927368,
      "learning_rate": 9.951830941630946e-06,
      "loss": 0.6163,
      "step": 1151
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1739448308944702,
      "learning_rate": 9.951688764803689e-06,
      "loss": 0.6276,
      "step": 1152
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2421715259552002,
      "learning_rate": 9.951546379478044e-06,
      "loss": 0.6289,
      "step": 1153
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1939436197280884,
      "learning_rate": 9.951403785660005e-06,
      "loss": 0.6502,
      "step": 1154
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0989609956741333,
      "learning_rate": 9.95126098335558e-06,
      "loss": 0.63,
      "step": 1155
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0621522665023804,
      "learning_rate": 9.951117972570776e-06,
      "loss": 0.5724,
      "step": 1156
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1018246412277222,
      "learning_rate": 9.95097475331162e-06,
      "loss": 0.6329,
      "step": 1157
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0743603706359863,
      "learning_rate": 9.950831325584138e-06,
      "loss": 0.5943,
      "step": 1158
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0884958505630493,
      "learning_rate": 9.950687689394373e-06,
      "loss": 0.5876,
      "step": 1159
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0881589651107788,
      "learning_rate": 9.950543844748372e-06,
      "loss": 0.6191,
      "step": 1160
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0929887294769287,
      "learning_rate": 9.950399791652193e-06,
      "loss": 0.6043,
      "step": 1161
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2016053199768066,
      "learning_rate": 9.950255530111896e-06,
      "loss": 0.6424,
      "step": 1162
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0238940715789795,
      "learning_rate": 9.950111060133562e-06,
      "loss": 0.5765,
      "step": 1163
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9373612999916077,
      "learning_rate": 9.94996638172327e-06,
      "loss": 0.5776,
      "step": 1164
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0880025625228882,
      "learning_rate": 9.949821494887116e-06,
      "loss": 0.6174,
      "step": 1165
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.065089225769043,
      "learning_rate": 9.949676399631197e-06,
      "loss": 0.5716,
      "step": 1166
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1390105485916138,
      "learning_rate": 9.949531095961621e-06,
      "loss": 0.6253,
      "step": 1167
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0902137756347656,
      "learning_rate": 9.94938558388451e-06,
      "loss": 0.5996,
      "step": 1168
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.119095802307129,
      "learning_rate": 9.94923986340599e-06,
      "loss": 0.5923,
      "step": 1169
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0107195377349854,
      "learning_rate": 9.949093934532196e-06,
      "loss": 0.5501,
      "step": 1170
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0740958452224731,
      "learning_rate": 9.948947797269275e-06,
      "loss": 0.5829,
      "step": 1171
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0222264528274536,
      "learning_rate": 9.948801451623376e-06,
      "loss": 0.5853,
      "step": 1172
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1215888261795044,
      "learning_rate": 9.948654897600664e-06,
      "loss": 0.6162,
      "step": 1173
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0881133079528809,
      "learning_rate": 9.94850813520731e-06,
      "loss": 0.5895,
      "step": 1174
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0932121276855469,
      "learning_rate": 9.948361164449493e-06,
      "loss": 0.5604,
      "step": 1175
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.211357593536377,
      "learning_rate": 9.948213985333403e-06,
      "loss": 0.6113,
      "step": 1176
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2227258682250977,
      "learning_rate": 9.948066597865234e-06,
      "loss": 0.5909,
      "step": 1177
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0413223505020142,
      "learning_rate": 9.947919002051194e-06,
      "loss": 0.5908,
      "step": 1178
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1543649435043335,
      "learning_rate": 9.947771197897495e-06,
      "loss": 0.588,
      "step": 1179
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0861040353775024,
      "learning_rate": 9.947623185410366e-06,
      "loss": 0.5964,
      "step": 1180
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1817620992660522,
      "learning_rate": 9.947474964596036e-06,
      "loss": 0.6389,
      "step": 1181
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1223866939544678,
      "learning_rate": 9.947326535460744e-06,
      "loss": 0.5958,
      "step": 1182
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0928974151611328,
      "learning_rate": 9.947177898010745e-06,
      "loss": 0.5861,
      "step": 1183
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0815249681472778,
      "learning_rate": 9.947029052252293e-06,
      "loss": 0.5637,
      "step": 1184
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0170737504959106,
      "learning_rate": 9.946879998191656e-06,
      "loss": 0.5843,
      "step": 1185
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1782039403915405,
      "learning_rate": 9.946730735835112e-06,
      "loss": 0.6589,
      "step": 1186
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1698777675628662,
      "learning_rate": 9.946581265188947e-06,
      "loss": 0.5822,
      "step": 1187
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.050987958908081,
      "learning_rate": 9.946431586259451e-06,
      "loss": 0.5849,
      "step": 1188
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0946595668792725,
      "learning_rate": 9.946281699052928e-06,
      "loss": 0.6106,
      "step": 1189
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1560449600219727,
      "learning_rate": 9.946131603575691e-06,
      "loss": 0.6777,
      "step": 1190
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1142282485961914,
      "learning_rate": 9.945981299834058e-06,
      "loss": 0.606,
      "step": 1191
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2300156354904175,
      "learning_rate": 9.945830787834358e-06,
      "loss": 0.6271,
      "step": 1192
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1337618827819824,
      "learning_rate": 9.945680067582928e-06,
      "loss": 0.5671,
      "step": 1193
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0776329040527344,
      "learning_rate": 9.945529139086116e-06,
      "loss": 0.5696,
      "step": 1194
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1155674457550049,
      "learning_rate": 9.945378002350277e-06,
      "loss": 0.6009,
      "step": 1195
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.103833794593811,
      "learning_rate": 9.945226657381773e-06,
      "loss": 0.6125,
      "step": 1196
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9649964570999146,
      "learning_rate": 9.945075104186978e-06,
      "loss": 0.557,
      "step": 1197
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.988289475440979,
      "learning_rate": 9.944923342772272e-06,
      "loss": 0.5916,
      "step": 1198
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1717194318771362,
      "learning_rate": 9.944771373144047e-06,
      "loss": 0.5698,
      "step": 1199
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0416381359100342,
      "learning_rate": 9.944619195308701e-06,
      "loss": 0.5561,
      "step": 1200
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1190505027770996,
      "learning_rate": 9.944466809272642e-06,
      "loss": 0.5663,
      "step": 1201
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0151177644729614,
      "learning_rate": 9.944314215042286e-06,
      "loss": 0.5623,
      "step": 1202
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0351370573043823,
      "learning_rate": 9.944161412624059e-06,
      "loss": 0.5844,
      "step": 1203
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0102064609527588,
      "learning_rate": 9.944008402024395e-06,
      "loss": 0.6177,
      "step": 1204
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.182949423789978,
      "learning_rate": 9.943855183249734e-06,
      "loss": 0.6217,
      "step": 1205
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0727534294128418,
      "learning_rate": 9.94370175630653e-06,
      "loss": 0.6389,
      "step": 1206
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1393942832946777,
      "learning_rate": 9.943548121201243e-06,
      "loss": 0.6057,
      "step": 1207
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0908935070037842,
      "learning_rate": 9.943394277940344e-06,
      "loss": 0.5453,
      "step": 1208
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0910968780517578,
      "learning_rate": 9.943240226530306e-06,
      "loss": 0.6553,
      "step": 1209
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0977840423583984,
      "learning_rate": 9.94308596697762e-06,
      "loss": 0.5948,
      "step": 1210
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1505475044250488,
      "learning_rate": 9.942931499288779e-06,
      "loss": 0.6209,
      "step": 1211
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0492063760757446,
      "learning_rate": 9.942776823470288e-06,
      "loss": 0.5464,
      "step": 1212
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0802040100097656,
      "learning_rate": 9.94262193952866e-06,
      "loss": 0.5941,
      "step": 1213
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1597340106964111,
      "learning_rate": 9.942466847470415e-06,
      "loss": 0.594,
      "step": 1214
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9489971399307251,
      "learning_rate": 9.942311547302087e-06,
      "loss": 0.5147,
      "step": 1215
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1232295036315918,
      "learning_rate": 9.94215603903021e-06,
      "loss": 0.5816,
      "step": 1216
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1503745317459106,
      "learning_rate": 9.942000322661339e-06,
      "loss": 0.5968,
      "step": 1217
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1849077939987183,
      "learning_rate": 9.941844398202022e-06,
      "loss": 0.5556,
      "step": 1218
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0492712259292603,
      "learning_rate": 9.941688265658832e-06,
      "loss": 0.5506,
      "step": 1219
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1029094457626343,
      "learning_rate": 9.941531925038337e-06,
      "loss": 0.5889,
      "step": 1220
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1172541379928589,
      "learning_rate": 9.941375376347124e-06,
      "loss": 0.6159,
      "step": 1221
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0389809608459473,
      "learning_rate": 9.941218619591783e-06,
      "loss": 0.6011,
      "step": 1222
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.18142569065094,
      "learning_rate": 9.941061654778917e-06,
      "loss": 0.5498,
      "step": 1223
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2676929235458374,
      "learning_rate": 9.940904481915132e-06,
      "loss": 0.5948,
      "step": 1224
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0667624473571777,
      "learning_rate": 9.940747101007049e-06,
      "loss": 0.5855,
      "step": 1225
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.029423475265503,
      "learning_rate": 9.940589512061292e-06,
      "loss": 0.574,
      "step": 1226
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1424000263214111,
      "learning_rate": 9.940431715084498e-06,
      "loss": 0.5789,
      "step": 1227
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0252925157546997,
      "learning_rate": 9.94027371008331e-06,
      "loss": 0.5667,
      "step": 1228
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1843030452728271,
      "learning_rate": 9.940115497064383e-06,
      "loss": 0.6177,
      "step": 1229
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3379067182540894,
      "learning_rate": 9.939957076034379e-06,
      "loss": 0.6186,
      "step": 1230
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.186153769493103,
      "learning_rate": 9.939798446999965e-06,
      "loss": 0.5924,
      "step": 1231
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1364245414733887,
      "learning_rate": 9.939639609967825e-06,
      "loss": 0.617,
      "step": 1232
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0593096017837524,
      "learning_rate": 9.939480564944642e-06,
      "loss": 0.5803,
      "step": 1233
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1644937992095947,
      "learning_rate": 9.939321311937117e-06,
      "loss": 0.5769,
      "step": 1234
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1884273290634155,
      "learning_rate": 9.939161850951955e-06,
      "loss": 0.6122,
      "step": 1235
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0510512590408325,
      "learning_rate": 9.939002181995869e-06,
      "loss": 0.5604,
      "step": 1236
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0551425218582153,
      "learning_rate": 9.938842305075583e-06,
      "loss": 0.5719,
      "step": 1237
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1491656303405762,
      "learning_rate": 9.938682220197828e-06,
      "loss": 0.5445,
      "step": 1238
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.017390251159668,
      "learning_rate": 9.938521927369344e-06,
      "loss": 0.554,
      "step": 1239
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0788060426712036,
      "learning_rate": 9.938361426596883e-06,
      "loss": 0.5958,
      "step": 1240
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1188304424285889,
      "learning_rate": 9.938200717887202e-06,
      "loss": 0.5762,
      "step": 1241
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1995320320129395,
      "learning_rate": 9.938039801247066e-06,
      "loss": 0.5796,
      "step": 1242
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2387287616729736,
      "learning_rate": 9.937878676683254e-06,
      "loss": 0.6327,
      "step": 1243
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0643160343170166,
      "learning_rate": 9.937717344202548e-06,
      "loss": 0.5255,
      "step": 1244
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1715853214263916,
      "learning_rate": 9.93755580381174e-06,
      "loss": 0.6288,
      "step": 1245
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.027347445487976,
      "learning_rate": 9.937394055517635e-06,
      "loss": 0.591,
      "step": 1246
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1184815168380737,
      "learning_rate": 9.937232099327044e-06,
      "loss": 0.5302,
      "step": 1247
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.129241943359375,
      "learning_rate": 9.937069935246782e-06,
      "loss": 0.5922,
      "step": 1248
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2312674522399902,
      "learning_rate": 9.93690756328368e-06,
      "loss": 0.5578,
      "step": 1249
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.081891655921936,
      "learning_rate": 9.936744983444576e-06,
      "loss": 0.6104,
      "step": 1250
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0679914951324463,
      "learning_rate": 9.936582195736314e-06,
      "loss": 0.5763,
      "step": 1251
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1055591106414795,
      "learning_rate": 9.936419200165748e-06,
      "loss": 0.5504,
      "step": 1252
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1147511005401611,
      "learning_rate": 9.936255996739743e-06,
      "loss": 0.574,
      "step": 1253
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0842479467391968,
      "learning_rate": 9.93609258546517e-06,
      "loss": 0.6122,
      "step": 1254
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0599509477615356,
      "learning_rate": 9.93592896634891e-06,
      "loss": 0.6165,
      "step": 1255
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1376314163208008,
      "learning_rate": 9.93576513939785e-06,
      "loss": 0.5898,
      "step": 1256
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0422910451889038,
      "learning_rate": 9.935601104618892e-06,
      "loss": 0.6605,
      "step": 1257
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.023500919342041,
      "learning_rate": 9.93543686201894e-06,
      "loss": 0.5934,
      "step": 1258
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1389398574829102,
      "learning_rate": 9.935272411604913e-06,
      "loss": 0.6282,
      "step": 1259
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0828518867492676,
      "learning_rate": 9.935107753383733e-06,
      "loss": 0.5374,
      "step": 1260
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1542357206344604,
      "learning_rate": 9.93494288736233e-06,
      "loss": 0.6057,
      "step": 1261
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2249398231506348,
      "learning_rate": 9.934777813547653e-06,
      "loss": 0.5892,
      "step": 1262
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1498955488204956,
      "learning_rate": 9.934612531946648e-06,
      "loss": 0.5485,
      "step": 1263
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1004254817962646,
      "learning_rate": 9.934447042566275e-06,
      "loss": 0.5916,
      "step": 1264
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1617814302444458,
      "learning_rate": 9.934281345413504e-06,
      "loss": 0.6255,
      "step": 1265
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.06760835647583,
      "learning_rate": 9.934115440495311e-06,
      "loss": 0.6103,
      "step": 1266
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1368690729141235,
      "learning_rate": 9.93394932781868e-06,
      "loss": 0.6045,
      "step": 1267
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1010313034057617,
      "learning_rate": 9.933783007390608e-06,
      "loss": 0.6089,
      "step": 1268
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1142655611038208,
      "learning_rate": 9.933616479218095e-06,
      "loss": 0.576,
      "step": 1269
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0613747835159302,
      "learning_rate": 9.933449743308155e-06,
      "loss": 0.6072,
      "step": 1270
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1401052474975586,
      "learning_rate": 9.93328279966781e-06,
      "loss": 0.6172,
      "step": 1271
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1933995485305786,
      "learning_rate": 9.933115648304087e-06,
      "loss": 0.5495,
      "step": 1272
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9564380645751953,
      "learning_rate": 9.932948289224025e-06,
      "loss": 0.569,
      "step": 1273
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.006587028503418,
      "learning_rate": 9.932780722434671e-06,
      "loss": 0.5972,
      "step": 1274
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0362249612808228,
      "learning_rate": 9.932612947943084e-06,
      "loss": 0.5653,
      "step": 1275
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0872751474380493,
      "learning_rate": 9.932444965756321e-06,
      "loss": 0.5605,
      "step": 1276
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.135711431503296,
      "learning_rate": 9.93227677588146e-06,
      "loss": 0.6212,
      "step": 1277
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0566486120224,
      "learning_rate": 9.932108378325582e-06,
      "loss": 0.5669,
      "step": 1278
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.031200885772705,
      "learning_rate": 9.931939773095779e-06,
      "loss": 0.5683,
      "step": 1279
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1347169876098633,
      "learning_rate": 9.93177096019915e-06,
      "loss": 0.5966,
      "step": 1280
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0358705520629883,
      "learning_rate": 9.9316019396428e-06,
      "loss": 0.6257,
      "step": 1281
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0359333753585815,
      "learning_rate": 9.931432711433849e-06,
      "loss": 0.6045,
      "step": 1282
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1425514221191406,
      "learning_rate": 9.93126327557942e-06,
      "loss": 0.6403,
      "step": 1283
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1299494504928589,
      "learning_rate": 9.931093632086651e-06,
      "loss": 0.5855,
      "step": 1284
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0689961910247803,
      "learning_rate": 9.930923780962683e-06,
      "loss": 0.5565,
      "step": 1285
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1040059328079224,
      "learning_rate": 9.930753722214668e-06,
      "loss": 0.6126,
      "step": 1286
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9784469604492188,
      "learning_rate": 9.930583455849766e-06,
      "loss": 0.5312,
      "step": 1287
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1913020610809326,
      "learning_rate": 9.930412981875148e-06,
      "loss": 0.6228,
      "step": 1288
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.05826735496521,
      "learning_rate": 9.93024230029799e-06,
      "loss": 0.6485,
      "step": 1289
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.064443826675415,
      "learning_rate": 9.93007141112548e-06,
      "loss": 0.5932,
      "step": 1290
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.113539218902588,
      "learning_rate": 9.929900314364813e-06,
      "loss": 0.6224,
      "step": 1291
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0498301982879639,
      "learning_rate": 9.929729010023195e-06,
      "loss": 0.601,
      "step": 1292
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1142476797103882,
      "learning_rate": 9.929557498107836e-06,
      "loss": 0.6249,
      "step": 1293
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0876604318618774,
      "learning_rate": 9.929385778625959e-06,
      "loss": 0.6059,
      "step": 1294
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0598934888839722,
      "learning_rate": 9.929213851584798e-06,
      "loss": 0.6144,
      "step": 1295
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0487703084945679,
      "learning_rate": 9.929041716991587e-06,
      "loss": 0.5732,
      "step": 1296
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0446213483810425,
      "learning_rate": 9.928869374853576e-06,
      "loss": 0.5645,
      "step": 1297
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1094882488250732,
      "learning_rate": 9.928696825178021e-06,
      "loss": 0.5925,
      "step": 1298
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0140398740768433,
      "learning_rate": 9.92852406797219e-06,
      "loss": 0.6101,
      "step": 1299
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.152634859085083,
      "learning_rate": 9.928351103243356e-06,
      "loss": 0.6287,
      "step": 1300
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0498493909835815,
      "learning_rate": 9.928177930998801e-06,
      "loss": 0.5741,
      "step": 1301
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1100025177001953,
      "learning_rate": 9.928004551245818e-06,
      "loss": 0.5462,
      "step": 1302
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0770983695983887,
      "learning_rate": 9.927830963991704e-06,
      "loss": 0.581,
      "step": 1303
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0831111669540405,
      "learning_rate": 9.927657169243773e-06,
      "loss": 0.6322,
      "step": 1304
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0878232717514038,
      "learning_rate": 9.92748316700934e-06,
      "loss": 0.6321,
      "step": 1305
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.059654712677002,
      "learning_rate": 9.927308957295733e-06,
      "loss": 0.572,
      "step": 1306
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0426253080368042,
      "learning_rate": 9.927134540110286e-06,
      "loss": 0.5878,
      "step": 1307
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9689581394195557,
      "learning_rate": 9.926959915460344e-06,
      "loss": 0.5506,
      "step": 1308
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0300748348236084,
      "learning_rate": 9.926785083353258e-06,
      "loss": 0.5293,
      "step": 1309
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1261236667633057,
      "learning_rate": 9.926610043796394e-06,
      "loss": 0.5615,
      "step": 1310
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0079063177108765,
      "learning_rate": 9.926434796797117e-06,
      "loss": 0.561,
      "step": 1311
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2044016122817993,
      "learning_rate": 9.92625934236281e-06,
      "loss": 0.6302,
      "step": 1312
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.182726263999939,
      "learning_rate": 9.92608368050086e-06,
      "loss": 0.6028,
      "step": 1313
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0777925252914429,
      "learning_rate": 9.925907811218661e-06,
      "loss": 0.5818,
      "step": 1314
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1328976154327393,
      "learning_rate": 9.925731734523621e-06,
      "loss": 0.6168,
      "step": 1315
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0863475799560547,
      "learning_rate": 9.925555450423153e-06,
      "loss": 0.5425,
      "step": 1316
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1228018999099731,
      "learning_rate": 9.92537895892468e-06,
      "loss": 0.5735,
      "step": 1317
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.138264536857605,
      "learning_rate": 9.925202260035632e-06,
      "loss": 0.6271,
      "step": 1318
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0448517799377441,
      "learning_rate": 9.925025353763452e-06,
      "loss": 0.5501,
      "step": 1319
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.146152138710022,
      "learning_rate": 9.924848240115585e-06,
      "loss": 0.6045,
      "step": 1320
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.060131311416626,
      "learning_rate": 9.924670919099493e-06,
      "loss": 0.5421,
      "step": 1321
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1032346487045288,
      "learning_rate": 9.92449339072264e-06,
      "loss": 0.5954,
      "step": 1322
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.110339879989624,
      "learning_rate": 9.924315654992501e-06,
      "loss": 0.5808,
      "step": 1323
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0799542665481567,
      "learning_rate": 9.924137711916559e-06,
      "loss": 0.568,
      "step": 1324
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0312904119491577,
      "learning_rate": 9.92395956150231e-06,
      "loss": 0.5882,
      "step": 1325
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0492658615112305,
      "learning_rate": 9.923781203757253e-06,
      "loss": 0.5699,
      "step": 1326
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.170120358467102,
      "learning_rate": 9.923602638688897e-06,
      "loss": 0.5883,
      "step": 1327
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1021220684051514,
      "learning_rate": 9.923423866304761e-06,
      "loss": 0.5832,
      "step": 1328
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0523005723953247,
      "learning_rate": 9.923244886612375e-06,
      "loss": 0.6064,
      "step": 1329
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0522792339324951,
      "learning_rate": 9.923065699619273e-06,
      "loss": 0.5573,
      "step": 1330
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0156751871109009,
      "learning_rate": 9.922886305333e-06,
      "loss": 0.575,
      "step": 1331
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.085269808769226,
      "learning_rate": 9.922706703761111e-06,
      "loss": 0.5673,
      "step": 1332
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0672869682312012,
      "learning_rate": 9.922526894911166e-06,
      "loss": 0.567,
      "step": 1333
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0538702011108398,
      "learning_rate": 9.922346878790739e-06,
      "loss": 0.6103,
      "step": 1334
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0380349159240723,
      "learning_rate": 9.922166655407408e-06,
      "loss": 0.6059,
      "step": 1335
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0586581230163574,
      "learning_rate": 9.921986224768762e-06,
      "loss": 0.573,
      "step": 1336
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1432409286499023,
      "learning_rate": 9.9218055868824e-06,
      "loss": 0.6387,
      "step": 1337
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0543259382247925,
      "learning_rate": 9.921624741755924e-06,
      "loss": 0.5625,
      "step": 1338
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.185113549232483,
      "learning_rate": 9.921443689396952e-06,
      "loss": 0.5878,
      "step": 1339
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0658711194992065,
      "learning_rate": 9.921262429813107e-06,
      "loss": 0.539,
      "step": 1340
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1457328796386719,
      "learning_rate": 9.921080963012021e-06,
      "loss": 0.5799,
      "step": 1341
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0452888011932373,
      "learning_rate": 9.920899289001335e-06,
      "loss": 0.6373,
      "step": 1342
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0156432390213013,
      "learning_rate": 9.9207174077887e-06,
      "loss": 0.5544,
      "step": 1343
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1745818853378296,
      "learning_rate": 9.92053531938177e-06,
      "loss": 0.5478,
      "step": 1344
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.074892520904541,
      "learning_rate": 9.920353023788216e-06,
      "loss": 0.6117,
      "step": 1345
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.111828327178955,
      "learning_rate": 9.920170521015714e-06,
      "loss": 0.5599,
      "step": 1346
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.118725299835205,
      "learning_rate": 9.919987811071946e-06,
      "loss": 0.6519,
      "step": 1347
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1201225519180298,
      "learning_rate": 9.919804893964607e-06,
      "loss": 0.6582,
      "step": 1348
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.105054259300232,
      "learning_rate": 9.9196217697014e-06,
      "loss": 0.6288,
      "step": 1349
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.117285132408142,
      "learning_rate": 9.919438438290032e-06,
      "loss": 0.6296,
      "step": 1350
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.033210277557373,
      "learning_rate": 9.919254899738227e-06,
      "loss": 0.575,
      "step": 1351
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0252848863601685,
      "learning_rate": 9.91907115405371e-06,
      "loss": 0.5776,
      "step": 1352
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9571742415428162,
      "learning_rate": 9.918887201244219e-06,
      "loss": 0.5863,
      "step": 1353
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0897350311279297,
      "learning_rate": 9.918703041317498e-06,
      "loss": 0.6026,
      "step": 1354
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0628808736801147,
      "learning_rate": 9.918518674281305e-06,
      "loss": 0.5792,
      "step": 1355
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1064058542251587,
      "learning_rate": 9.9183341001434e-06,
      "loss": 0.6186,
      "step": 1356
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.124521255493164,
      "learning_rate": 9.918149318911557e-06,
      "loss": 0.5886,
      "step": 1357
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1114414930343628,
      "learning_rate": 9.917964330593553e-06,
      "loss": 0.5855,
      "step": 1358
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0328484773635864,
      "learning_rate": 9.917779135197181e-06,
      "loss": 0.5917,
      "step": 1359
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0707226991653442,
      "learning_rate": 9.917593732730236e-06,
      "loss": 0.5941,
      "step": 1360
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1221344470977783,
      "learning_rate": 9.917408123200527e-06,
      "loss": 0.6313,
      "step": 1361
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.060730218887329,
      "learning_rate": 9.917222306615868e-06,
      "loss": 0.5358,
      "step": 1362
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1551522016525269,
      "learning_rate": 9.917036282984084e-06,
      "loss": 0.6038,
      "step": 1363
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0436571836471558,
      "learning_rate": 9.916850052313007e-06,
      "loss": 0.617,
      "step": 1364
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9944663643836975,
      "learning_rate": 9.916663614610478e-06,
      "loss": 0.5487,
      "step": 1365
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0552699565887451,
      "learning_rate": 9.916476969884348e-06,
      "loss": 0.5983,
      "step": 1366
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0966901779174805,
      "learning_rate": 9.916290118142478e-06,
      "loss": 0.5831,
      "step": 1367
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.059079647064209,
      "learning_rate": 9.91610305939273e-06,
      "loss": 0.5876,
      "step": 1368
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0341980457305908,
      "learning_rate": 9.915915793642987e-06,
      "loss": 0.5768,
      "step": 1369
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1835851669311523,
      "learning_rate": 9.91572832090113e-06,
      "loss": 0.5718,
      "step": 1370
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1765505075454712,
      "learning_rate": 9.915540641175055e-06,
      "loss": 0.613,
      "step": 1371
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0547789335250854,
      "learning_rate": 9.915352754472662e-06,
      "loss": 0.5908,
      "step": 1372
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1082268953323364,
      "learning_rate": 9.915164660801865e-06,
      "loss": 0.5664,
      "step": 1373
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1055669784545898,
      "learning_rate": 9.914976360170583e-06,
      "loss": 0.5496,
      "step": 1374
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0965393781661987,
      "learning_rate": 9.914787852586744e-06,
      "loss": 0.5578,
      "step": 1375
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1463366746902466,
      "learning_rate": 9.914599138058285e-06,
      "loss": 0.5767,
      "step": 1376
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0385583639144897,
      "learning_rate": 9.914410216593154e-06,
      "loss": 0.5495,
      "step": 1377
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1779868602752686,
      "learning_rate": 9.914221088199304e-06,
      "loss": 0.5763,
      "step": 1378
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1094473600387573,
      "learning_rate": 9.9140317528847e-06,
      "loss": 0.5735,
      "step": 1379
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0603336095809937,
      "learning_rate": 9.913842210657314e-06,
      "loss": 0.585,
      "step": 1380
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1335699558258057,
      "learning_rate": 9.913652461525126e-06,
      "loss": 0.6302,
      "step": 1381
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.112284541130066,
      "learning_rate": 9.913462505496126e-06,
      "loss": 0.5903,
      "step": 1382
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1852585077285767,
      "learning_rate": 9.913272342578312e-06,
      "loss": 0.5635,
      "step": 1383
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.136877179145813,
      "learning_rate": 9.913081972779692e-06,
      "loss": 0.5878,
      "step": 1384
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0918060541152954,
      "learning_rate": 9.912891396108281e-06,
      "loss": 0.5938,
      "step": 1385
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0967477560043335,
      "learning_rate": 9.912700612572106e-06,
      "loss": 0.5435,
      "step": 1386
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1458653211593628,
      "learning_rate": 9.912509622179197e-06,
      "loss": 0.6112,
      "step": 1387
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0895942449569702,
      "learning_rate": 9.912318424937596e-06,
      "loss": 0.6049,
      "step": 1388
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0798182487487793,
      "learning_rate": 9.912127020855356e-06,
      "loss": 0.6021,
      "step": 1389
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1087937355041504,
      "learning_rate": 9.911935409940536e-06,
      "loss": 0.6122,
      "step": 1390
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1287997961044312,
      "learning_rate": 9.911743592201203e-06,
      "loss": 0.5944,
      "step": 1391
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0558440685272217,
      "learning_rate": 9.911551567645433e-06,
      "loss": 0.5626,
      "step": 1392
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1150189638137817,
      "learning_rate": 9.911359336281312e-06,
      "loss": 0.6096,
      "step": 1393
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1409717798233032,
      "learning_rate": 9.911166898116935e-06,
      "loss": 0.6173,
      "step": 1394
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0079151391983032,
      "learning_rate": 9.910974253160405e-06,
      "loss": 0.5734,
      "step": 1395
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0540634393692017,
      "learning_rate": 9.910781401419835e-06,
      "loss": 0.5882,
      "step": 1396
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.054129958152771,
      "learning_rate": 9.910588342903342e-06,
      "loss": 0.6047,
      "step": 1397
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0486072301864624,
      "learning_rate": 9.910395077619057e-06,
      "loss": 0.5836,
      "step": 1398
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0438059568405151,
      "learning_rate": 9.910201605575116e-06,
      "loss": 0.4912,
      "step": 1399
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0598132610321045,
      "learning_rate": 9.910007926779669e-06,
      "loss": 0.5759,
      "step": 1400
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0906940698623657,
      "learning_rate": 9.909814041240867e-06,
      "loss": 0.6319,
      "step": 1401
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1012886762619019,
      "learning_rate": 9.909619948966875e-06,
      "loss": 0.5676,
      "step": 1402
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0419704914093018,
      "learning_rate": 9.909425649965869e-06,
      "loss": 0.5474,
      "step": 1403
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0559276342391968,
      "learning_rate": 9.909231144246026e-06,
      "loss": 0.5799,
      "step": 1404
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1001919507980347,
      "learning_rate": 9.909036431815538e-06,
      "loss": 0.5837,
      "step": 1405
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1216944456100464,
      "learning_rate": 9.908841512682602e-06,
      "loss": 0.5731,
      "step": 1406
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.017932653427124,
      "learning_rate": 9.908646386855427e-06,
      "loss": 0.5706,
      "step": 1407
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0552278757095337,
      "learning_rate": 9.90845105434223e-06,
      "loss": 0.6122,
      "step": 1408
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1080493927001953,
      "learning_rate": 9.908255515151232e-06,
      "loss": 0.6026,
      "step": 1409
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0634651184082031,
      "learning_rate": 9.90805976929067e-06,
      "loss": 0.5535,
      "step": 1410
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1341966390609741,
      "learning_rate": 9.907863816768786e-06,
      "loss": 0.6083,
      "step": 1411
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.068042278289795,
      "learning_rate": 9.907667657593828e-06,
      "loss": 0.579,
      "step": 1412
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.038496732711792,
      "learning_rate": 9.907471291774058e-06,
      "loss": 0.5722,
      "step": 1413
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.072607159614563,
      "learning_rate": 9.907274719317746e-06,
      "loss": 0.5825,
      "step": 1414
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0538018941879272,
      "learning_rate": 9.907077940233162e-06,
      "loss": 0.5521,
      "step": 1415
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.216343879699707,
      "learning_rate": 9.906880954528601e-06,
      "loss": 0.6778,
      "step": 1416
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.142187237739563,
      "learning_rate": 9.90668376221235e-06,
      "loss": 0.6006,
      "step": 1417
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.062034010887146,
      "learning_rate": 9.906486363292718e-06,
      "loss": 0.5901,
      "step": 1418
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0838314294815063,
      "learning_rate": 9.906288757778012e-06,
      "loss": 0.6226,
      "step": 1419
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0943511724472046,
      "learning_rate": 9.906090945676552e-06,
      "loss": 0.571,
      "step": 1420
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1141382455825806,
      "learning_rate": 9.905892926996672e-06,
      "loss": 0.5907,
      "step": 1421
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9890863299369812,
      "learning_rate": 9.905694701746706e-06,
      "loss": 0.6038,
      "step": 1422
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0769665241241455,
      "learning_rate": 9.905496269935002e-06,
      "loss": 0.6017,
      "step": 1423
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9992973208427429,
      "learning_rate": 9.905297631569915e-06,
      "loss": 0.5853,
      "step": 1424
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0979317426681519,
      "learning_rate": 9.905098786659809e-06,
      "loss": 0.5775,
      "step": 1425
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0635846853256226,
      "learning_rate": 9.904899735213058e-06,
      "loss": 0.6112,
      "step": 1426
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0106208324432373,
      "learning_rate": 9.90470047723804e-06,
      "loss": 0.5671,
      "step": 1427
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9854536056518555,
      "learning_rate": 9.904501012743149e-06,
      "loss": 0.5015,
      "step": 1428
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1171709299087524,
      "learning_rate": 9.90430134173678e-06,
      "loss": 0.5806,
      "step": 1429
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0839475393295288,
      "learning_rate": 9.904101464227342e-06,
      "loss": 0.5587,
      "step": 1430
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0651477575302124,
      "learning_rate": 9.903901380223254e-06,
      "loss": 0.5947,
      "step": 1431
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.050728440284729,
      "learning_rate": 9.903701089732937e-06,
      "loss": 0.5744,
      "step": 1432
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0288559198379517,
      "learning_rate": 9.903500592764825e-06,
      "loss": 0.5407,
      "step": 1433
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1495304107666016,
      "learning_rate": 9.903299889327362e-06,
      "loss": 0.5824,
      "step": 1434
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.187200903892517,
      "learning_rate": 9.903098979428998e-06,
      "loss": 0.5627,
      "step": 1435
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1792247295379639,
      "learning_rate": 9.902897863078192e-06,
      "loss": 0.5171,
      "step": 1436
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0808206796646118,
      "learning_rate": 9.902696540283414e-06,
      "loss": 0.6191,
      "step": 1437
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.024139404296875,
      "learning_rate": 9.90249501105314e-06,
      "loss": 0.5325,
      "step": 1438
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0681065320968628,
      "learning_rate": 9.902293275395854e-06,
      "loss": 0.5886,
      "step": 1439
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2235044240951538,
      "learning_rate": 9.902091333320053e-06,
      "loss": 0.5487,
      "step": 1440
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1250619888305664,
      "learning_rate": 9.90188918483424e-06,
      "loss": 0.6068,
      "step": 1441
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2732850313186646,
      "learning_rate": 9.901686829946924e-06,
      "loss": 0.6125,
      "step": 1442
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1575543880462646,
      "learning_rate": 9.901484268666628e-06,
      "loss": 0.5941,
      "step": 1443
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.035243034362793,
      "learning_rate": 9.90128150100188e-06,
      "loss": 0.6241,
      "step": 1444
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.078845739364624,
      "learning_rate": 9.90107852696122e-06,
      "loss": 0.5723,
      "step": 1445
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1393600702285767,
      "learning_rate": 9.900875346553192e-06,
      "loss": 0.6079,
      "step": 1446
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1262428760528564,
      "learning_rate": 9.900671959786352e-06,
      "loss": 0.5905,
      "step": 1447
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2057019472122192,
      "learning_rate": 9.900468366669264e-06,
      "loss": 0.5814,
      "step": 1448
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1640501022338867,
      "learning_rate": 9.900264567210501e-06,
      "loss": 0.5562,
      "step": 1449
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9975053071975708,
      "learning_rate": 9.900060561418643e-06,
      "loss": 0.5484,
      "step": 1450
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0894286632537842,
      "learning_rate": 9.89985634930228e-06,
      "loss": 0.6026,
      "step": 1451
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0827438831329346,
      "learning_rate": 9.899651930870014e-06,
      "loss": 0.5798,
      "step": 1452
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.058447241783142,
      "learning_rate": 9.899447306130447e-06,
      "loss": 0.5445,
      "step": 1453
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3698556423187256,
      "learning_rate": 9.8992424750922e-06,
      "loss": 0.6093,
      "step": 1454
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1240925788879395,
      "learning_rate": 9.899037437763894e-06,
      "loss": 0.5647,
      "step": 1455
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1698404550552368,
      "learning_rate": 9.898832194154165e-06,
      "loss": 0.5791,
      "step": 1456
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0323277711868286,
      "learning_rate": 9.898626744271654e-06,
      "loss": 0.5268,
      "step": 1457
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0751768350601196,
      "learning_rate": 9.898421088125012e-06,
      "loss": 0.5584,
      "step": 1458
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.116385817527771,
      "learning_rate": 9.898215225722899e-06,
      "loss": 0.6126,
      "step": 1459
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0593472719192505,
      "learning_rate": 9.898009157073982e-06,
      "loss": 0.608,
      "step": 1460
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0496641397476196,
      "learning_rate": 9.897802882186938e-06,
      "loss": 0.5774,
      "step": 1461
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.032878041267395,
      "learning_rate": 9.897596401070452e-06,
      "loss": 0.5633,
      "step": 1462
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.102704644203186,
      "learning_rate": 9.89738971373322e-06,
      "loss": 0.5533,
      "step": 1463
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.116784691810608,
      "learning_rate": 9.897182820183944e-06,
      "loss": 0.6055,
      "step": 1464
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0715374946594238,
      "learning_rate": 9.896975720431334e-06,
      "loss": 0.5584,
      "step": 1465
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0144976377487183,
      "learning_rate": 9.896768414484115e-06,
      "loss": 0.5891,
      "step": 1466
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0812373161315918,
      "learning_rate": 9.896560902351009e-06,
      "loss": 0.5897,
      "step": 1467
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1514919996261597,
      "learning_rate": 9.89635318404076e-06,
      "loss": 0.6151,
      "step": 1468
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.000230073928833,
      "learning_rate": 9.896145259562111e-06,
      "loss": 0.5991,
      "step": 1469
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0490227937698364,
      "learning_rate": 9.895937128923816e-06,
      "loss": 0.5768,
      "step": 1470
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0367496013641357,
      "learning_rate": 9.895728792134642e-06,
      "loss": 0.5864,
      "step": 1471
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0676687955856323,
      "learning_rate": 9.895520249203358e-06,
      "loss": 0.6064,
      "step": 1472
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0376131534576416,
      "learning_rate": 9.895311500138749e-06,
      "loss": 0.5652,
      "step": 1473
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0196382999420166,
      "learning_rate": 9.8951025449496e-06,
      "loss": 0.5893,
      "step": 1474
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0555390119552612,
      "learning_rate": 9.894893383644713e-06,
      "loss": 0.5366,
      "step": 1475
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.116921067237854,
      "learning_rate": 9.894684016232893e-06,
      "loss": 0.5993,
      "step": 1476
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.143212914466858,
      "learning_rate": 9.894474442722956e-06,
      "loss": 0.6333,
      "step": 1477
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0427284240722656,
      "learning_rate": 9.89426466312373e-06,
      "loss": 0.5722,
      "step": 1478
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1189481019973755,
      "learning_rate": 9.89405467744404e-06,
      "loss": 0.6584,
      "step": 1479
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0095423460006714,
      "learning_rate": 9.893844485692736e-06,
      "loss": 0.5483,
      "step": 1480
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0975712537765503,
      "learning_rate": 9.893634087878665e-06,
      "loss": 0.5858,
      "step": 1481
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0731724500656128,
      "learning_rate": 9.893423484010685e-06,
      "loss": 0.5715,
      "step": 1482
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0699676275253296,
      "learning_rate": 9.893212674097666e-06,
      "loss": 0.5669,
      "step": 1483
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.993610680103302,
      "learning_rate": 9.893001658148482e-06,
      "loss": 0.5196,
      "step": 1484
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9979503750801086,
      "learning_rate": 9.892790436172022e-06,
      "loss": 0.5465,
      "step": 1485
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0339465141296387,
      "learning_rate": 9.892579008177176e-06,
      "loss": 0.595,
      "step": 1486
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0727976560592651,
      "learning_rate": 9.892367374172849e-06,
      "loss": 0.6281,
      "step": 1487
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.01498281955719,
      "learning_rate": 9.89215553416795e-06,
      "loss": 0.5745,
      "step": 1488
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0790021419525146,
      "learning_rate": 9.8919434881714e-06,
      "loss": 0.5655,
      "step": 1489
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9969554543495178,
      "learning_rate": 9.891731236192127e-06,
      "loss": 0.6151,
      "step": 1490
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0288242101669312,
      "learning_rate": 9.89151877823907e-06,
      "loss": 0.5744,
      "step": 1491
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0811976194381714,
      "learning_rate": 9.891306114321175e-06,
      "loss": 0.5768,
      "step": 1492
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0616317987442017,
      "learning_rate": 9.891093244447393e-06,
      "loss": 0.5762,
      "step": 1493
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1114658117294312,
      "learning_rate": 9.890880168626691e-06,
      "loss": 0.5799,
      "step": 1494
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.052419662475586,
      "learning_rate": 9.890666886868038e-06,
      "loss": 0.5779,
      "step": 1495
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.951212465763092,
      "learning_rate": 9.890453399180415e-06,
      "loss": 0.5546,
      "step": 1496
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1681593656539917,
      "learning_rate": 9.890239705572815e-06,
      "loss": 0.6047,
      "step": 1497
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0570625066757202,
      "learning_rate": 9.89002580605423e-06,
      "loss": 0.5864,
      "step": 1498
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1256753206253052,
      "learning_rate": 9.88981170063367e-06,
      "loss": 0.5726,
      "step": 1499
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0458052158355713,
      "learning_rate": 9.88959738932015e-06,
      "loss": 0.5125,
      "step": 1500
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1477687358856201,
      "learning_rate": 9.889382872122693e-06,
      "loss": 0.6334,
      "step": 1501
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1043277978897095,
      "learning_rate": 9.889168149050334e-06,
      "loss": 0.5889,
      "step": 1502
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1527658700942993,
      "learning_rate": 9.88895322011211e-06,
      "loss": 0.6627,
      "step": 1503
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1285358667373657,
      "learning_rate": 9.888738085317075e-06,
      "loss": 0.632,
      "step": 1504
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0251461267471313,
      "learning_rate": 9.888522744674286e-06,
      "loss": 0.5727,
      "step": 1505
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.096802830696106,
      "learning_rate": 9.888307198192808e-06,
      "loss": 0.5645,
      "step": 1506
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0713390111923218,
      "learning_rate": 9.888091445881723e-06,
      "loss": 0.5978,
      "step": 1507
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0598140954971313,
      "learning_rate": 9.887875487750108e-06,
      "loss": 0.5756,
      "step": 1508
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0787073373794556,
      "learning_rate": 9.887659323807062e-06,
      "loss": 0.5913,
      "step": 1509
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0909522771835327,
      "learning_rate": 9.887442954061684e-06,
      "loss": 0.6353,
      "step": 1510
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0643187761306763,
      "learning_rate": 9.887226378523085e-06,
      "loss": 0.578,
      "step": 1511
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0913480520248413,
      "learning_rate": 9.887009597200385e-06,
      "loss": 0.5661,
      "step": 1512
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0126315355300903,
      "learning_rate": 9.88679261010271e-06,
      "loss": 0.5045,
      "step": 1513
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0305216312408447,
      "learning_rate": 9.886575417239202e-06,
      "loss": 0.5679,
      "step": 1514
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0515673160552979,
      "learning_rate": 9.886358018619e-06,
      "loss": 0.5889,
      "step": 1515
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0206800699234009,
      "learning_rate": 9.886140414251259e-06,
      "loss": 0.5633,
      "step": 1516
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1397572755813599,
      "learning_rate": 9.885922604145143e-06,
      "loss": 0.6153,
      "step": 1517
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0395382642745972,
      "learning_rate": 9.885704588309825e-06,
      "loss": 0.5592,
      "step": 1518
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1337170600891113,
      "learning_rate": 9.885486366754482e-06,
      "loss": 0.6303,
      "step": 1519
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.12993323802948,
      "learning_rate": 9.885267939488303e-06,
      "loss": 0.6014,
      "step": 1520
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0222861766815186,
      "learning_rate": 9.885049306520487e-06,
      "loss": 0.5193,
      "step": 1521
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1298507452011108,
      "learning_rate": 9.884830467860238e-06,
      "loss": 0.5602,
      "step": 1522
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1178910732269287,
      "learning_rate": 9.88461142351677e-06,
      "loss": 0.6111,
      "step": 1523
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9981340765953064,
      "learning_rate": 9.884392173499308e-06,
      "loss": 0.5449,
      "step": 1524
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1069059371948242,
      "learning_rate": 9.884172717817085e-06,
      "loss": 0.601,
      "step": 1525
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0417429208755493,
      "learning_rate": 9.883953056479336e-06,
      "loss": 0.5891,
      "step": 1526
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.144789457321167,
      "learning_rate": 9.883733189495316e-06,
      "loss": 0.6119,
      "step": 1527
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0751010179519653,
      "learning_rate": 9.88351311687428e-06,
      "loss": 0.5567,
      "step": 1528
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0349299907684326,
      "learning_rate": 9.883292838625495e-06,
      "loss": 0.5613,
      "step": 1529
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0517909526824951,
      "learning_rate": 9.883072354758237e-06,
      "loss": 0.5476,
      "step": 1530
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.044845461845398,
      "learning_rate": 9.88285166528179e-06,
      "loss": 0.6208,
      "step": 1531
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0600078105926514,
      "learning_rate": 9.882630770205444e-06,
      "loss": 0.5358,
      "step": 1532
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0708039999008179,
      "learning_rate": 9.882409669538503e-06,
      "loss": 0.5602,
      "step": 1533
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1167932748794556,
      "learning_rate": 9.882188363290273e-06,
      "loss": 0.5941,
      "step": 1534
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1405528783798218,
      "learning_rate": 9.881966851470077e-06,
      "loss": 0.5814,
      "step": 1535
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0730372667312622,
      "learning_rate": 9.881745134087239e-06,
      "loss": 0.5897,
      "step": 1536
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1045047044754028,
      "learning_rate": 9.881523211151097e-06,
      "loss": 0.582,
      "step": 1537
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1381570100784302,
      "learning_rate": 9.881301082670992e-06,
      "loss": 0.5851,
      "step": 1538
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1380250453948975,
      "learning_rate": 9.881078748656282e-06,
      "loss": 0.6164,
      "step": 1539
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0847206115722656,
      "learning_rate": 9.880856209116324e-06,
      "loss": 0.562,
      "step": 1540
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1043504476547241,
      "learning_rate": 9.880633464060492e-06,
      "loss": 0.5713,
      "step": 1541
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0896272659301758,
      "learning_rate": 9.880410513498163e-06,
      "loss": 0.5626,
      "step": 1542
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1659765243530273,
      "learning_rate": 9.880187357438722e-06,
      "loss": 0.5996,
      "step": 1543
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1116617918014526,
      "learning_rate": 9.87996399589157e-06,
      "loss": 0.5873,
      "step": 1544
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0397089719772339,
      "learning_rate": 9.87974042886611e-06,
      "loss": 0.5574,
      "step": 1545
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.982353925704956,
      "learning_rate": 9.879516656371758e-06,
      "loss": 0.5099,
      "step": 1546
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1460829973220825,
      "learning_rate": 9.879292678417934e-06,
      "loss": 0.6181,
      "step": 1547
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0767769813537598,
      "learning_rate": 9.879068495014068e-06,
      "loss": 0.5771,
      "step": 1548
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0753417015075684,
      "learning_rate": 9.878844106169601e-06,
      "loss": 0.6469,
      "step": 1549
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1642966270446777,
      "learning_rate": 9.87861951189398e-06,
      "loss": 0.6111,
      "step": 1550
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0993573665618896,
      "learning_rate": 9.878394712196665e-06,
      "loss": 0.6171,
      "step": 1551
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0451866388320923,
      "learning_rate": 9.878169707087116e-06,
      "loss": 0.5912,
      "step": 1552
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9942476749420166,
      "learning_rate": 9.877944496574813e-06,
      "loss": 0.5278,
      "step": 1553
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0837749242782593,
      "learning_rate": 9.877719080669235e-06,
      "loss": 0.5482,
      "step": 1554
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.187171220779419,
      "learning_rate": 9.877493459379876e-06,
      "loss": 0.6004,
      "step": 1555
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1294541358947754,
      "learning_rate": 9.877267632716235e-06,
      "loss": 0.6328,
      "step": 1556
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1029921770095825,
      "learning_rate": 9.87704160068782e-06,
      "loss": 0.5721,
      "step": 1557
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1500641107559204,
      "learning_rate": 9.87681536330415e-06,
      "loss": 0.5731,
      "step": 1558
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.218934416770935,
      "learning_rate": 9.87658892057475e-06,
      "loss": 0.6042,
      "step": 1559
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0704762935638428,
      "learning_rate": 9.876362272509154e-06,
      "loss": 0.6435,
      "step": 1560
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.109381914138794,
      "learning_rate": 9.876135419116908e-06,
      "loss": 0.5703,
      "step": 1561
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0808093547821045,
      "learning_rate": 9.87590836040756e-06,
      "loss": 0.5565,
      "step": 1562
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0995346307754517,
      "learning_rate": 9.875681096390676e-06,
      "loss": 0.5791,
      "step": 1563
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0648128986358643,
      "learning_rate": 9.87545362707582e-06,
      "loss": 0.5752,
      "step": 1564
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0962284803390503,
      "learning_rate": 9.875225952472574e-06,
      "loss": 0.5804,
      "step": 1565
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.032683253288269,
      "learning_rate": 9.874998072590521e-06,
      "loss": 0.6071,
      "step": 1566
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0552096366882324,
      "learning_rate": 9.874769987439259e-06,
      "loss": 0.611,
      "step": 1567
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.04661226272583,
      "learning_rate": 9.87454169702839e-06,
      "loss": 0.5924,
      "step": 1568
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.108020544052124,
      "learning_rate": 9.87431320136753e-06,
      "loss": 0.5242,
      "step": 1569
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.098593831062317,
      "learning_rate": 9.874084500466295e-06,
      "loss": 0.5901,
      "step": 1570
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0470693111419678,
      "learning_rate": 9.873855594334319e-06,
      "loss": 0.6177,
      "step": 1571
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0568416118621826,
      "learning_rate": 9.873626482981238e-06,
      "loss": 0.5679,
      "step": 1572
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.067253589630127,
      "learning_rate": 9.873397166416698e-06,
      "loss": 0.5495,
      "step": 1573
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.996239960193634,
      "learning_rate": 9.87316764465036e-06,
      "loss": 0.5381,
      "step": 1574
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1866554021835327,
      "learning_rate": 9.872937917691883e-06,
      "loss": 0.5227,
      "step": 1575
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0590612888336182,
      "learning_rate": 9.872707985550942e-06,
      "loss": 0.5189,
      "step": 1576
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1451884508132935,
      "learning_rate": 9.872477848237221e-06,
      "loss": 0.588,
      "step": 1577
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1452605724334717,
      "learning_rate": 9.872247505760405e-06,
      "loss": 0.5537,
      "step": 1578
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2145410776138306,
      "learning_rate": 9.872016958130197e-06,
      "loss": 0.5707,
      "step": 1579
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0972182750701904,
      "learning_rate": 9.871786205356303e-06,
      "loss": 0.5754,
      "step": 1580
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0583133697509766,
      "learning_rate": 9.871555247448442e-06,
      "loss": 0.6119,
      "step": 1581
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1397663354873657,
      "learning_rate": 9.871324084416332e-06,
      "loss": 0.5531,
      "step": 1582
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2657743692398071,
      "learning_rate": 9.871092716269714e-06,
      "loss": 0.5982,
      "step": 1583
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0894304513931274,
      "learning_rate": 9.870861143018327e-06,
      "loss": 0.5701,
      "step": 1584
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0850940942764282,
      "learning_rate": 9.87062936467192e-06,
      "loss": 0.5933,
      "step": 1585
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.080039381980896,
      "learning_rate": 9.870397381240256e-06,
      "loss": 0.5614,
      "step": 1586
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0982593297958374,
      "learning_rate": 9.870165192733101e-06,
      "loss": 0.5794,
      "step": 1587
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0418531894683838,
      "learning_rate": 9.869932799160232e-06,
      "loss": 0.6145,
      "step": 1588
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0308923721313477,
      "learning_rate": 9.869700200531431e-06,
      "loss": 0.5992,
      "step": 1589
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0501437187194824,
      "learning_rate": 9.869467396856499e-06,
      "loss": 0.5468,
      "step": 1590
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.054578423500061,
      "learning_rate": 9.869234388145232e-06,
      "loss": 0.6308,
      "step": 1591
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2149218320846558,
      "learning_rate": 9.869001174407444e-06,
      "loss": 0.5391,
      "step": 1592
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9719024300575256,
      "learning_rate": 9.868767755652955e-06,
      "loss": 0.5596,
      "step": 1593
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.015842080116272,
      "learning_rate": 9.868534131891594e-06,
      "loss": 0.5588,
      "step": 1594
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0591557025909424,
      "learning_rate": 9.868300303133195e-06,
      "loss": 0.6244,
      "step": 1595
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.069869875907898,
      "learning_rate": 9.868066269387609e-06,
      "loss": 0.5485,
      "step": 1596
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0887794494628906,
      "learning_rate": 9.867832030664685e-06,
      "loss": 0.5765,
      "step": 1597
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2797043323516846,
      "learning_rate": 9.867597586974288e-06,
      "loss": 0.607,
      "step": 1598
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.023294448852539,
      "learning_rate": 9.86736293832629e-06,
      "loss": 0.5669,
      "step": 1599
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0748317241668701,
      "learning_rate": 9.86712808473057e-06,
      "loss": 0.6009,
      "step": 1600
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0411274433135986,
      "learning_rate": 9.86689302619702e-06,
      "loss": 0.5721,
      "step": 1601
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.059205412864685,
      "learning_rate": 9.866657762735534e-06,
      "loss": 0.5968,
      "step": 1602
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.054073452949524,
      "learning_rate": 9.866422294356019e-06,
      "loss": 0.6236,
      "step": 1603
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9893490076065063,
      "learning_rate": 9.866186621068391e-06,
      "loss": 0.5158,
      "step": 1604
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1270335912704468,
      "learning_rate": 9.865950742882574e-06,
      "loss": 0.548,
      "step": 1605
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.09773588180542,
      "learning_rate": 9.865714659808497e-06,
      "loss": 0.5779,
      "step": 1606
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0233718156814575,
      "learning_rate": 9.865478371856102e-06,
      "loss": 0.5779,
      "step": 1607
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0285214185714722,
      "learning_rate": 9.86524187903534e-06,
      "loss": 0.6001,
      "step": 1608
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0676337480545044,
      "learning_rate": 9.865005181356166e-06,
      "loss": 0.587,
      "step": 1609
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0382795333862305,
      "learning_rate": 9.864768278828548e-06,
      "loss": 0.5448,
      "step": 1610
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0121814012527466,
      "learning_rate": 9.864531171462462e-06,
      "loss": 0.5373,
      "step": 1611
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0314677953720093,
      "learning_rate": 9.86429385926789e-06,
      "loss": 0.6135,
      "step": 1612
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1360394954681396,
      "learning_rate": 9.864056342254827e-06,
      "loss": 0.5943,
      "step": 1613
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.08867609500885,
      "learning_rate": 9.86381862043327e-06,
      "loss": 0.6075,
      "step": 1614
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1060529947280884,
      "learning_rate": 9.863580693813232e-06,
      "loss": 0.5825,
      "step": 1615
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0926846265792847,
      "learning_rate": 9.86334256240473e-06,
      "loss": 0.598,
      "step": 1616
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.007134199142456,
      "learning_rate": 9.86310422621779e-06,
      "loss": 0.5843,
      "step": 1617
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.060012698173523,
      "learning_rate": 9.86286568526245e-06,
      "loss": 0.6095,
      "step": 1618
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0170177221298218,
      "learning_rate": 9.862626939548751e-06,
      "loss": 0.6206,
      "step": 1619
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.03775155544281,
      "learning_rate": 9.862387989086749e-06,
      "loss": 0.5696,
      "step": 1620
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.032655119895935,
      "learning_rate": 9.862148833886504e-06,
      "loss": 0.5922,
      "step": 1621
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0248726606369019,
      "learning_rate": 9.861909473958084e-06,
      "loss": 0.5668,
      "step": 1622
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0710545778274536,
      "learning_rate": 9.861669909311571e-06,
      "loss": 0.567,
      "step": 1623
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9834234118461609,
      "learning_rate": 9.861430139957052e-06,
      "loss": 0.5864,
      "step": 1624
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1512020826339722,
      "learning_rate": 9.861190165904617e-06,
      "loss": 0.6048,
      "step": 1625
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1149070262908936,
      "learning_rate": 9.860949987164379e-06,
      "loss": 0.5597,
      "step": 1626
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0812784433364868,
      "learning_rate": 9.860709603746445e-06,
      "loss": 0.592,
      "step": 1627
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0941449403762817,
      "learning_rate": 9.86046901566094e-06,
      "loss": 0.575,
      "step": 1628
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1726043224334717,
      "learning_rate": 9.860228222917992e-06,
      "loss": 0.5678,
      "step": 1629
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0753968954086304,
      "learning_rate": 9.859987225527742e-06,
      "loss": 0.5704,
      "step": 1630
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.135108470916748,
      "learning_rate": 9.859746023500337e-06,
      "loss": 0.6198,
      "step": 1631
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0584384202957153,
      "learning_rate": 9.85950461684593e-06,
      "loss": 0.5889,
      "step": 1632
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0583887100219727,
      "learning_rate": 9.85926300557469e-06,
      "loss": 0.5879,
      "step": 1633
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0071109533309937,
      "learning_rate": 9.85902118969679e-06,
      "loss": 0.5767,
      "step": 1634
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0471796989440918,
      "learning_rate": 9.85877916922241e-06,
      "loss": 0.5655,
      "step": 1635
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1224594116210938,
      "learning_rate": 9.858536944161743e-06,
      "loss": 0.6142,
      "step": 1636
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.083262324333191,
      "learning_rate": 9.858294514524987e-06,
      "loss": 0.5738,
      "step": 1637
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0365420579910278,
      "learning_rate": 9.858051880322347e-06,
      "loss": 0.5379,
      "step": 1638
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0462414026260376,
      "learning_rate": 9.857809041564044e-06,
      "loss": 0.592,
      "step": 1639
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0216164588928223,
      "learning_rate": 9.857565998260302e-06,
      "loss": 0.6077,
      "step": 1640
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0120216608047485,
      "learning_rate": 9.857322750421353e-06,
      "loss": 0.5515,
      "step": 1641
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0469900369644165,
      "learning_rate": 9.857079298057442e-06,
      "loss": 0.5897,
      "step": 1642
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0659940242767334,
      "learning_rate": 9.856835641178816e-06,
      "loss": 0.6174,
      "step": 1643
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0688961744308472,
      "learning_rate": 9.856591779795738e-06,
      "loss": 0.584,
      "step": 1644
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0651742219924927,
      "learning_rate": 9.856347713918475e-06,
      "loss": 0.6084,
      "step": 1645
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9851531386375427,
      "learning_rate": 9.856103443557304e-06,
      "loss": 0.5539,
      "step": 1646
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1691269874572754,
      "learning_rate": 9.85585896872251e-06,
      "loss": 0.5984,
      "step": 1647
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0338088274002075,
      "learning_rate": 9.855614289424386e-06,
      "loss": 0.5695,
      "step": 1648
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0682861804962158,
      "learning_rate": 9.855369405673236e-06,
      "loss": 0.5168,
      "step": 1649
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0562790632247925,
      "learning_rate": 9.855124317479372e-06,
      "loss": 0.5706,
      "step": 1650
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1324230432510376,
      "learning_rate": 9.854879024853113e-06,
      "loss": 0.5845,
      "step": 1651
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1204321384429932,
      "learning_rate": 9.854633527804787e-06,
      "loss": 0.5954,
      "step": 1652
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9424839019775391,
      "learning_rate": 9.85438782634473e-06,
      "loss": 0.5087,
      "step": 1653
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9705440998077393,
      "learning_rate": 9.854141920483289e-06,
      "loss": 0.5626,
      "step": 1654
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1535807847976685,
      "learning_rate": 9.853895810230818e-06,
      "loss": 0.5677,
      "step": 1655
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1456506252288818,
      "learning_rate": 9.853649495597682e-06,
      "loss": 0.6238,
      "step": 1656
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.103397011756897,
      "learning_rate": 9.853402976594248e-06,
      "loss": 0.6069,
      "step": 1657
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0972193479537964,
      "learning_rate": 9.8531562532309e-06,
      "loss": 0.5484,
      "step": 1658
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0873202085494995,
      "learning_rate": 9.852909325518022e-06,
      "loss": 0.6027,
      "step": 1659
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.017956256866455,
      "learning_rate": 9.852662193466019e-06,
      "loss": 0.5701,
      "step": 1660
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.097472906112671,
      "learning_rate": 9.852414857085288e-06,
      "loss": 0.6403,
      "step": 1661
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1169770956039429,
      "learning_rate": 9.85216731638625e-06,
      "loss": 0.5763,
      "step": 1662
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0574263334274292,
      "learning_rate": 9.851919571379326e-06,
      "loss": 0.6297,
      "step": 1663
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0826281309127808,
      "learning_rate": 9.851671622074947e-06,
      "loss": 0.5682,
      "step": 1664
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1474690437316895,
      "learning_rate": 9.851423468483554e-06,
      "loss": 0.6094,
      "step": 1665
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1534571647644043,
      "learning_rate": 9.851175110615594e-06,
      "loss": 0.5387,
      "step": 1666
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0650722980499268,
      "learning_rate": 9.850926548481528e-06,
      "loss": 0.5899,
      "step": 1667
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1716303825378418,
      "learning_rate": 9.850677782091818e-06,
      "loss": 0.5806,
      "step": 1668
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.033402442932129,
      "learning_rate": 9.850428811456943e-06,
      "loss": 0.5319,
      "step": 1669
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.113662600517273,
      "learning_rate": 9.850179636587383e-06,
      "loss": 0.5838,
      "step": 1670
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1086348295211792,
      "learning_rate": 9.849930257493632e-06,
      "loss": 0.5578,
      "step": 1671
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1946237087249756,
      "learning_rate": 9.849680674186188e-06,
      "loss": 0.6278,
      "step": 1672
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0539735555648804,
      "learning_rate": 9.849430886675564e-06,
      "loss": 0.5906,
      "step": 1673
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0256834030151367,
      "learning_rate": 9.849180894972272e-06,
      "loss": 0.5803,
      "step": 1674
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9212612509727478,
      "learning_rate": 9.848930699086846e-06,
      "loss": 0.577,
      "step": 1675
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0151751041412354,
      "learning_rate": 9.848680299029813e-06,
      "loss": 0.5676,
      "step": 1676
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0259836912155151,
      "learning_rate": 9.848429694811721e-06,
      "loss": 0.528,
      "step": 1677
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1119564771652222,
      "learning_rate": 9.84817888644312e-06,
      "loss": 0.5951,
      "step": 1678
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.052847981452942,
      "learning_rate": 9.847927873934573e-06,
      "loss": 0.5732,
      "step": 1679
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1761375665664673,
      "learning_rate": 9.847676657296647e-06,
      "loss": 0.5885,
      "step": 1680
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.073297142982483,
      "learning_rate": 9.847425236539922e-06,
      "loss": 0.5852,
      "step": 1681
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0152772665023804,
      "learning_rate": 9.847173611674982e-06,
      "loss": 0.5493,
      "step": 1682
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1353917121887207,
      "learning_rate": 9.846921782712424e-06,
      "loss": 0.5567,
      "step": 1683
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0314478874206543,
      "learning_rate": 9.846669749662851e-06,
      "loss": 0.5533,
      "step": 1684
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0658608675003052,
      "learning_rate": 9.846417512536874e-06,
      "loss": 0.5625,
      "step": 1685
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.116602897644043,
      "learning_rate": 9.846165071345118e-06,
      "loss": 0.6012,
      "step": 1686
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1800258159637451,
      "learning_rate": 9.845912426098206e-06,
      "loss": 0.6041,
      "step": 1687
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0146981477737427,
      "learning_rate": 9.845659576806781e-06,
      "loss": 0.5607,
      "step": 1688
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0185143947601318,
      "learning_rate": 9.845406523481488e-06,
      "loss": 0.5872,
      "step": 1689
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0636152029037476,
      "learning_rate": 9.845153266132981e-06,
      "loss": 0.54,
      "step": 1690
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0499494075775146,
      "learning_rate": 9.844899804771927e-06,
      "loss": 0.5694,
      "step": 1691
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1946227550506592,
      "learning_rate": 9.844646139408995e-06,
      "loss": 0.593,
      "step": 1692
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9941321015357971,
      "learning_rate": 9.844392270054868e-06,
      "loss": 0.5635,
      "step": 1693
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.088547706604004,
      "learning_rate": 9.844138196720236e-06,
      "loss": 0.615,
      "step": 1694
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0187320709228516,
      "learning_rate": 9.843883919415795e-06,
      "loss": 0.6168,
      "step": 1695
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0746089220046997,
      "learning_rate": 9.843629438152252e-06,
      "loss": 0.5816,
      "step": 1696
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0524471998214722,
      "learning_rate": 9.843374752940323e-06,
      "loss": 0.5683,
      "step": 1697
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0667989253997803,
      "learning_rate": 9.843119863790733e-06,
      "loss": 0.6443,
      "step": 1698
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1471716165542603,
      "learning_rate": 9.842864770714213e-06,
      "loss": 0.5554,
      "step": 1699
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.115005612373352,
      "learning_rate": 9.842609473721505e-06,
      "loss": 0.5617,
      "step": 1700
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0730042457580566,
      "learning_rate": 9.842353972823358e-06,
      "loss": 0.5653,
      "step": 1701
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0105443000793457,
      "learning_rate": 9.842098268030532e-06,
      "loss": 0.5291,
      "step": 1702
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1139228343963623,
      "learning_rate": 9.84184235935379e-06,
      "loss": 0.5783,
      "step": 1703
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0208879709243774,
      "learning_rate": 9.84158624680391e-06,
      "loss": 0.5841,
      "step": 1704
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0257160663604736,
      "learning_rate": 9.841329930391678e-06,
      "loss": 0.5669,
      "step": 1705
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3174355030059814,
      "learning_rate": 9.841073410127884e-06,
      "loss": 0.6071,
      "step": 1706
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1715240478515625,
      "learning_rate": 9.840816686023329e-06,
      "loss": 0.5801,
      "step": 1707
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9790030121803284,
      "learning_rate": 9.840559758088821e-06,
      "loss": 0.5325,
      "step": 1708
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0704594850540161,
      "learning_rate": 9.840302626335182e-06,
      "loss": 0.5574,
      "step": 1709
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.012350082397461,
      "learning_rate": 9.84004529077324e-06,
      "loss": 0.5779,
      "step": 1710
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9783762693405151,
      "learning_rate": 9.839787751413825e-06,
      "loss": 0.5429,
      "step": 1711
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0305521488189697,
      "learning_rate": 9.839530008267785e-06,
      "loss": 0.5534,
      "step": 1712
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0552220344543457,
      "learning_rate": 9.839272061345974e-06,
      "loss": 0.5872,
      "step": 1713
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0879662036895752,
      "learning_rate": 9.839013910659249e-06,
      "loss": 0.5924,
      "step": 1714
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1004040241241455,
      "learning_rate": 9.838755556218483e-06,
      "loss": 0.5854,
      "step": 1715
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0836282968521118,
      "learning_rate": 9.838496998034552e-06,
      "loss": 0.6243,
      "step": 1716
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1114145517349243,
      "learning_rate": 9.838238236118344e-06,
      "loss": 0.5672,
      "step": 1717
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0886434316635132,
      "learning_rate": 9.837979270480758e-06,
      "loss": 0.5725,
      "step": 1718
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9595142602920532,
      "learning_rate": 9.837720101132692e-06,
      "loss": 0.6296,
      "step": 1719
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0831856727600098,
      "learning_rate": 9.837460728085062e-06,
      "loss": 0.5772,
      "step": 1720
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0812630653381348,
      "learning_rate": 9.83720115134879e-06,
      "loss": 0.639,
      "step": 1721
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9859281778335571,
      "learning_rate": 9.836941370934806e-06,
      "loss": 0.556,
      "step": 1722
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.047037959098816,
      "learning_rate": 9.836681386854045e-06,
      "loss": 0.5702,
      "step": 1723
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0959748029708862,
      "learning_rate": 9.836421199117456e-06,
      "loss": 0.5845,
      "step": 1724
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0794752836227417,
      "learning_rate": 9.836160807735997e-06,
      "loss": 0.6148,
      "step": 1725
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0430015325546265,
      "learning_rate": 9.83590021272063e-06,
      "loss": 0.559,
      "step": 1726
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1524336338043213,
      "learning_rate": 9.835639414082327e-06,
      "loss": 0.583,
      "step": 1727
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0416326522827148,
      "learning_rate": 9.83537841183207e-06,
      "loss": 0.5753,
      "step": 1728
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.080689549446106,
      "learning_rate": 9.83511720598085e-06,
      "loss": 0.5485,
      "step": 1729
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.15471351146698,
      "learning_rate": 9.834855796539665e-06,
      "loss": 0.5558,
      "step": 1730
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0688234567642212,
      "learning_rate": 9.834594183519521e-06,
      "loss": 0.5585,
      "step": 1731
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0754145383834839,
      "learning_rate": 9.834332366931435e-06,
      "loss": 0.5745,
      "step": 1732
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0678609609603882,
      "learning_rate": 9.834070346786428e-06,
      "loss": 0.5691,
      "step": 1733
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.062833309173584,
      "learning_rate": 9.833808123095538e-06,
      "loss": 0.5819,
      "step": 1734
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1182148456573486,
      "learning_rate": 9.833545695869802e-06,
      "loss": 0.6185,
      "step": 1735
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1162580251693726,
      "learning_rate": 9.833283065120272e-06,
      "loss": 0.5718,
      "step": 1736
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0736720561981201,
      "learning_rate": 9.833020230858005e-06,
      "loss": 0.6132,
      "step": 1737
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0325653553009033,
      "learning_rate": 9.832757193094072e-06,
      "loss": 0.5879,
      "step": 1738
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9937186241149902,
      "learning_rate": 9.832493951839541e-06,
      "loss": 0.5699,
      "step": 1739
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.101816177368164,
      "learning_rate": 9.832230507105504e-06,
      "loss": 0.5643,
      "step": 1740
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0643540620803833,
      "learning_rate": 9.831966858903049e-06,
      "loss": 0.591,
      "step": 1741
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9806624054908752,
      "learning_rate": 9.83170300724328e-06,
      "loss": 0.5871,
      "step": 1742
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0082716941833496,
      "learning_rate": 9.831438952137304e-06,
      "loss": 0.5493,
      "step": 1743
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0105382204055786,
      "learning_rate": 9.831174693596241e-06,
      "loss": 0.5441,
      "step": 1744
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1338965892791748,
      "learning_rate": 9.83091023163122e-06,
      "loss": 0.6051,
      "step": 1745
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0223342180252075,
      "learning_rate": 9.830645566253374e-06,
      "loss": 0.5614,
      "step": 1746
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.089674711227417,
      "learning_rate": 9.830380697473848e-06,
      "loss": 0.5997,
      "step": 1747
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0419564247131348,
      "learning_rate": 9.830115625303793e-06,
      "loss": 0.6263,
      "step": 1748
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0378230810165405,
      "learning_rate": 9.829850349754373e-06,
      "loss": 0.5786,
      "step": 1749
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0939370393753052,
      "learning_rate": 9.829584870836756e-06,
      "loss": 0.6189,
      "step": 1750
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.079451084136963,
      "learning_rate": 9.82931918856212e-06,
      "loss": 0.5546,
      "step": 1751
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.10829496383667,
      "learning_rate": 9.829053302941656e-06,
      "loss": 0.5974,
      "step": 1752
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0517836809158325,
      "learning_rate": 9.828787213986554e-06,
      "loss": 0.6085,
      "step": 1753
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0262974500656128,
      "learning_rate": 9.82852092170802e-06,
      "loss": 0.5534,
      "step": 1754
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0490713119506836,
      "learning_rate": 9.82825442611727e-06,
      "loss": 0.5534,
      "step": 1755
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1421825885772705,
      "learning_rate": 9.82798772722552e-06,
      "loss": 0.6172,
      "step": 1756
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0918395519256592,
      "learning_rate": 9.827720825044003e-06,
      "loss": 0.5853,
      "step": 1757
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0919643640518188,
      "learning_rate": 9.827453719583957e-06,
      "loss": 0.5493,
      "step": 1758
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.004308819770813,
      "learning_rate": 9.827186410856627e-06,
      "loss": 0.5275,
      "step": 1759
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0599530935287476,
      "learning_rate": 9.82691889887327e-06,
      "loss": 0.5225,
      "step": 1760
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0359498262405396,
      "learning_rate": 9.82665118364515e-06,
      "loss": 0.5409,
      "step": 1761
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1455105543136597,
      "learning_rate": 9.82638326518354e-06,
      "loss": 0.5428,
      "step": 1762
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1107579469680786,
      "learning_rate": 9.826115143499721e-06,
      "loss": 0.6133,
      "step": 1763
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0757449865341187,
      "learning_rate": 9.82584681860498e-06,
      "loss": 0.5667,
      "step": 1764
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.064499855041504,
      "learning_rate": 9.82557829051062e-06,
      "loss": 0.5712,
      "step": 1765
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.087536334991455,
      "learning_rate": 9.825309559227944e-06,
      "loss": 0.6336,
      "step": 1766
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1417089700698853,
      "learning_rate": 9.825040624768267e-06,
      "loss": 0.5945,
      "step": 1767
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0964943170547485,
      "learning_rate": 9.824771487142917e-06,
      "loss": 0.6186,
      "step": 1768
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.009015440940857,
      "learning_rate": 9.824502146363222e-06,
      "loss": 0.5442,
      "step": 1769
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0028083324432373,
      "learning_rate": 9.824232602440524e-06,
      "loss": 0.5788,
      "step": 1770
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9938005805015564,
      "learning_rate": 9.823962855386175e-06,
      "loss": 0.597,
      "step": 1771
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0942137241363525,
      "learning_rate": 9.823692905211533e-06,
      "loss": 0.5339,
      "step": 1772
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0058943033218384,
      "learning_rate": 9.823422751927961e-06,
      "loss": 0.5573,
      "step": 1773
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0122864246368408,
      "learning_rate": 9.823152395546836e-06,
      "loss": 0.5873,
      "step": 1774
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0176011323928833,
      "learning_rate": 9.822881836079543e-06,
      "loss": 0.5539,
      "step": 1775
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0070388317108154,
      "learning_rate": 9.822611073537474e-06,
      "loss": 0.5853,
      "step": 1776
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0597985982894897,
      "learning_rate": 9.822340107932028e-06,
      "loss": 0.5133,
      "step": 1777
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1082496643066406,
      "learning_rate": 9.822068939274616e-06,
      "loss": 0.5884,
      "step": 1778
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0341330766677856,
      "learning_rate": 9.821797567576656e-06,
      "loss": 0.5426,
      "step": 1779
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0901429653167725,
      "learning_rate": 9.821525992849575e-06,
      "loss": 0.6166,
      "step": 1780
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1223825216293335,
      "learning_rate": 9.821254215104808e-06,
      "loss": 0.6138,
      "step": 1781
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0060911178588867,
      "learning_rate": 9.820982234353795e-06,
      "loss": 0.5476,
      "step": 1782
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1112552881240845,
      "learning_rate": 9.820710050607994e-06,
      "loss": 0.5866,
      "step": 1783
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0314642190933228,
      "learning_rate": 9.820437663878862e-06,
      "loss": 0.5702,
      "step": 1784
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0946764945983887,
      "learning_rate": 9.820165074177867e-06,
      "loss": 0.6082,
      "step": 1785
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0887361764907837,
      "learning_rate": 9.819892281516491e-06,
      "loss": 0.6036,
      "step": 1786
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.141646146774292,
      "learning_rate": 9.819619285906217e-06,
      "loss": 0.5512,
      "step": 1787
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1829429864883423,
      "learning_rate": 9.819346087358542e-06,
      "loss": 0.6157,
      "step": 1788
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1907756328582764,
      "learning_rate": 9.819072685884969e-06,
      "loss": 0.5512,
      "step": 1789
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1075764894485474,
      "learning_rate": 9.818799081497008e-06,
      "loss": 0.6017,
      "step": 1790
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0502817630767822,
      "learning_rate": 9.818525274206184e-06,
      "loss": 0.5918,
      "step": 1791
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.004138469696045,
      "learning_rate": 9.818251264024018e-06,
      "loss": 0.6126,
      "step": 1792
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0952894687652588,
      "learning_rate": 9.817977050962058e-06,
      "loss": 0.5419,
      "step": 1793
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0882782936096191,
      "learning_rate": 9.817702635031842e-06,
      "loss": 0.591,
      "step": 1794
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0567734241485596,
      "learning_rate": 9.817428016244928e-06,
      "loss": 0.5676,
      "step": 1795
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0806546211242676,
      "learning_rate": 9.81715319461288e-06,
      "loss": 0.6661,
      "step": 1796
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9864494204521179,
      "learning_rate": 9.816878170147268e-06,
      "loss": 0.6047,
      "step": 1797
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1121294498443604,
      "learning_rate": 9.816602942859672e-06,
      "loss": 0.6089,
      "step": 1798
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1498067378997803,
      "learning_rate": 9.816327512761683e-06,
      "loss": 0.5679,
      "step": 1799
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.135475993156433,
      "learning_rate": 9.816051879864896e-06,
      "loss": 0.5329,
      "step": 1800
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0519531965255737,
      "learning_rate": 9.81577604418092e-06,
      "loss": 0.5243,
      "step": 1801
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.032646894454956,
      "learning_rate": 9.815500005721365e-06,
      "loss": 0.575,
      "step": 1802
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1332684755325317,
      "learning_rate": 9.815223764497859e-06,
      "loss": 0.5767,
      "step": 1803
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9929540753364563,
      "learning_rate": 9.814947320522031e-06,
      "loss": 0.559,
      "step": 1804
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1531939506530762,
      "learning_rate": 9.81467067380552e-06,
      "loss": 0.5943,
      "step": 1805
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0956352949142456,
      "learning_rate": 9.814393824359975e-06,
      "loss": 0.5878,
      "step": 1806
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9945195913314819,
      "learning_rate": 9.814116772197058e-06,
      "loss": 0.5425,
      "step": 1807
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1511505842208862,
      "learning_rate": 9.813839517328428e-06,
      "loss": 0.5698,
      "step": 1808
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.056097388267517,
      "learning_rate": 9.813562059765762e-06,
      "loss": 0.5904,
      "step": 1809
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0813708305358887,
      "learning_rate": 9.813284399520744e-06,
      "loss": 0.5949,
      "step": 1810
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0767086744308472,
      "learning_rate": 9.813006536605063e-06,
      "loss": 0.584,
      "step": 1811
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1276978254318237,
      "learning_rate": 9.812728471030421e-06,
      "loss": 0.606,
      "step": 1812
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0942857265472412,
      "learning_rate": 9.812450202808525e-06,
      "loss": 0.5775,
      "step": 1813
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9871548414230347,
      "learning_rate": 9.812171731951092e-06,
      "loss": 0.5556,
      "step": 1814
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0339603424072266,
      "learning_rate": 9.811893058469848e-06,
      "loss": 0.5699,
      "step": 1815
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0050936937332153,
      "learning_rate": 9.811614182376527e-06,
      "loss": 0.6113,
      "step": 1816
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0542008876800537,
      "learning_rate": 9.811335103682872e-06,
      "loss": 0.5385,
      "step": 1817
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9927748441696167,
      "learning_rate": 9.81105582240063e-06,
      "loss": 0.597,
      "step": 1818
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0616050958633423,
      "learning_rate": 9.810776338541566e-06,
      "loss": 0.6027,
      "step": 1819
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0716493129730225,
      "learning_rate": 9.810496652117445e-06,
      "loss": 0.5842,
      "step": 1820
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9858418107032776,
      "learning_rate": 9.810216763140046e-06,
      "loss": 0.5235,
      "step": 1821
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0361227989196777,
      "learning_rate": 9.809936671621151e-06,
      "loss": 0.5259,
      "step": 1822
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0617575645446777,
      "learning_rate": 9.809656377572556e-06,
      "loss": 0.6021,
      "step": 1823
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0023410320281982,
      "learning_rate": 9.809375881006063e-06,
      "loss": 0.5129,
      "step": 1824
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9909228682518005,
      "learning_rate": 9.809095181933482e-06,
      "loss": 0.5426,
      "step": 1825
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1261311769485474,
      "learning_rate": 9.808814280366632e-06,
      "loss": 0.6176,
      "step": 1826
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1135318279266357,
      "learning_rate": 9.808533176317341e-06,
      "loss": 0.6037,
      "step": 1827
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1510497331619263,
      "learning_rate": 9.808251869797445e-06,
      "loss": 0.6103,
      "step": 1828
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0855205059051514,
      "learning_rate": 9.807970360818791e-06,
      "loss": 0.5569,
      "step": 1829
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1300290822982788,
      "learning_rate": 9.80768864939323e-06,
      "loss": 0.6362,
      "step": 1830
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.05135178565979,
      "learning_rate": 9.807406735532625e-06,
      "loss": 0.5439,
      "step": 1831
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.057826042175293,
      "learning_rate": 9.807124619248847e-06,
      "loss": 0.6045,
      "step": 1832
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2218470573425293,
      "learning_rate": 9.806842300553772e-06,
      "loss": 0.5813,
      "step": 1833
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0777051448822021,
      "learning_rate": 9.806559779459291e-06,
      "loss": 0.6229,
      "step": 1834
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9983317255973816,
      "learning_rate": 9.806277055977299e-06,
      "loss": 0.5392,
      "step": 1835
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1468912363052368,
      "learning_rate": 9.8059941301197e-06,
      "loss": 0.5798,
      "step": 1836
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0529228448867798,
      "learning_rate": 9.805711001898406e-06,
      "loss": 0.5773,
      "step": 1837
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.017897605895996,
      "learning_rate": 9.805427671325339e-06,
      "loss": 0.5667,
      "step": 1838
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.046511173248291,
      "learning_rate": 9.80514413841243e-06,
      "loss": 0.5286,
      "step": 1839
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2047935724258423,
      "learning_rate": 9.804860403171617e-06,
      "loss": 0.6145,
      "step": 1840
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1296573877334595,
      "learning_rate": 9.804576465614848e-06,
      "loss": 0.5614,
      "step": 1841
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.065781593322754,
      "learning_rate": 9.804292325754079e-06,
      "loss": 0.6024,
      "step": 1842
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0602785348892212,
      "learning_rate": 9.804007983601271e-06,
      "loss": 0.5905,
      "step": 1843
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0478906631469727,
      "learning_rate": 9.8037234391684e-06,
      "loss": 0.5914,
      "step": 1844
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0980864763259888,
      "learning_rate": 9.803438692467446e-06,
      "loss": 0.5525,
      "step": 1845
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.053962230682373,
      "learning_rate": 9.8031537435104e-06,
      "loss": 0.5454,
      "step": 1846
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0422577857971191,
      "learning_rate": 9.802868592309255e-06,
      "loss": 0.5807,
      "step": 1847
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1173555850982666,
      "learning_rate": 9.802583238876024e-06,
      "loss": 0.5915,
      "step": 1848
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0564782619476318,
      "learning_rate": 9.80229768322272e-06,
      "loss": 0.5566,
      "step": 1849
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0177700519561768,
      "learning_rate": 9.802011925361366e-06,
      "loss": 0.5884,
      "step": 1850
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0435445308685303,
      "learning_rate": 9.801725965303995e-06,
      "loss": 0.5258,
      "step": 1851
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1057952642440796,
      "learning_rate": 9.801439803062646e-06,
      "loss": 0.5988,
      "step": 1852
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9988928437232971,
      "learning_rate": 9.801153438649371e-06,
      "loss": 0.5789,
      "step": 1853
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.096474051475525,
      "learning_rate": 9.800866872076227e-06,
      "loss": 0.578,
      "step": 1854
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0434516668319702,
      "learning_rate": 9.80058010335528e-06,
      "loss": 0.5812,
      "step": 1855
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1052302122116089,
      "learning_rate": 9.800293132498603e-06,
      "loss": 0.5821,
      "step": 1856
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.084574580192566,
      "learning_rate": 9.800005959518284e-06,
      "loss": 0.5597,
      "step": 1857
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1198168992996216,
      "learning_rate": 9.79971858442641e-06,
      "loss": 0.6072,
      "step": 1858
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.142138957977295,
      "learning_rate": 9.799431007235086e-06,
      "loss": 0.5874,
      "step": 1859
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0381388664245605,
      "learning_rate": 9.799143227956416e-06,
      "loss": 0.6147,
      "step": 1860
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1210684776306152,
      "learning_rate": 9.798855246602522e-06,
      "loss": 0.6059,
      "step": 1861
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0363600254058838,
      "learning_rate": 9.798567063185525e-06,
      "loss": 0.549,
      "step": 1862
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0627700090408325,
      "learning_rate": 9.798278677717562e-06,
      "loss": 0.5713,
      "step": 1863
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0229395627975464,
      "learning_rate": 9.797990090210777e-06,
      "loss": 0.5821,
      "step": 1864
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0014104843139648,
      "learning_rate": 9.79770130067732e-06,
      "loss": 0.6146,
      "step": 1865
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9721370339393616,
      "learning_rate": 9.797412309129351e-06,
      "loss": 0.5623,
      "step": 1866
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0202564001083374,
      "learning_rate": 9.79712311557904e-06,
      "loss": 0.5536,
      "step": 1867
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.028006911277771,
      "learning_rate": 9.79683372003856e-06,
      "loss": 0.5573,
      "step": 1868
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0714778900146484,
      "learning_rate": 9.796544122520101e-06,
      "loss": 0.5889,
      "step": 1869
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.029428482055664,
      "learning_rate": 9.796254323035854e-06,
      "loss": 0.5298,
      "step": 1870
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0187203884124756,
      "learning_rate": 9.795964321598023e-06,
      "loss": 0.5737,
      "step": 1871
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9776306748390198,
      "learning_rate": 9.795674118218819e-06,
      "loss": 0.4964,
      "step": 1872
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0021920204162598,
      "learning_rate": 9.795383712910458e-06,
      "loss": 0.5602,
      "step": 1873
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.052433729171753,
      "learning_rate": 9.795093105685175e-06,
      "loss": 0.5904,
      "step": 1874
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0357136726379395,
      "learning_rate": 9.794802296555198e-06,
      "loss": 0.5716,
      "step": 1875
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9910592436790466,
      "learning_rate": 9.79451128553278e-06,
      "loss": 0.5199,
      "step": 1876
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1537259817123413,
      "learning_rate": 9.794220072630168e-06,
      "loss": 0.6142,
      "step": 1877
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2031731605529785,
      "learning_rate": 9.793928657859627e-06,
      "loss": 0.6193,
      "step": 1878
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.170706868171692,
      "learning_rate": 9.793637041233428e-06,
      "loss": 0.6113,
      "step": 1879
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0904954671859741,
      "learning_rate": 9.793345222763847e-06,
      "loss": 0.5832,
      "step": 1880
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0491715669631958,
      "learning_rate": 9.793053202463176e-06,
      "loss": 0.5481,
      "step": 1881
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0722602605819702,
      "learning_rate": 9.792760980343708e-06,
      "loss": 0.5864,
      "step": 1882
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1385523080825806,
      "learning_rate": 9.792468556417746e-06,
      "loss": 0.5359,
      "step": 1883
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0855581760406494,
      "learning_rate": 9.792175930697608e-06,
      "loss": 0.6192,
      "step": 1884
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.002265453338623,
      "learning_rate": 9.79188310319561e-06,
      "loss": 0.5793,
      "step": 1885
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0687613487243652,
      "learning_rate": 9.791590073924086e-06,
      "loss": 0.6658,
      "step": 1886
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0373985767364502,
      "learning_rate": 9.79129684289537e-06,
      "loss": 0.5612,
      "step": 1887
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9813857674598694,
      "learning_rate": 9.791003410121815e-06,
      "loss": 0.5701,
      "step": 1888
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0157337188720703,
      "learning_rate": 9.79070977561577e-06,
      "loss": 0.5682,
      "step": 1889
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9713268876075745,
      "learning_rate": 9.790415939389604e-06,
      "loss": 0.57,
      "step": 1890
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0119757652282715,
      "learning_rate": 9.790121901455687e-06,
      "loss": 0.5453,
      "step": 1891
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.069533348083496,
      "learning_rate": 9.7898276618264e-06,
      "loss": 0.544,
      "step": 1892
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9527726173400879,
      "learning_rate": 9.789533220514132e-06,
      "loss": 0.5392,
      "step": 1893
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0631836652755737,
      "learning_rate": 9.789238577531284e-06,
      "loss": 0.6161,
      "step": 1894
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0867925882339478,
      "learning_rate": 9.788943732890258e-06,
      "loss": 0.5382,
      "step": 1895
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0896707773208618,
      "learning_rate": 9.788648686603472e-06,
      "loss": 0.5852,
      "step": 1896
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0752573013305664,
      "learning_rate": 9.788353438683346e-06,
      "loss": 0.5952,
      "step": 1897
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0917840003967285,
      "learning_rate": 9.788057989142317e-06,
      "loss": 0.6199,
      "step": 1898
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9953743815422058,
      "learning_rate": 9.787762337992821e-06,
      "loss": 0.5669,
      "step": 1899
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1098178625106812,
      "learning_rate": 9.78746648524731e-06,
      "loss": 0.6031,
      "step": 1900
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0404541492462158,
      "learning_rate": 9.787170430918239e-06,
      "loss": 0.5569,
      "step": 1901
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1098657846450806,
      "learning_rate": 9.786874175018073e-06,
      "loss": 0.5853,
      "step": 1902
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0735530853271484,
      "learning_rate": 9.78657771755929e-06,
      "loss": 0.5813,
      "step": 1903
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9700368642807007,
      "learning_rate": 9.786281058554369e-06,
      "loss": 0.5414,
      "step": 1904
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.979625940322876,
      "learning_rate": 9.785984198015804e-06,
      "loss": 0.5367,
      "step": 1905
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0052965879440308,
      "learning_rate": 9.785687135956092e-06,
      "loss": 0.5731,
      "step": 1906
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0384953022003174,
      "learning_rate": 9.785389872387745e-06,
      "loss": 0.5167,
      "step": 1907
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0627541542053223,
      "learning_rate": 9.785092407323276e-06,
      "loss": 0.6133,
      "step": 1908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.117289423942566,
      "learning_rate": 9.784794740775212e-06,
      "loss": 0.5983,
      "step": 1909
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.080491542816162,
      "learning_rate": 9.784496872756086e-06,
      "loss": 0.5615,
      "step": 1910
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.043665885925293,
      "learning_rate": 9.784198803278442e-06,
      "loss": 0.5818,
      "step": 1911
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0099648237228394,
      "learning_rate": 9.78390053235483e-06,
      "loss": 0.5487,
      "step": 1912
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1236435174942017,
      "learning_rate": 9.783602059997808e-06,
      "loss": 0.5977,
      "step": 1913
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0785053968429565,
      "learning_rate": 9.783303386219942e-06,
      "loss": 0.5789,
      "step": 1914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0824027061462402,
      "learning_rate": 9.783004511033814e-06,
      "loss": 0.6267,
      "step": 1915
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1740599870681763,
      "learning_rate": 9.782705434452002e-06,
      "loss": 0.633,
      "step": 1916
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.023955225944519,
      "learning_rate": 9.782406156487104e-06,
      "loss": 0.598,
      "step": 1917
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1687668561935425,
      "learning_rate": 9.782106677151717e-06,
      "loss": 0.6052,
      "step": 1918
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0945203304290771,
      "learning_rate": 9.781806996458456e-06,
      "loss": 0.5986,
      "step": 1919
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1078828573226929,
      "learning_rate": 9.781507114419937e-06,
      "loss": 0.5578,
      "step": 1920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0594239234924316,
      "learning_rate": 9.781207031048785e-06,
      "loss": 0.5605,
      "step": 1921
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0740500688552856,
      "learning_rate": 9.78090674635764e-06,
      "loss": 0.5738,
      "step": 1922
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0668847560882568,
      "learning_rate": 9.780606260359141e-06,
      "loss": 0.5996,
      "step": 1923
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.119848608970642,
      "learning_rate": 9.780305573065945e-06,
      "loss": 0.6466,
      "step": 1924
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9590592384338379,
      "learning_rate": 9.78000468449071e-06,
      "loss": 0.5362,
      "step": 1925
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.09816575050354,
      "learning_rate": 9.779703594646106e-06,
      "loss": 0.6452,
      "step": 1926
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1173896789550781,
      "learning_rate": 9.779402303544811e-06,
      "loss": 0.5727,
      "step": 1927
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.010911464691162,
      "learning_rate": 9.77910081119951e-06,
      "loss": 0.5318,
      "step": 1928
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2026013135910034,
      "learning_rate": 9.7787991176229e-06,
      "loss": 0.6171,
      "step": 1929
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9925667643547058,
      "learning_rate": 9.778497222827685e-06,
      "loss": 0.547,
      "step": 1930
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0244348049163818,
      "learning_rate": 9.778195126826574e-06,
      "loss": 0.6138,
      "step": 1931
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0038400888442993,
      "learning_rate": 9.777892829632288e-06,
      "loss": 0.5836,
      "step": 1932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0449421405792236,
      "learning_rate": 9.777590331257557e-06,
      "loss": 0.6516,
      "step": 1933
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0840104818344116,
      "learning_rate": 9.777287631715117e-06,
      "loss": 0.5375,
      "step": 1934
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0799813270568848,
      "learning_rate": 9.776984731017714e-06,
      "loss": 0.6516,
      "step": 1935
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.014874815940857,
      "learning_rate": 9.7766816291781e-06,
      "loss": 0.536,
      "step": 1936
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1212245225906372,
      "learning_rate": 9.77637832620904e-06,
      "loss": 0.598,
      "step": 1937
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0710190534591675,
      "learning_rate": 9.776074822123306e-06,
      "loss": 0.5727,
      "step": 1938
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0462238788604736,
      "learning_rate": 9.775771116933674e-06,
      "loss": 0.5412,
      "step": 1939
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0557196140289307,
      "learning_rate": 9.775467210652936e-06,
      "loss": 0.5848,
      "step": 1940
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.044198751449585,
      "learning_rate": 9.775163103293885e-06,
      "loss": 0.5564,
      "step": 1941
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0481051206588745,
      "learning_rate": 9.774858794869328e-06,
      "loss": 0.6043,
      "step": 1942
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0210567712783813,
      "learning_rate": 9.774554285392078e-06,
      "loss": 0.5476,
      "step": 1943
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0004888772964478,
      "learning_rate": 9.774249574874957e-06,
      "loss": 0.5579,
      "step": 1944
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9968801736831665,
      "learning_rate": 9.773944663330793e-06,
      "loss": 0.5568,
      "step": 1945
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0612200498580933,
      "learning_rate": 9.773639550772428e-06,
      "loss": 0.5521,
      "step": 1946
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0417052507400513,
      "learning_rate": 9.773334237212707e-06,
      "loss": 0.5736,
      "step": 1947
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0855478048324585,
      "learning_rate": 9.773028722664486e-06,
      "loss": 0.5767,
      "step": 1948
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.009902834892273,
      "learning_rate": 9.77272300714063e-06,
      "loss": 0.6113,
      "step": 1949
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0125980377197266,
      "learning_rate": 9.77241709065401e-06,
      "loss": 0.5467,
      "step": 1950
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0341662168502808,
      "learning_rate": 9.772110973217512e-06,
      "loss": 0.578,
      "step": 1951
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0842918157577515,
      "learning_rate": 9.77180465484402e-06,
      "loss": 0.5719,
      "step": 1952
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9895298480987549,
      "learning_rate": 9.771498135546433e-06,
      "loss": 0.5673,
      "step": 1953
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0436815023422241,
      "learning_rate": 9.77119141533766e-06,
      "loss": 0.5575,
      "step": 1954
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.059710144996643,
      "learning_rate": 9.770884494230614e-06,
      "loss": 0.5649,
      "step": 1955
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9809030890464783,
      "learning_rate": 9.770577372238217e-06,
      "loss": 0.5313,
      "step": 1956
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0639938116073608,
      "learning_rate": 9.770270049373403e-06,
      "loss": 0.5909,
      "step": 1957
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0074024200439453,
      "learning_rate": 9.769962525649112e-06,
      "loss": 0.5131,
      "step": 1958
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0866878032684326,
      "learning_rate": 9.769654801078294e-06,
      "loss": 0.6138,
      "step": 1959
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0164234638214111,
      "learning_rate": 9.769346875673903e-06,
      "loss": 0.5365,
      "step": 1960
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0905537605285645,
      "learning_rate": 9.769038749448907e-06,
      "loss": 0.5678,
      "step": 1961
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0142149925231934,
      "learning_rate": 9.76873042241628e-06,
      "loss": 0.5378,
      "step": 1962
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0266367197036743,
      "learning_rate": 9.768421894589003e-06,
      "loss": 0.5878,
      "step": 1963
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1256475448608398,
      "learning_rate": 9.76811316598007e-06,
      "loss": 0.5872,
      "step": 1964
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.02944016456604,
      "learning_rate": 9.767804236602476e-06,
      "loss": 0.5647,
      "step": 1965
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.980645477771759,
      "learning_rate": 9.767495106469233e-06,
      "loss": 0.573,
      "step": 1966
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1279990673065186,
      "learning_rate": 9.767185775593356e-06,
      "loss": 0.5803,
      "step": 1967
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1235218048095703,
      "learning_rate": 9.76687624398787e-06,
      "loss": 0.5421,
      "step": 1968
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0594100952148438,
      "learning_rate": 9.766566511665808e-06,
      "loss": 0.5899,
      "step": 1969
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0320079326629639,
      "learning_rate": 9.766256578640212e-06,
      "loss": 0.5809,
      "step": 1970
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0690884590148926,
      "learning_rate": 9.76594644492413e-06,
      "loss": 0.5566,
      "step": 1971
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.002901554107666,
      "learning_rate": 9.765636110530626e-06,
      "loss": 0.5923,
      "step": 1972
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1202044486999512,
      "learning_rate": 9.765325575472761e-06,
      "loss": 0.571,
      "step": 1973
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0211033821105957,
      "learning_rate": 9.765014839763616e-06,
      "loss": 0.589,
      "step": 1974
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.101736307144165,
      "learning_rate": 9.764703903416271e-06,
      "loss": 0.545,
      "step": 1975
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0086767673492432,
      "learning_rate": 9.76439276644382e-06,
      "loss": 0.5584,
      "step": 1976
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0515953302383423,
      "learning_rate": 9.764081428859363e-06,
      "loss": 0.6101,
      "step": 1977
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0618976354599,
      "learning_rate": 9.763769890676011e-06,
      "loss": 0.5512,
      "step": 1978
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1033425331115723,
      "learning_rate": 9.76345815190688e-06,
      "loss": 0.6218,
      "step": 1979
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.13877272605896,
      "learning_rate": 9.763146212565097e-06,
      "loss": 0.5939,
      "step": 1980
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0750939846038818,
      "learning_rate": 9.762834072663798e-06,
      "loss": 0.5518,
      "step": 1981
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9559831023216248,
      "learning_rate": 9.762521732216124e-06,
      "loss": 0.5434,
      "step": 1982
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0375754833221436,
      "learning_rate": 9.762209191235227e-06,
      "loss": 0.6282,
      "step": 1983
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.04959237575531,
      "learning_rate": 9.761896449734269e-06,
      "loss": 0.5665,
      "step": 1984
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.168744683265686,
      "learning_rate": 9.761583507726416e-06,
      "loss": 0.5913,
      "step": 1985
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2400338649749756,
      "learning_rate": 9.761270365224846e-06,
      "loss": 0.5966,
      "step": 1986
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9545580744743347,
      "learning_rate": 9.760957022242746e-06,
      "loss": 0.5043,
      "step": 1987
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.086821436882019,
      "learning_rate": 9.760643478793305e-06,
      "loss": 0.6122,
      "step": 1988
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0779457092285156,
      "learning_rate": 9.760329734889729e-06,
      "loss": 0.5736,
      "step": 1989
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0653327703475952,
      "learning_rate": 9.760015790545227e-06,
      "loss": 0.552,
      "step": 1990
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.119644284248352,
      "learning_rate": 9.759701645773022e-06,
      "loss": 0.5557,
      "step": 1991
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.195134162902832,
      "learning_rate": 9.759387300586336e-06,
      "loss": 0.5542,
      "step": 1992
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8945276141166687,
      "learning_rate": 9.759072754998407e-06,
      "loss": 0.5006,
      "step": 1993
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1034178733825684,
      "learning_rate": 9.758758009022482e-06,
      "loss": 0.612,
      "step": 1994
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9998924732208252,
      "learning_rate": 9.758443062671809e-06,
      "loss": 0.5873,
      "step": 1995
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0209602117538452,
      "learning_rate": 9.758127915959655e-06,
      "loss": 0.5756,
      "step": 1996
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1128679513931274,
      "learning_rate": 9.757812568899285e-06,
      "loss": 0.6099,
      "step": 1997
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9929051399230957,
      "learning_rate": 9.75749702150398e-06,
      "loss": 0.5368,
      "step": 1998
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0805941820144653,
      "learning_rate": 9.757181273787024e-06,
      "loss": 0.5681,
      "step": 1999
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0125678777694702,
      "learning_rate": 9.756865325761715e-06,
      "loss": 0.5911,
      "step": 2000
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0813782215118408,
      "learning_rate": 9.756549177441354e-06,
      "loss": 0.5496,
      "step": 2001
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0482062101364136,
      "learning_rate": 9.756232828839256e-06,
      "loss": 0.5561,
      "step": 2002
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0184236764907837,
      "learning_rate": 9.755916279968738e-06,
      "loss": 0.583,
      "step": 2003
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0551586151123047,
      "learning_rate": 9.75559953084313e-06,
      "loss": 0.5462,
      "step": 2004
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.135002851486206,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.5428,
      "step": 2005
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0241683721542358,
      "learning_rate": 9.75496543188e-06,
      "loss": 0.5505,
      "step": 2006
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.01645827293396,
      "learning_rate": 9.754648082069181e-06,
      "loss": 0.5413,
      "step": 2007
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0453730821609497,
      "learning_rate": 9.75433053205667e-06,
      "loss": 0.5514,
      "step": 2008
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0232735872268677,
      "learning_rate": 9.754012781855837e-06,
      "loss": 0.5932,
      "step": 2009
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0743303298950195,
      "learning_rate": 9.753694831480067e-06,
      "loss": 0.5519,
      "step": 2010
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1090643405914307,
      "learning_rate": 9.753376680942744e-06,
      "loss": 0.5891,
      "step": 2011
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0725889205932617,
      "learning_rate": 9.753058330257263e-06,
      "loss": 0.5544,
      "step": 2012
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1309471130371094,
      "learning_rate": 9.752739779437032e-06,
      "loss": 0.573,
      "step": 2013
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1250039339065552,
      "learning_rate": 9.752421028495461e-06,
      "loss": 0.6168,
      "step": 2014
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0606460571289062,
      "learning_rate": 9.752102077445974e-06,
      "loss": 0.5707,
      "step": 2015
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0642883777618408,
      "learning_rate": 9.751782926302e-06,
      "loss": 0.5621,
      "step": 2016
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0262463092803955,
      "learning_rate": 9.751463575076977e-06,
      "loss": 0.5777,
      "step": 2017
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0222927331924438,
      "learning_rate": 9.75114402378435e-06,
      "loss": 0.5698,
      "step": 2018
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1447890996932983,
      "learning_rate": 9.75082427243758e-06,
      "loss": 0.6233,
      "step": 2019
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1772607564926147,
      "learning_rate": 9.750504321050126e-06,
      "loss": 0.6242,
      "step": 2020
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0555652379989624,
      "learning_rate": 9.75018416963546e-06,
      "loss": 0.5834,
      "step": 2021
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1782764196395874,
      "learning_rate": 9.749863818207061e-06,
      "loss": 0.5901,
      "step": 2022
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0955591201782227,
      "learning_rate": 9.749543266778424e-06,
      "loss": 0.5761,
      "step": 2023
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.125753402709961,
      "learning_rate": 9.749222515363041e-06,
      "loss": 0.5906,
      "step": 2024
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0516302585601807,
      "learning_rate": 9.748901563974418e-06,
      "loss": 0.5582,
      "step": 2025
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1180061101913452,
      "learning_rate": 9.748580412626072e-06,
      "loss": 0.6122,
      "step": 2026
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0065104961395264,
      "learning_rate": 9.748259061331524e-06,
      "loss": 0.5759,
      "step": 2027
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0845080614089966,
      "learning_rate": 9.747937510104305e-06,
      "loss": 0.5855,
      "step": 2028
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0901507139205933,
      "learning_rate": 9.74761575895795e-06,
      "loss": 0.5857,
      "step": 2029
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1333222389221191,
      "learning_rate": 9.747293807906017e-06,
      "loss": 0.5956,
      "step": 2030
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9672915935516357,
      "learning_rate": 9.746971656962053e-06,
      "loss": 0.5722,
      "step": 2031
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.082801342010498,
      "learning_rate": 9.746649306139627e-06,
      "loss": 0.5835,
      "step": 2032
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1733567714691162,
      "learning_rate": 9.74632675545231e-06,
      "loss": 0.5665,
      "step": 2033
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.02162504196167,
      "learning_rate": 9.746004004913688e-06,
      "loss": 0.5609,
      "step": 2034
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9524051547050476,
      "learning_rate": 9.745681054537345e-06,
      "loss": 0.5042,
      "step": 2035
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0968283414840698,
      "learning_rate": 9.745357904336882e-06,
      "loss": 0.5785,
      "step": 2036
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9993317723274231,
      "learning_rate": 9.745034554325905e-06,
      "loss": 0.5537,
      "step": 2037
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0851517915725708,
      "learning_rate": 9.74471100451803e-06,
      "loss": 0.5703,
      "step": 2038
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0083051919937134,
      "learning_rate": 9.744387254926882e-06,
      "loss": 0.559,
      "step": 2039
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1442487239837646,
      "learning_rate": 9.74406330556609e-06,
      "loss": 0.6185,
      "step": 2040
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1060651540756226,
      "learning_rate": 9.743739156449294e-06,
      "loss": 0.6161,
      "step": 2041
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0019123554229736,
      "learning_rate": 9.743414807590145e-06,
      "loss": 0.5745,
      "step": 2042
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0129235982894897,
      "learning_rate": 9.743090259002302e-06,
      "loss": 0.549,
      "step": 2043
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.088689923286438,
      "learning_rate": 9.742765510699425e-06,
      "loss": 0.5491,
      "step": 2044
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0228464603424072,
      "learning_rate": 9.742440562695194e-06,
      "loss": 0.5514,
      "step": 2045
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0378639698028564,
      "learning_rate": 9.742115415003288e-06,
      "loss": 0.553,
      "step": 2046
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0331372022628784,
      "learning_rate": 9.741790067637398e-06,
      "loss": 0.6551,
      "step": 2047
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0934799909591675,
      "learning_rate": 9.741464520611223e-06,
      "loss": 0.5955,
      "step": 2048
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.070549488067627,
      "learning_rate": 9.741138773938472e-06,
      "loss": 0.5467,
      "step": 2049
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.067488670349121,
      "learning_rate": 9.74081282763286e-06,
      "loss": 0.5599,
      "step": 2050
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0350698232650757,
      "learning_rate": 9.740486681708114e-06,
      "loss": 0.6225,
      "step": 2051
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0792306661605835,
      "learning_rate": 9.740160336177962e-06,
      "loss": 0.6127,
      "step": 2052
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0747098922729492,
      "learning_rate": 9.73983379105615e-06,
      "loss": 0.5754,
      "step": 2053
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.069152593612671,
      "learning_rate": 9.739507046356424e-06,
      "loss": 0.5852,
      "step": 2054
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9734413623809814,
      "learning_rate": 9.739180102092544e-06,
      "loss": 0.5735,
      "step": 2055
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1087932586669922,
      "learning_rate": 9.738852958278278e-06,
      "loss": 0.5878,
      "step": 2056
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0658721923828125,
      "learning_rate": 9.738525614927399e-06,
      "loss": 0.5843,
      "step": 2057
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9961588382720947,
      "learning_rate": 9.73819807205369e-06,
      "loss": 0.5266,
      "step": 2058
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0801833868026733,
      "learning_rate": 9.737870329670942e-06,
      "loss": 0.5932,
      "step": 2059
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1166014671325684,
      "learning_rate": 9.737542387792957e-06,
      "loss": 0.6081,
      "step": 2060
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0558866262435913,
      "learning_rate": 9.737214246433544e-06,
      "loss": 0.5787,
      "step": 2061
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1056846380233765,
      "learning_rate": 9.736885905606516e-06,
      "loss": 0.5789,
      "step": 2062
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0813485383987427,
      "learning_rate": 9.736557365325703e-06,
      "loss": 0.5877,
      "step": 2063
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.005872368812561,
      "learning_rate": 9.736228625604938e-06,
      "loss": 0.5553,
      "step": 2064
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0798795223236084,
      "learning_rate": 9.735899686458059e-06,
      "loss": 0.5727,
      "step": 2065
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9533892273902893,
      "learning_rate": 9.73557054789892e-06,
      "loss": 0.5545,
      "step": 2066
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9775428771972656,
      "learning_rate": 9.73524120994138e-06,
      "loss": 0.5799,
      "step": 2067
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0594699382781982,
      "learning_rate": 9.734911672599304e-06,
      "loss": 0.5669,
      "step": 2068
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.021916151046753,
      "learning_rate": 9.73458193588657e-06,
      "loss": 0.6152,
      "step": 2069
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0601035356521606,
      "learning_rate": 9.734251999817061e-06,
      "loss": 0.5698,
      "step": 2070
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.082911491394043,
      "learning_rate": 9.733921864404669e-06,
      "loss": 0.5278,
      "step": 2071
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1061042547225952,
      "learning_rate": 9.733591529663295e-06,
      "loss": 0.5808,
      "step": 2072
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0210843086242676,
      "learning_rate": 9.73326099560685e-06,
      "loss": 0.5361,
      "step": 2073
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9831228852272034,
      "learning_rate": 9.732930262249249e-06,
      "loss": 0.5207,
      "step": 2074
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.057084083557129,
      "learning_rate": 9.73259932960442e-06,
      "loss": 0.5674,
      "step": 2075
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0501519441604614,
      "learning_rate": 9.732268197686296e-06,
      "loss": 0.5809,
      "step": 2076
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0547102689743042,
      "learning_rate": 9.731936866508822e-06,
      "loss": 0.5508,
      "step": 2077
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1186622381210327,
      "learning_rate": 9.731605336085947e-06,
      "loss": 0.6001,
      "step": 2078
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0614229440689087,
      "learning_rate": 9.73127360643163e-06,
      "loss": 0.5409,
      "step": 2079
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.057842493057251,
      "learning_rate": 9.730941677559843e-06,
      "loss": 0.5829,
      "step": 2080
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0404309034347534,
      "learning_rate": 9.730609549484558e-06,
      "loss": 0.6032,
      "step": 2081
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9369000792503357,
      "learning_rate": 9.730277222219762e-06,
      "loss": 0.4888,
      "step": 2082
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1000438928604126,
      "learning_rate": 9.729944695779448e-06,
      "loss": 0.5853,
      "step": 2083
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0321012735366821,
      "learning_rate": 9.729611970177615e-06,
      "loss": 0.5321,
      "step": 2084
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0978355407714844,
      "learning_rate": 9.729279045428277e-06,
      "loss": 0.6109,
      "step": 2085
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0012578964233398,
      "learning_rate": 9.72894592154545e-06,
      "loss": 0.5483,
      "step": 2086
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0384838581085205,
      "learning_rate": 9.72861259854316e-06,
      "loss": 0.5335,
      "step": 2087
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0041470527648926,
      "learning_rate": 9.728279076435446e-06,
      "loss": 0.5387,
      "step": 2088
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.126891851425171,
      "learning_rate": 9.727945355236345e-06,
      "loss": 0.5387,
      "step": 2089
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1620920896530151,
      "learning_rate": 9.727611434959914e-06,
      "loss": 0.5745,
      "step": 2090
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0596435070037842,
      "learning_rate": 9.727277315620212e-06,
      "loss": 0.6089,
      "step": 2091
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9877673983573914,
      "learning_rate": 9.726942997231308e-06,
      "loss": 0.5555,
      "step": 2092
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0126142501831055,
      "learning_rate": 9.726608479807278e-06,
      "loss": 0.5542,
      "step": 2093
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0659465789794922,
      "learning_rate": 9.726273763362206e-06,
      "loss": 0.6002,
      "step": 2094
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.177112340927124,
      "learning_rate": 9.725938847910187e-06,
      "loss": 0.5862,
      "step": 2095
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0119010210037231,
      "learning_rate": 9.725603733465325e-06,
      "loss": 0.557,
      "step": 2096
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0778146982192993,
      "learning_rate": 9.725268420041728e-06,
      "loss": 0.6207,
      "step": 2097
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9919752478599548,
      "learning_rate": 9.724932907653516e-06,
      "loss": 0.5706,
      "step": 2098
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.003469705581665,
      "learning_rate": 9.724597196314817e-06,
      "loss": 0.5379,
      "step": 2099
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0249968767166138,
      "learning_rate": 9.724261286039766e-06,
      "loss": 0.55,
      "step": 2100
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1211967468261719,
      "learning_rate": 9.723925176842506e-06,
      "loss": 0.5692,
      "step": 2101
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1254615783691406,
      "learning_rate": 9.72358886873719e-06,
      "loss": 0.5689,
      "step": 2102
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9981034398078918,
      "learning_rate": 9.723252361737977e-06,
      "loss": 0.5425,
      "step": 2103
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9972788095474243,
      "learning_rate": 9.722915655859042e-06,
      "loss": 0.6044,
      "step": 2104
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0443592071533203,
      "learning_rate": 9.722578751114556e-06,
      "loss": 0.5497,
      "step": 2105
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0376752614974976,
      "learning_rate": 9.72224164751871e-06,
      "loss": 0.6096,
      "step": 2106
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2066547870635986,
      "learning_rate": 9.721904345085692e-06,
      "loss": 0.6374,
      "step": 2107
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0088324546813965,
      "learning_rate": 9.72156684382971e-06,
      "loss": 0.5344,
      "step": 2108
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0367845296859741,
      "learning_rate": 9.721229143764975e-06,
      "loss": 0.5277,
      "step": 2109
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0177240371704102,
      "learning_rate": 9.720891244905701e-06,
      "loss": 0.5958,
      "step": 2110
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.14991295337677,
      "learning_rate": 9.720553147266123e-06,
      "loss": 0.6143,
      "step": 2111
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1234478950500488,
      "learning_rate": 9.720214850860473e-06,
      "loss": 0.6052,
      "step": 2112
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1023095846176147,
      "learning_rate": 9.719876355702993e-06,
      "loss": 0.5674,
      "step": 2113
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0658434629440308,
      "learning_rate": 9.719537661807942e-06,
      "loss": 0.5703,
      "step": 2114
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0622062683105469,
      "learning_rate": 9.719198769189577e-06,
      "loss": 0.5308,
      "step": 2115
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0702844858169556,
      "learning_rate": 9.718859677862169e-06,
      "loss": 0.5833,
      "step": 2116
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0853898525238037,
      "learning_rate": 9.718520387839996e-06,
      "loss": 0.5786,
      "step": 2117
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0653886795043945,
      "learning_rate": 9.718180899137344e-06,
      "loss": 0.5863,
      "step": 2118
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0895500183105469,
      "learning_rate": 9.717841211768505e-06,
      "loss": 0.5423,
      "step": 2119
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0399247407913208,
      "learning_rate": 9.71750132574779e-06,
      "loss": 0.5686,
      "step": 2120
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1106464862823486,
      "learning_rate": 9.717161241089501e-06,
      "loss": 0.5797,
      "step": 2121
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0631537437438965,
      "learning_rate": 9.716820957807963e-06,
      "loss": 0.5606,
      "step": 2122
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.048399806022644,
      "learning_rate": 9.716480475917504e-06,
      "loss": 0.566,
      "step": 2123
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0127686262130737,
      "learning_rate": 9.716139795432459e-06,
      "loss": 0.5521,
      "step": 2124
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0522278547286987,
      "learning_rate": 9.715798916367174e-06,
      "loss": 0.571,
      "step": 2125
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0320959091186523,
      "learning_rate": 9.715457838736e-06,
      "loss": 0.5768,
      "step": 2126
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0515543222427368,
      "learning_rate": 9.715116562553302e-06,
      "loss": 0.5936,
      "step": 2127
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0121512413024902,
      "learning_rate": 9.714775087833446e-06,
      "loss": 0.5614,
      "step": 2128
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0705124139785767,
      "learning_rate": 9.714433414590816e-06,
      "loss": 0.565,
      "step": 2129
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0589114427566528,
      "learning_rate": 9.714091542839792e-06,
      "loss": 0.5875,
      "step": 2130
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.98140549659729,
      "learning_rate": 9.713749472594773e-06,
      "loss": 0.569,
      "step": 2131
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0898550748825073,
      "learning_rate": 9.713407203870163e-06,
      "loss": 0.5749,
      "step": 2132
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9971396923065186,
      "learning_rate": 9.713064736680372e-06,
      "loss": 0.5178,
      "step": 2133
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0567504167556763,
      "learning_rate": 9.71272207103982e-06,
      "loss": 0.536,
      "step": 2134
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1251214742660522,
      "learning_rate": 9.712379206962936e-06,
      "loss": 0.6351,
      "step": 2135
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9792860150337219,
      "learning_rate": 9.712036144464157e-06,
      "loss": 0.5423,
      "step": 2136
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0028667449951172,
      "learning_rate": 9.711692883557928e-06,
      "loss": 0.6023,
      "step": 2137
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.046597957611084,
      "learning_rate": 9.711349424258702e-06,
      "loss": 0.5918,
      "step": 2138
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0695263147354126,
      "learning_rate": 9.711005766580942e-06,
      "loss": 0.5404,
      "step": 2139
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0932008028030396,
      "learning_rate": 9.710661910539117e-06,
      "loss": 0.5448,
      "step": 2140
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0518196821212769,
      "learning_rate": 9.710317856147707e-06,
      "loss": 0.5634,
      "step": 2141
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0272433757781982,
      "learning_rate": 9.709973603421196e-06,
      "loss": 0.5293,
      "step": 2142
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0365588665008545,
      "learning_rate": 9.709629152374084e-06,
      "loss": 0.5599,
      "step": 2143
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0788600444793701,
      "learning_rate": 9.70928450302087e-06,
      "loss": 0.6063,
      "step": 2144
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.985220193862915,
      "learning_rate": 9.708939655376069e-06,
      "loss": 0.522,
      "step": 2145
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0706517696380615,
      "learning_rate": 9.708594609454201e-06,
      "loss": 0.5709,
      "step": 2146
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9971744418144226,
      "learning_rate": 9.708249365269793e-06,
      "loss": 0.5983,
      "step": 2147
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0823099613189697,
      "learning_rate": 9.707903922837382e-06,
      "loss": 0.599,
      "step": 2148
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.09287691116333,
      "learning_rate": 9.707558282171517e-06,
      "loss": 0.6628,
      "step": 2149
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9843802452087402,
      "learning_rate": 9.707212443286746e-06,
      "loss": 0.5796,
      "step": 2150
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0742542743682861,
      "learning_rate": 9.706866406197637e-06,
      "loss": 0.5566,
      "step": 2151
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0588560104370117,
      "learning_rate": 9.706520170918756e-06,
      "loss": 0.6191,
      "step": 2152
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0245299339294434,
      "learning_rate": 9.706173737464683e-06,
      "loss": 0.5365,
      "step": 2153
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.029465913772583,
      "learning_rate": 9.705827105850008e-06,
      "loss": 0.5687,
      "step": 2154
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1049184799194336,
      "learning_rate": 9.705480276089323e-06,
      "loss": 0.5191,
      "step": 2155
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0191420316696167,
      "learning_rate": 9.705133248197232e-06,
      "loss": 0.5487,
      "step": 2156
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0053186416625977,
      "learning_rate": 9.704786022188346e-06,
      "loss": 0.5622,
      "step": 2157
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0235744714736938,
      "learning_rate": 9.704438598077291e-06,
      "loss": 0.5317,
      "step": 2158
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0682446956634521,
      "learning_rate": 9.70409097587869e-06,
      "loss": 0.5349,
      "step": 2159
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.011461615562439,
      "learning_rate": 9.703743155607182e-06,
      "loss": 0.5326,
      "step": 2160
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0195093154907227,
      "learning_rate": 9.703395137277414e-06,
      "loss": 0.5612,
      "step": 2161
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0611745119094849,
      "learning_rate": 9.703046920904038e-06,
      "loss": 0.5622,
      "step": 2162
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2104378938674927,
      "learning_rate": 9.702698506501717e-06,
      "loss": 0.5635,
      "step": 2163
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.056689739227295,
      "learning_rate": 9.702349894085122e-06,
      "loss": 0.559,
      "step": 2164
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0749422311782837,
      "learning_rate": 9.702001083668931e-06,
      "loss": 0.5979,
      "step": 2165
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1210516691207886,
      "learning_rate": 9.701652075267832e-06,
      "loss": 0.6176,
      "step": 2166
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1960784196853638,
      "learning_rate": 9.701302868896518e-06,
      "loss": 0.5733,
      "step": 2167
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0504810810089111,
      "learning_rate": 9.700953464569698e-06,
      "loss": 0.5869,
      "step": 2168
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0147689580917358,
      "learning_rate": 9.700603862302078e-06,
      "loss": 0.5408,
      "step": 2169
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0584076642990112,
      "learning_rate": 9.700254062108383e-06,
      "loss": 0.5806,
      "step": 2170
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0558146238327026,
      "learning_rate": 9.69990406400334e-06,
      "loss": 0.5633,
      "step": 2171
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0875517129898071,
      "learning_rate": 9.699553868001688e-06,
      "loss": 0.6023,
      "step": 2172
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1330591440200806,
      "learning_rate": 9.699203474118168e-06,
      "loss": 0.5357,
      "step": 2173
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0651309490203857,
      "learning_rate": 9.69885288236754e-06,
      "loss": 0.5559,
      "step": 2174
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2233500480651855,
      "learning_rate": 9.698502092764562e-06,
      "loss": 0.5752,
      "step": 2175
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0270609855651855,
      "learning_rate": 9.698151105324006e-06,
      "loss": 0.5416,
      "step": 2176
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0004528760910034,
      "learning_rate": 9.697799920060651e-06,
      "loss": 0.5924,
      "step": 2177
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0074985027313232,
      "learning_rate": 9.697448536989284e-06,
      "loss": 0.5565,
      "step": 2178
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.044296145439148,
      "learning_rate": 9.697096956124699e-06,
      "loss": 0.5802,
      "step": 2179
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1109081506729126,
      "learning_rate": 9.696745177481703e-06,
      "loss": 0.5349,
      "step": 2180
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9400250911712646,
      "learning_rate": 9.696393201075105e-06,
      "loss": 0.5466,
      "step": 2181
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1142268180847168,
      "learning_rate": 9.696041026919727e-06,
      "loss": 0.535,
      "step": 2182
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0524247884750366,
      "learning_rate": 9.695688655030397e-06,
      "loss": 0.616,
      "step": 2183
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0073829889297485,
      "learning_rate": 9.695336085421953e-06,
      "loss": 0.5719,
      "step": 2184
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0474032163619995,
      "learning_rate": 9.694983318109242e-06,
      "loss": 0.5499,
      "step": 2185
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0479087829589844,
      "learning_rate": 9.694630353107115e-06,
      "loss": 0.616,
      "step": 2186
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.073878288269043,
      "learning_rate": 9.694277190430437e-06,
      "loss": 0.5741,
      "step": 2187
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0107192993164062,
      "learning_rate": 9.693923830094074e-06,
      "loss": 0.5786,
      "step": 2188
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0605086088180542,
      "learning_rate": 9.693570272112908e-06,
      "loss": 0.5929,
      "step": 2189
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0389643907546997,
      "learning_rate": 9.693216516501827e-06,
      "loss": 0.5538,
      "step": 2190
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0293034315109253,
      "learning_rate": 9.692862563275725e-06,
      "loss": 0.6193,
      "step": 2191
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.003514051437378,
      "learning_rate": 9.692508412449505e-06,
      "loss": 0.5897,
      "step": 2192
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0086450576782227,
      "learning_rate": 9.692154064038079e-06,
      "loss": 0.5623,
      "step": 2193
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0342178344726562,
      "learning_rate": 9.691799518056369e-06,
      "loss": 0.6051,
      "step": 2194
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0912590026855469,
      "learning_rate": 9.691444774519302e-06,
      "loss": 0.5518,
      "step": 2195
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.103579044342041,
      "learning_rate": 9.691089833441818e-06,
      "loss": 0.6017,
      "step": 2196
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0409106016159058,
      "learning_rate": 9.69073469483886e-06,
      "loss": 0.5614,
      "step": 2197
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.062774896621704,
      "learning_rate": 9.690379358725379e-06,
      "loss": 0.6216,
      "step": 2198
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.027420997619629,
      "learning_rate": 9.69002382511634e-06,
      "loss": 0.5999,
      "step": 2199
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0598483085632324,
      "learning_rate": 9.689668094026716e-06,
      "loss": 0.5388,
      "step": 2200
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1370553970336914,
      "learning_rate": 9.689312165471483e-06,
      "loss": 0.5359,
      "step": 2201
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0518770217895508,
      "learning_rate": 9.688956039465626e-06,
      "loss": 0.567,
      "step": 2202
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.046511173248291,
      "learning_rate": 9.688599716024141e-06,
      "loss": 0.569,
      "step": 2203
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0349102020263672,
      "learning_rate": 9.688243195162033e-06,
      "loss": 0.5654,
      "step": 2204
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0427666902542114,
      "learning_rate": 9.687886476894314e-06,
      "loss": 0.5174,
      "step": 2205
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.213161587715149,
      "learning_rate": 9.687529561236004e-06,
      "loss": 0.6052,
      "step": 2206
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.065405249595642,
      "learning_rate": 9.687172448202129e-06,
      "loss": 0.5917,
      "step": 2207
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9554116725921631,
      "learning_rate": 9.68681513780773e-06,
      "loss": 0.5801,
      "step": 2208
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1155883073806763,
      "learning_rate": 9.686457630067848e-06,
      "loss": 0.5936,
      "step": 2209
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0245048999786377,
      "learning_rate": 9.686099924997538e-06,
      "loss": 0.5474,
      "step": 2210
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9610018730163574,
      "learning_rate": 9.685742022611864e-06,
      "loss": 0.5173,
      "step": 2211
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.137954592704773,
      "learning_rate": 9.685383922925892e-06,
      "loss": 0.5747,
      "step": 2212
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0669362545013428,
      "learning_rate": 9.685025625954703e-06,
      "loss": 0.5971,
      "step": 2213
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.135469675064087,
      "learning_rate": 9.684667131713381e-06,
      "loss": 0.5644,
      "step": 2214
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.065121054649353,
      "learning_rate": 9.684308440217026e-06,
      "loss": 0.6075,
      "step": 2215
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0996694564819336,
      "learning_rate": 9.683949551480736e-06,
      "loss": 0.515,
      "step": 2216
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.113033413887024,
      "learning_rate": 9.683590465519625e-06,
      "loss": 0.6083,
      "step": 2217
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0447050333023071,
      "learning_rate": 9.683231182348813e-06,
      "loss": 0.5751,
      "step": 2218
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9936871528625488,
      "learning_rate": 9.682871701983428e-06,
      "loss": 0.5505,
      "step": 2219
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1078193187713623,
      "learning_rate": 9.682512024438607e-06,
      "loss": 0.5774,
      "step": 2220
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0609996318817139,
      "learning_rate": 9.682152149729491e-06,
      "loss": 0.6056,
      "step": 2221
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.070388674736023,
      "learning_rate": 9.681792077871238e-06,
      "loss": 0.5829,
      "step": 2222
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1255348920822144,
      "learning_rate": 9.681431808879007e-06,
      "loss": 0.5679,
      "step": 2223
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1114442348480225,
      "learning_rate": 9.681071342767967e-06,
      "loss": 0.592,
      "step": 2224
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1285667419433594,
      "learning_rate": 9.6807106795533e-06,
      "loss": 0.5819,
      "step": 2225
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.109213948249817,
      "learning_rate": 9.680349819250185e-06,
      "loss": 0.5491,
      "step": 2226
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0387630462646484,
      "learning_rate": 9.679988761873824e-06,
      "loss": 0.5465,
      "step": 2227
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0273832082748413,
      "learning_rate": 9.679627507439416e-06,
      "loss": 0.581,
      "step": 2228
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0205281972885132,
      "learning_rate": 9.679266055962174e-06,
      "loss": 0.61,
      "step": 2229
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9858378171920776,
      "learning_rate": 9.678904407457314e-06,
      "loss": 0.555,
      "step": 2230
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1034581661224365,
      "learning_rate": 9.678542561940067e-06,
      "loss": 0.6115,
      "step": 2231
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1124613285064697,
      "learning_rate": 9.678180519425669e-06,
      "loss": 0.5678,
      "step": 2232
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0323601961135864,
      "learning_rate": 9.677818279929363e-06,
      "loss": 0.5149,
      "step": 2233
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.946808934211731,
      "learning_rate": 9.677455843466402e-06,
      "loss": 0.5688,
      "step": 2234
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0611376762390137,
      "learning_rate": 9.677093210052048e-06,
      "loss": 0.5322,
      "step": 2235
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0465165376663208,
      "learning_rate": 9.676730379701567e-06,
      "loss": 0.5893,
      "step": 2236
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0118218660354614,
      "learning_rate": 9.676367352430242e-06,
      "loss": 0.6004,
      "step": 2237
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0092300176620483,
      "learning_rate": 9.676004128253354e-06,
      "loss": 0.571,
      "step": 2238
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0186326503753662,
      "learning_rate": 9.675640707186199e-06,
      "loss": 0.5769,
      "step": 2239
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0467184782028198,
      "learning_rate": 9.67527708924408e-06,
      "loss": 0.5376,
      "step": 2240
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.021779179573059,
      "learning_rate": 9.674913274442305e-06,
      "loss": 0.6079,
      "step": 2241
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9716930985450745,
      "learning_rate": 9.674549262796196e-06,
      "loss": 0.5796,
      "step": 2242
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9714882969856262,
      "learning_rate": 9.674185054321079e-06,
      "loss": 0.5152,
      "step": 2243
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0956681966781616,
      "learning_rate": 9.67382064903229e-06,
      "loss": 0.5704,
      "step": 2244
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1594781875610352,
      "learning_rate": 9.67345604694517e-06,
      "loss": 0.5553,
      "step": 2245
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.062707543373108,
      "learning_rate": 9.673091248075077e-06,
      "loss": 0.5664,
      "step": 2246
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0908327102661133,
      "learning_rate": 9.672726252437368e-06,
      "loss": 0.566,
      "step": 2247
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0008841753005981,
      "learning_rate": 9.67236106004741e-06,
      "loss": 0.5607,
      "step": 2248
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9752715229988098,
      "learning_rate": 9.671995670920582e-06,
      "loss": 0.5441,
      "step": 2249
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0136269330978394,
      "learning_rate": 9.671630085072268e-06,
      "loss": 0.5602,
      "step": 2250
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0056018829345703,
      "learning_rate": 9.671264302517864e-06,
      "loss": 0.5857,
      "step": 2251
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9630345106124878,
      "learning_rate": 9.67089832327277e-06,
      "loss": 0.5591,
      "step": 2252
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1637920141220093,
      "learning_rate": 9.670532147352399e-06,
      "loss": 0.6148,
      "step": 2253
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0402402877807617,
      "learning_rate": 9.670165774772164e-06,
      "loss": 0.5934,
      "step": 2254
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0115119218826294,
      "learning_rate": 9.669799205547494e-06,
      "loss": 0.5944,
      "step": 2255
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.013441562652588,
      "learning_rate": 9.669432439693827e-06,
      "loss": 0.6018,
      "step": 2256
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0764124393463135,
      "learning_rate": 9.669065477226602e-06,
      "loss": 0.612,
      "step": 2257
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0073105096817017,
      "learning_rate": 9.668698318161271e-06,
      "loss": 0.5506,
      "step": 2258
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0922359228134155,
      "learning_rate": 9.668330962513297e-06,
      "loss": 0.5587,
      "step": 2259
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2077372074127197,
      "learning_rate": 9.667963410298147e-06,
      "loss": 0.5915,
      "step": 2260
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0957609415054321,
      "learning_rate": 9.667595661531294e-06,
      "loss": 0.5817,
      "step": 2261
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9836629033088684,
      "learning_rate": 9.667227716228228e-06,
      "loss": 0.5945,
      "step": 2262
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9690822958946228,
      "learning_rate": 9.666859574404434e-06,
      "loss": 0.5601,
      "step": 2263
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0340832471847534,
      "learning_rate": 9.666491236075423e-06,
      "loss": 0.5599,
      "step": 2264
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0833946466445923,
      "learning_rate": 9.666122701256697e-06,
      "loss": 0.6062,
      "step": 2265
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0714364051818848,
      "learning_rate": 9.665753969963779e-06,
      "loss": 0.5708,
      "step": 2266
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.003904104232788,
      "learning_rate": 9.66538504221219e-06,
      "loss": 0.6108,
      "step": 2267
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.98911052942276,
      "learning_rate": 9.665015918017467e-06,
      "loss": 0.5416,
      "step": 2268
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.054664969444275,
      "learning_rate": 9.664646597395151e-06,
      "loss": 0.5697,
      "step": 2269
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0036427974700928,
      "learning_rate": 9.664277080360796e-06,
      "loss": 0.6049,
      "step": 2270
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0508859157562256,
      "learning_rate": 9.663907366929958e-06,
      "loss": 0.5559,
      "step": 2271
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.055727243423462,
      "learning_rate": 9.663537457118206e-06,
      "loss": 0.6226,
      "step": 2272
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1699100732803345,
      "learning_rate": 9.663167350941114e-06,
      "loss": 0.6372,
      "step": 2273
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0462695360183716,
      "learning_rate": 9.662797048414267e-06,
      "loss": 0.6348,
      "step": 2274
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1070884466171265,
      "learning_rate": 9.662426549553257e-06,
      "loss": 0.5869,
      "step": 2275
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0312139987945557,
      "learning_rate": 9.662055854373684e-06,
      "loss": 0.6051,
      "step": 2276
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9913875460624695,
      "learning_rate": 9.661684962891158e-06,
      "loss": 0.5672,
      "step": 2277
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9986532926559448,
      "learning_rate": 9.661313875121294e-06,
      "loss": 0.5188,
      "step": 2278
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0187432765960693,
      "learning_rate": 9.66094259107972e-06,
      "loss": 0.6049,
      "step": 2279
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1970367431640625,
      "learning_rate": 9.660571110782066e-06,
      "loss": 0.5716,
      "step": 2280
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0717835426330566,
      "learning_rate": 9.660199434243977e-06,
      "loss": 0.5828,
      "step": 2281
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.001078486442566,
      "learning_rate": 9.6598275614811e-06,
      "loss": 0.5683,
      "step": 2282
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0684194564819336,
      "learning_rate": 9.659455492509096e-06,
      "loss": 0.5513,
      "step": 2283
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.065119981765747,
      "learning_rate": 9.659083227343628e-06,
      "loss": 0.5637,
      "step": 2284
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1782958507537842,
      "learning_rate": 9.658710766000375e-06,
      "loss": 0.5661,
      "step": 2285
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0934861898422241,
      "learning_rate": 9.658338108495018e-06,
      "loss": 0.5459,
      "step": 2286
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0259177684783936,
      "learning_rate": 9.65796525484325e-06,
      "loss": 0.5708,
      "step": 2287
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0861341953277588,
      "learning_rate": 9.657592205060766e-06,
      "loss": 0.5637,
      "step": 2288
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0681545734405518,
      "learning_rate": 9.657218959163278e-06,
      "loss": 0.5259,
      "step": 2289
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.044212818145752,
      "learning_rate": 9.656845517166502e-06,
      "loss": 0.5527,
      "step": 2290
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1174644231796265,
      "learning_rate": 9.656471879086158e-06,
      "loss": 0.57,
      "step": 2291
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0047990083694458,
      "learning_rate": 9.656098044937985e-06,
      "loss": 0.5822,
      "step": 2292
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.07584547996521,
      "learning_rate": 9.65572401473772e-06,
      "loss": 0.6041,
      "step": 2293
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.054582118988037,
      "learning_rate": 9.655349788501112e-06,
      "loss": 0.5843,
      "step": 2294
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.116593599319458,
      "learning_rate": 9.654975366243919e-06,
      "loss": 0.5613,
      "step": 2295
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9938434362411499,
      "learning_rate": 9.654600747981908e-06,
      "loss": 0.5017,
      "step": 2296
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0228008031845093,
      "learning_rate": 9.654225933730852e-06,
      "loss": 0.5806,
      "step": 2297
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0263339281082153,
      "learning_rate": 9.653850923506532e-06,
      "loss": 0.5742,
      "step": 2298
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0496147871017456,
      "learning_rate": 9.653475717324739e-06,
      "loss": 0.5722,
      "step": 2299
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0120099782943726,
      "learning_rate": 9.65310031520127e-06,
      "loss": 0.6022,
      "step": 2300
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0658786296844482,
      "learning_rate": 9.652724717151938e-06,
      "loss": 0.6067,
      "step": 2301
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.041401743888855,
      "learning_rate": 9.652348923192551e-06,
      "loss": 0.5763,
      "step": 2302
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9860716462135315,
      "learning_rate": 9.651972933338935e-06,
      "loss": 0.5306,
      "step": 2303
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0807832479476929,
      "learning_rate": 9.651596747606924e-06,
      "loss": 0.6203,
      "step": 2304
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0927594900131226,
      "learning_rate": 9.651220366012354e-06,
      "loss": 0.566,
      "step": 2305
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0677021741867065,
      "learning_rate": 9.650843788571076e-06,
      "loss": 0.582,
      "step": 2306
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1915913820266724,
      "learning_rate": 9.650467015298943e-06,
      "loss": 0.5704,
      "step": 2307
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0427172183990479,
      "learning_rate": 9.650090046211822e-06,
      "loss": 0.6025,
      "step": 2308
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9778934121131897,
      "learning_rate": 9.649712881325587e-06,
      "loss": 0.518,
      "step": 2309
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0721766948699951,
      "learning_rate": 9.649335520656118e-06,
      "loss": 0.6194,
      "step": 2310
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.088017463684082,
      "learning_rate": 9.648957964219303e-06,
      "loss": 0.6091,
      "step": 2311
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0366744995117188,
      "learning_rate": 9.64858021203104e-06,
      "loss": 0.589,
      "step": 2312
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9899066686630249,
      "learning_rate": 9.648202264107239e-06,
      "loss": 0.5605,
      "step": 2313
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0178853273391724,
      "learning_rate": 9.647824120463806e-06,
      "loss": 0.5583,
      "step": 2314
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.095003604888916,
      "learning_rate": 9.64744578111667e-06,
      "loss": 0.5185,
      "step": 2315
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1024019718170166,
      "learning_rate": 9.647067246081761e-06,
      "loss": 0.5716,
      "step": 2316
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0408539772033691,
      "learning_rate": 9.646688515375014e-06,
      "loss": 0.5683,
      "step": 2317
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0439146757125854,
      "learning_rate": 9.646309589012379e-06,
      "loss": 0.547,
      "step": 2318
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9975500106811523,
      "learning_rate": 9.64593046700981e-06,
      "loss": 0.5251,
      "step": 2319
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0552396774291992,
      "learning_rate": 9.645551149383272e-06,
      "loss": 0.5575,
      "step": 2320
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.975261926651001,
      "learning_rate": 9.645171636148736e-06,
      "loss": 0.5052,
      "step": 2321
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0283842086791992,
      "learning_rate": 9.644791927322182e-06,
      "loss": 0.5466,
      "step": 2322
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.044140338897705,
      "learning_rate": 9.644412022919597e-06,
      "loss": 0.5374,
      "step": 2323
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0010985136032104,
      "learning_rate": 9.644031922956979e-06,
      "loss": 0.5567,
      "step": 2324
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0314868688583374,
      "learning_rate": 9.64365162745033e-06,
      "loss": 0.5878,
      "step": 2325
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9741953015327454,
      "learning_rate": 9.643271136415668e-06,
      "loss": 0.5589,
      "step": 2326
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0657076835632324,
      "learning_rate": 9.642890449869008e-06,
      "loss": 0.5935,
      "step": 2327
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9860471487045288,
      "learning_rate": 9.642509567826386e-06,
      "loss": 0.5435,
      "step": 2328
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.027422308921814,
      "learning_rate": 9.642128490303834e-06,
      "loss": 0.5383,
      "step": 2329
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9340013265609741,
      "learning_rate": 9.6417472173174e-06,
      "loss": 0.5688,
      "step": 2330
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9668096899986267,
      "learning_rate": 9.64136574888314e-06,
      "loss": 0.5719,
      "step": 2331
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0264288187026978,
      "learning_rate": 9.640984085017113e-06,
      "loss": 0.5957,
      "step": 2332
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9968881011009216,
      "learning_rate": 9.640602225735391e-06,
      "loss": 0.6009,
      "step": 2333
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9973297119140625,
      "learning_rate": 9.640220171054054e-06,
      "loss": 0.5623,
      "step": 2334
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2151033878326416,
      "learning_rate": 9.639837920989188e-06,
      "loss": 0.5627,
      "step": 2335
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1902861595153809,
      "learning_rate": 9.639455475556887e-06,
      "loss": 0.5959,
      "step": 2336
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0719233751296997,
      "learning_rate": 9.639072834773254e-06,
      "loss": 0.5984,
      "step": 2337
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.039047360420227,
      "learning_rate": 9.638689998654404e-06,
      "loss": 0.5333,
      "step": 2338
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0445502996444702,
      "learning_rate": 9.638306967216453e-06,
      "loss": 0.584,
      "step": 2339
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9414918422698975,
      "learning_rate": 9.637923740475534e-06,
      "loss": 0.5178,
      "step": 2340
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0898582935333252,
      "learning_rate": 9.637540318447778e-06,
      "loss": 0.5821,
      "step": 2341
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0142707824707031,
      "learning_rate": 9.637156701149333e-06,
      "loss": 0.5971,
      "step": 2342
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0771269798278809,
      "learning_rate": 9.636772888596352e-06,
      "loss": 0.5993,
      "step": 2343
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9535529613494873,
      "learning_rate": 9.636388880804991e-06,
      "loss": 0.5602,
      "step": 2344
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0608489513397217,
      "learning_rate": 9.636004677791427e-06,
      "loss": 0.5983,
      "step": 2345
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.132565975189209,
      "learning_rate": 9.635620279571833e-06,
      "loss": 0.6043,
      "step": 2346
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0791984796524048,
      "learning_rate": 9.635235686162395e-06,
      "loss": 0.5979,
      "step": 2347
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9959856867790222,
      "learning_rate": 9.634850897579304e-06,
      "loss": 0.5548,
      "step": 2348
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0113996267318726,
      "learning_rate": 9.63446591383877e-06,
      "loss": 0.5516,
      "step": 2349
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0375350713729858,
      "learning_rate": 9.634080734956993e-06,
      "loss": 0.5548,
      "step": 2350
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0703330039978027,
      "learning_rate": 9.633695360950202e-06,
      "loss": 0.5933,
      "step": 2351
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0566142797470093,
      "learning_rate": 9.633309791834617e-06,
      "loss": 0.5393,
      "step": 2352
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2384819984436035,
      "learning_rate": 9.632924027626474e-06,
      "loss": 0.5905,
      "step": 2353
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.102569341659546,
      "learning_rate": 9.632538068342018e-06,
      "loss": 0.5665,
      "step": 2354
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0893784761428833,
      "learning_rate": 9.632151913997498e-06,
      "loss": 0.5453,
      "step": 2355
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1036884784698486,
      "learning_rate": 9.631765564609177e-06,
      "loss": 0.5205,
      "step": 2356
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1188225746154785,
      "learning_rate": 9.63137902019332e-06,
      "loss": 0.5471,
      "step": 2357
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0827534198760986,
      "learning_rate": 9.630992280766202e-06,
      "loss": 0.536,
      "step": 2358
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1823593378067017,
      "learning_rate": 9.630605346344113e-06,
      "loss": 0.5388,
      "step": 2359
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.026418924331665,
      "learning_rate": 9.630218216943338e-06,
      "loss": 0.6167,
      "step": 2360
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1206985712051392,
      "learning_rate": 9.629830892580183e-06,
      "loss": 0.6046,
      "step": 2361
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1533458232879639,
      "learning_rate": 9.629443373270954e-06,
      "loss": 0.5696,
      "step": 2362
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.19341242313385,
      "learning_rate": 9.62905565903197e-06,
      "loss": 0.6008,
      "step": 2363
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0197237730026245,
      "learning_rate": 9.628667749879555e-06,
      "loss": 0.5629,
      "step": 2364
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0176202058792114,
      "learning_rate": 9.628279645830044e-06,
      "loss": 0.5628,
      "step": 2365
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0141323804855347,
      "learning_rate": 9.627891346899775e-06,
      "loss": 0.5293,
      "step": 2366
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0090693235397339,
      "learning_rate": 9.627502853105104e-06,
      "loss": 0.5858,
      "step": 2367
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.059515357017517,
      "learning_rate": 9.627114164462385e-06,
      "loss": 0.5623,
      "step": 2368
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0840575695037842,
      "learning_rate": 9.626725280987985e-06,
      "loss": 0.6196,
      "step": 2369
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.001650333404541,
      "learning_rate": 9.626336202698277e-06,
      "loss": 0.5397,
      "step": 2370
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0970371961593628,
      "learning_rate": 9.625946929609647e-06,
      "loss": 0.537,
      "step": 2371
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.096764087677002,
      "learning_rate": 9.625557461738484e-06,
      "loss": 0.6295,
      "step": 2372
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0875904560089111,
      "learning_rate": 9.625167799101188e-06,
      "loss": 0.6304,
      "step": 2373
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1247825622558594,
      "learning_rate": 9.624777941714165e-06,
      "loss": 0.6608,
      "step": 2374
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0841033458709717,
      "learning_rate": 9.624387889593832e-06,
      "loss": 0.5454,
      "step": 2375
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9857283234596252,
      "learning_rate": 9.62399764275661e-06,
      "loss": 0.5596,
      "step": 2376
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0509493350982666,
      "learning_rate": 9.623607201218934e-06,
      "loss": 0.587,
      "step": 2377
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0771265029907227,
      "learning_rate": 9.623216564997244e-06,
      "loss": 0.6089,
      "step": 2378
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9977293014526367,
      "learning_rate": 9.622825734107987e-06,
      "loss": 0.5601,
      "step": 2379
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1041595935821533,
      "learning_rate": 9.62243470856762e-06,
      "loss": 0.6054,
      "step": 2380
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0072989463806152,
      "learning_rate": 9.622043488392607e-06,
      "loss": 0.5584,
      "step": 2381
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9942018985748291,
      "learning_rate": 9.621652073599423e-06,
      "loss": 0.5868,
      "step": 2382
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9523398280143738,
      "learning_rate": 9.621260464204548e-06,
      "loss": 0.5531,
      "step": 2383
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.998120129108429,
      "learning_rate": 9.620868660224468e-06,
      "loss": 0.5677,
      "step": 2384
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0065315961837769,
      "learning_rate": 9.620476661675685e-06,
      "loss": 0.5598,
      "step": 2385
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.123533844947815,
      "learning_rate": 9.620084468574704e-06,
      "loss": 0.5741,
      "step": 2386
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.026013731956482,
      "learning_rate": 9.619692080938039e-06,
      "loss": 0.532,
      "step": 2387
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.102170467376709,
      "learning_rate": 9.61929949878221e-06,
      "loss": 0.608,
      "step": 2388
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0668294429779053,
      "learning_rate": 9.618906722123748e-06,
      "loss": 0.5807,
      "step": 2389
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1137778759002686,
      "learning_rate": 9.618513750979193e-06,
      "loss": 0.5795,
      "step": 2390
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0429565906524658,
      "learning_rate": 9.61812058536509e-06,
      "loss": 0.5191,
      "step": 2391
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0552812814712524,
      "learning_rate": 9.617727225297994e-06,
      "loss": 0.5823,
      "step": 2392
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0765811204910278,
      "learning_rate": 9.617333670794468e-06,
      "loss": 0.5655,
      "step": 2393
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.030846357345581,
      "learning_rate": 9.616939921871087e-06,
      "loss": 0.5564,
      "step": 2394
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1203022003173828,
      "learning_rate": 9.616545978544424e-06,
      "loss": 0.5641,
      "step": 2395
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0004371404647827,
      "learning_rate": 9.616151840831069e-06,
      "loss": 0.5221,
      "step": 2396
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0611863136291504,
      "learning_rate": 9.61575750874762e-06,
      "loss": 0.5545,
      "step": 2397
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9985150098800659,
      "learning_rate": 9.615362982310679e-06,
      "loss": 0.4831,
      "step": 2398
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.041325330734253,
      "learning_rate": 9.614968261536858e-06,
      "loss": 0.5563,
      "step": 2399
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.217852234840393,
      "learning_rate": 9.61457334644278e-06,
      "loss": 0.617,
      "step": 2400
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1040374040603638,
      "learning_rate": 9.61417823704507e-06,
      "loss": 0.5679,
      "step": 2401
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.114501953125,
      "learning_rate": 9.613782933360365e-06,
      "loss": 0.608,
      "step": 2402
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0527658462524414,
      "learning_rate": 9.613387435405312e-06,
      "loss": 0.4935,
      "step": 2403
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1038916110992432,
      "learning_rate": 9.612991743196562e-06,
      "loss": 0.5531,
      "step": 2404
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9707751870155334,
      "learning_rate": 9.612595856750776e-06,
      "loss": 0.5677,
      "step": 2405
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.034013271331787,
      "learning_rate": 9.612199776084627e-06,
      "loss": 0.5734,
      "step": 2406
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1263413429260254,
      "learning_rate": 9.611803501214789e-06,
      "loss": 0.6142,
      "step": 2407
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9992231726646423,
      "learning_rate": 9.61140703215795e-06,
      "loss": 0.5411,
      "step": 2408
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0508081912994385,
      "learning_rate": 9.611010368930801e-06,
      "loss": 0.5605,
      "step": 2409
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0325145721435547,
      "learning_rate": 9.610613511550047e-06,
      "loss": 0.5872,
      "step": 2410
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0452853441238403,
      "learning_rate": 9.610216460032398e-06,
      "loss": 0.5944,
      "step": 2411
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0934975147247314,
      "learning_rate": 9.60981921439457e-06,
      "loss": 0.6453,
      "step": 2412
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0208979845046997,
      "learning_rate": 9.609421774653291e-06,
      "loss": 0.622,
      "step": 2413
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0178954601287842,
      "learning_rate": 9.609024140825299e-06,
      "loss": 0.5192,
      "step": 2414
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1866627931594849,
      "learning_rate": 9.608626312927331e-06,
      "loss": 0.6222,
      "step": 2415
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0078473091125488,
      "learning_rate": 9.608228290976143e-06,
      "loss": 0.5801,
      "step": 2416
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.073379397392273,
      "learning_rate": 9.607830074988491e-06,
      "loss": 0.6375,
      "step": 2417
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0799641609191895,
      "learning_rate": 9.607431664981144e-06,
      "loss": 0.5561,
      "step": 2418
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1281019449234009,
      "learning_rate": 9.607033060970878e-06,
      "loss": 0.5997,
      "step": 2419
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0619001388549805,
      "learning_rate": 9.606634262974477e-06,
      "loss": 0.5907,
      "step": 2420
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9821775555610657,
      "learning_rate": 9.606235271008732e-06,
      "loss": 0.5351,
      "step": 2421
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1251587867736816,
      "learning_rate": 9.605836085090445e-06,
      "loss": 0.5694,
      "step": 2422
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.228594183921814,
      "learning_rate": 9.605436705236421e-06,
      "loss": 0.5574,
      "step": 2423
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9809057712554932,
      "learning_rate": 9.60503713146348e-06,
      "loss": 0.577,
      "step": 2424
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1365041732788086,
      "learning_rate": 9.604637363788444e-06,
      "loss": 0.574,
      "step": 2425
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9982814788818359,
      "learning_rate": 9.604237402228149e-06,
      "loss": 0.622,
      "step": 2426
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.141382098197937,
      "learning_rate": 9.603837246799431e-06,
      "loss": 0.5937,
      "step": 2427
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.042654275894165,
      "learning_rate": 9.603436897519145e-06,
      "loss": 0.5735,
      "step": 2428
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0142533779144287,
      "learning_rate": 9.603036354404145e-06,
      "loss": 0.5442,
      "step": 2429
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9460806846618652,
      "learning_rate": 9.602635617471295e-06,
      "loss": 0.5689,
      "step": 2430
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9694622159004211,
      "learning_rate": 9.602234686737473e-06,
      "loss": 0.6085,
      "step": 2431
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9917210340499878,
      "learning_rate": 9.601833562219556e-06,
      "loss": 0.562,
      "step": 2432
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9378957152366638,
      "learning_rate": 9.601432243934437e-06,
      "loss": 0.5098,
      "step": 2433
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0544147491455078,
      "learning_rate": 9.601030731899014e-06,
      "loss": 0.5531,
      "step": 2434
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9746567010879517,
      "learning_rate": 9.600629026130192e-06,
      "loss": 0.5252,
      "step": 2435
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0485756397247314,
      "learning_rate": 9.600227126644887e-06,
      "loss": 0.6327,
      "step": 2436
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0036715269088745,
      "learning_rate": 9.59982503346002e-06,
      "loss": 0.5437,
      "step": 2437
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0754265785217285,
      "learning_rate": 9.599422746592522e-06,
      "loss": 0.6003,
      "step": 2438
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.062813639640808,
      "learning_rate": 9.599020266059334e-06,
      "loss": 0.6354,
      "step": 2439
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9763906002044678,
      "learning_rate": 9.5986175918774e-06,
      "loss": 0.5905,
      "step": 2440
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0599114894866943,
      "learning_rate": 9.598214724063678e-06,
      "loss": 0.6055,
      "step": 2441
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.984308123588562,
      "learning_rate": 9.597811662635128e-06,
      "loss": 0.5885,
      "step": 2442
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2546172142028809,
      "learning_rate": 9.597408407608725e-06,
      "loss": 0.6079,
      "step": 2443
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.020815134048462,
      "learning_rate": 9.597004959001447e-06,
      "loss": 0.5654,
      "step": 2444
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1067922115325928,
      "learning_rate": 9.596601316830282e-06,
      "loss": 0.5599,
      "step": 2445
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0436384677886963,
      "learning_rate": 9.596197481112225e-06,
      "loss": 0.5519,
      "step": 2446
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9777306914329529,
      "learning_rate": 9.59579345186428e-06,
      "loss": 0.5408,
      "step": 2447
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1045746803283691,
      "learning_rate": 9.595389229103464e-06,
      "loss": 0.5752,
      "step": 2448
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.973228931427002,
      "learning_rate": 9.594984812846792e-06,
      "loss": 0.5467,
      "step": 2449
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0757728815078735,
      "learning_rate": 9.594580203111294e-06,
      "loss": 0.5394,
      "step": 2450
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0427677631378174,
      "learning_rate": 9.594175399914008e-06,
      "loss": 0.5478,
      "step": 2451
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0544366836547852,
      "learning_rate": 9.593770403271977e-06,
      "loss": 0.5963,
      "step": 2452
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0926449298858643,
      "learning_rate": 9.593365213202255e-06,
      "loss": 0.5747,
      "step": 2453
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9562935829162598,
      "learning_rate": 9.592959829721903e-06,
      "loss": 0.5514,
      "step": 2454
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.050257682800293,
      "learning_rate": 9.59255425284799e-06,
      "loss": 0.5929,
      "step": 2455
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0073059797286987,
      "learning_rate": 9.592148482597595e-06,
      "loss": 0.5564,
      "step": 2456
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0695605278015137,
      "learning_rate": 9.591742518987802e-06,
      "loss": 0.6154,
      "step": 2457
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.957906186580658,
      "learning_rate": 9.591336362035703e-06,
      "loss": 0.5076,
      "step": 2458
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9516271352767944,
      "learning_rate": 9.590930011758403e-06,
      "loss": 0.5469,
      "step": 2459
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0699506998062134,
      "learning_rate": 9.590523468173011e-06,
      "loss": 0.5738,
      "step": 2460
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1135766506195068,
      "learning_rate": 9.590116731296646e-06,
      "loss": 0.5231,
      "step": 2461
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0049141645431519,
      "learning_rate": 9.589709801146432e-06,
      "loss": 0.5416,
      "step": 2462
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.034530758857727,
      "learning_rate": 9.589302677739506e-06,
      "loss": 0.5573,
      "step": 2463
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0589345693588257,
      "learning_rate": 9.588895361093009e-06,
      "loss": 0.5911,
      "step": 2464
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.033722162246704,
      "learning_rate": 9.588487851224091e-06,
      "loss": 0.5711,
      "step": 2465
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.143248200416565,
      "learning_rate": 9.588080148149912e-06,
      "loss": 0.5856,
      "step": 2466
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1562327146530151,
      "learning_rate": 9.587672251887639e-06,
      "loss": 0.6477,
      "step": 2467
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9303533434867859,
      "learning_rate": 9.587264162454447e-06,
      "loss": 0.5134,
      "step": 2468
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.045608639717102,
      "learning_rate": 9.586855879867519e-06,
      "loss": 0.5697,
      "step": 2469
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0757079124450684,
      "learning_rate": 9.586447404144046e-06,
      "loss": 0.6234,
      "step": 2470
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.096331000328064,
      "learning_rate": 9.58603873530123e-06,
      "loss": 0.5649,
      "step": 2471
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0169209241867065,
      "learning_rate": 9.585629873356273e-06,
      "loss": 0.4537,
      "step": 2472
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0567876100540161,
      "learning_rate": 9.585220818326395e-06,
      "loss": 0.5838,
      "step": 2473
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.066560983657837,
      "learning_rate": 9.58481157022882e-06,
      "loss": 0.5419,
      "step": 2474
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0556069612503052,
      "learning_rate": 9.584402129080779e-06,
      "loss": 0.575,
      "step": 2475
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9920236468315125,
      "learning_rate": 9.583992494899513e-06,
      "loss": 0.55,
      "step": 2476
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0362834930419922,
      "learning_rate": 9.583582667702269e-06,
      "loss": 0.5787,
      "step": 2477
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0171829462051392,
      "learning_rate": 9.583172647506305e-06,
      "loss": 0.546,
      "step": 2478
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0156619548797607,
      "learning_rate": 9.582762434328883e-06,
      "loss": 0.5497,
      "step": 2479
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.990680456161499,
      "learning_rate": 9.582352028187278e-06,
      "loss": 0.6005,
      "step": 2480
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9822958111763,
      "learning_rate": 9.581941429098769e-06,
      "loss": 0.5028,
      "step": 2481
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0820964574813843,
      "learning_rate": 9.581530637080647e-06,
      "loss": 0.5871,
      "step": 2482
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1711864471435547,
      "learning_rate": 9.581119652150208e-06,
      "loss": 0.5747,
      "step": 2483
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1426204442977905,
      "learning_rate": 9.580708474324755e-06,
      "loss": 0.5489,
      "step": 2484
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0703874826431274,
      "learning_rate": 9.580297103621605e-06,
      "loss": 0.5749,
      "step": 2485
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0250228643417358,
      "learning_rate": 9.579885540058079e-06,
      "loss": 0.5777,
      "step": 2486
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.002596139907837,
      "learning_rate": 9.579473783651503e-06,
      "loss": 0.5351,
      "step": 2487
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1740162372589111,
      "learning_rate": 9.579061834419217e-06,
      "loss": 0.6129,
      "step": 2488
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1330641508102417,
      "learning_rate": 9.578649692378567e-06,
      "loss": 0.6367,
      "step": 2489
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1116493940353394,
      "learning_rate": 9.578237357546907e-06,
      "loss": 0.5447,
      "step": 2490
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0301001071929932,
      "learning_rate": 9.577824829941597e-06,
      "loss": 0.5909,
      "step": 2491
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0472018718719482,
      "learning_rate": 9.577412109580009e-06,
      "loss": 0.5484,
      "step": 2492
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0376631021499634,
      "learning_rate": 9.57699919647952e-06,
      "loss": 0.6091,
      "step": 2493
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1155240535736084,
      "learning_rate": 9.576586090657519e-06,
      "loss": 0.5844,
      "step": 2494
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.144724726676941,
      "learning_rate": 9.576172792131397e-06,
      "loss": 0.5911,
      "step": 2495
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0386004447937012,
      "learning_rate": 9.57575930091856e-06,
      "loss": 0.5678,
      "step": 2496
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0411202907562256,
      "learning_rate": 9.575345617036413e-06,
      "loss": 0.5939,
      "step": 2497
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.019171118736267,
      "learning_rate": 9.574931740502383e-06,
      "loss": 0.567,
      "step": 2498
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0830127000808716,
      "learning_rate": 9.57451767133389e-06,
      "loss": 0.5962,
      "step": 2499
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.036673665046692,
      "learning_rate": 9.57410340954837e-06,
      "loss": 0.5726,
      "step": 2500
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9850932955741882,
      "learning_rate": 9.57368895516327e-06,
      "loss": 0.5759,
      "step": 2501
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1040841341018677,
      "learning_rate": 9.573274308196037e-06,
      "loss": 0.554,
      "step": 2502
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9730511903762817,
      "learning_rate": 9.572859468664133e-06,
      "loss": 0.5287,
      "step": 2503
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0098161697387695,
      "learning_rate": 9.572444436585025e-06,
      "loss": 0.6103,
      "step": 2504
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0337492227554321,
      "learning_rate": 9.572029211976189e-06,
      "loss": 0.5723,
      "step": 2505
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9878047704696655,
      "learning_rate": 9.571613794855105e-06,
      "loss": 0.5733,
      "step": 2506
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0235744714736938,
      "learning_rate": 9.57119818523927e-06,
      "loss": 0.547,
      "step": 2507
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.025876522064209,
      "learning_rate": 9.570782383146183e-06,
      "loss": 0.5847,
      "step": 2508
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1984429359436035,
      "learning_rate": 9.570366388593347e-06,
      "loss": 0.6102,
      "step": 2509
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.136934518814087,
      "learning_rate": 9.569950201598283e-06,
      "loss": 0.5599,
      "step": 2510
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0273244380950928,
      "learning_rate": 9.569533822178513e-06,
      "loss": 0.5903,
      "step": 2511
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9874576330184937,
      "learning_rate": 9.569117250351571e-06,
      "loss": 0.5544,
      "step": 2512
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.027095079421997,
      "learning_rate": 9.568700486134996e-06,
      "loss": 0.5663,
      "step": 2513
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0826996564865112,
      "learning_rate": 9.568283529546336e-06,
      "loss": 0.5274,
      "step": 2514
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0489357709884644,
      "learning_rate": 9.56786638060315e-06,
      "loss": 0.5844,
      "step": 2515
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0651934146881104,
      "learning_rate": 9.567449039323e-06,
      "loss": 0.6068,
      "step": 2516
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9960207939147949,
      "learning_rate": 9.56703150572346e-06,
      "loss": 0.5678,
      "step": 2517
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.075217843055725,
      "learning_rate": 9.56661377982211e-06,
      "loss": 0.5541,
      "step": 2518
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0845835208892822,
      "learning_rate": 9.566195861636542e-06,
      "loss": 0.6073,
      "step": 2519
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0700103044509888,
      "learning_rate": 9.56577775118435e-06,
      "loss": 0.5778,
      "step": 2520
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1105780601501465,
      "learning_rate": 9.56535944848314e-06,
      "loss": 0.6404,
      "step": 2521
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.961463987827301,
      "learning_rate": 9.564940953550525e-06,
      "loss": 0.5808,
      "step": 2522
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9934834837913513,
      "learning_rate": 9.564522266404127e-06,
      "loss": 0.5474,
      "step": 2523
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0321897268295288,
      "learning_rate": 9.564103387061575e-06,
      "loss": 0.5176,
      "step": 2524
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2021677494049072,
      "learning_rate": 9.563684315540507e-06,
      "loss": 0.543,
      "step": 2525
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1043858528137207,
      "learning_rate": 9.563265051858569e-06,
      "loss": 0.5861,
      "step": 2526
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.026528239250183,
      "learning_rate": 9.562845596033413e-06,
      "loss": 0.612,
      "step": 2527
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0612083673477173,
      "learning_rate": 9.562425948082702e-06,
      "loss": 0.5459,
      "step": 2528
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9829614758491516,
      "learning_rate": 9.562006108024106e-06,
      "loss": 0.5407,
      "step": 2529
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1292877197265625,
      "learning_rate": 9.561586075875304e-06,
      "loss": 0.5878,
      "step": 2530
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.096125602722168,
      "learning_rate": 9.56116585165398e-06,
      "loss": 0.5269,
      "step": 2531
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1311432123184204,
      "learning_rate": 9.560745435377828e-06,
      "loss": 0.5386,
      "step": 2532
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0907361507415771,
      "learning_rate": 9.560324827064553e-06,
      "loss": 0.5681,
      "step": 2533
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0219597816467285,
      "learning_rate": 9.559904026731862e-06,
      "loss": 0.5279,
      "step": 2534
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9601796865463257,
      "learning_rate": 9.559483034397477e-06,
      "loss": 0.5515,
      "step": 2535
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9654351472854614,
      "learning_rate": 9.559061850079121e-06,
      "loss": 0.5096,
      "step": 2536
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0824660062789917,
      "learning_rate": 9.558640473794533e-06,
      "loss": 0.5346,
      "step": 2537
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0912461280822754,
      "learning_rate": 9.558218905561452e-06,
      "loss": 0.5576,
      "step": 2538
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0075145959854126,
      "learning_rate": 9.557797145397629e-06,
      "loss": 0.5801,
      "step": 2539
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.053517460823059,
      "learning_rate": 9.557375193320824e-06,
      "loss": 0.5874,
      "step": 2540
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0725799798965454,
      "learning_rate": 9.556953049348803e-06,
      "loss": 0.5453,
      "step": 2541
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0406394004821777,
      "learning_rate": 9.556530713499341e-06,
      "loss": 0.5786,
      "step": 2542
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1403875350952148,
      "learning_rate": 9.556108185790223e-06,
      "loss": 0.639,
      "step": 2543
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0901226997375488,
      "learning_rate": 9.55568546623924e-06,
      "loss": 0.613,
      "step": 2544
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.057421326637268,
      "learning_rate": 9.555262554864188e-06,
      "loss": 0.5585,
      "step": 2545
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.002384901046753,
      "learning_rate": 9.554839451682876e-06,
      "loss": 0.5107,
      "step": 2546
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.971389889717102,
      "learning_rate": 9.554416156713121e-06,
      "loss": 0.5524,
      "step": 2547
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.013134241104126,
      "learning_rate": 9.553992669972744e-06,
      "loss": 0.5461,
      "step": 2548
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9925364255905151,
      "learning_rate": 9.55356899147958e-06,
      "loss": 0.568,
      "step": 2549
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0054972171783447,
      "learning_rate": 9.553145121251465e-06,
      "loss": 0.5726,
      "step": 2550
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.061987042427063,
      "learning_rate": 9.552721059306248e-06,
      "loss": 0.5742,
      "step": 2551
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0200176239013672,
      "learning_rate": 9.552296805661787e-06,
      "loss": 0.5705,
      "step": 2552
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1196894645690918,
      "learning_rate": 9.551872360335941e-06,
      "loss": 0.5836,
      "step": 2553
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0350748300552368,
      "learning_rate": 9.551447723346587e-06,
      "loss": 0.5628,
      "step": 2554
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0239752531051636,
      "learning_rate": 9.5510228947116e-06,
      "loss": 0.572,
      "step": 2555
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.016221284866333,
      "learning_rate": 9.550597874448874e-06,
      "loss": 0.5742,
      "step": 2556
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0235182046890259,
      "learning_rate": 9.5501726625763e-06,
      "loss": 0.5836,
      "step": 2557
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0047327280044556,
      "learning_rate": 9.549747259111786e-06,
      "loss": 0.5531,
      "step": 2558
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0238168239593506,
      "learning_rate": 9.54932166407324e-06,
      "loss": 0.5564,
      "step": 2559
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0174661874771118,
      "learning_rate": 9.548895877478585e-06,
      "loss": 0.5585,
      "step": 2560
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0724985599517822,
      "learning_rate": 9.54846989934575e-06,
      "loss": 0.6014,
      "step": 2561
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3581907749176025,
      "learning_rate": 9.54804372969267e-06,
      "loss": 0.573,
      "step": 2562
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0653820037841797,
      "learning_rate": 9.54761736853729e-06,
      "loss": 0.5453,
      "step": 2563
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9831539392471313,
      "learning_rate": 9.547190815897563e-06,
      "loss": 0.5298,
      "step": 2564
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1154634952545166,
      "learning_rate": 9.54676407179145e-06,
      "loss": 0.6071,
      "step": 2565
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0205329656600952,
      "learning_rate": 9.546337136236916e-06,
      "loss": 0.5596,
      "step": 2566
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0515823364257812,
      "learning_rate": 9.545910009251945e-06,
      "loss": 0.5866,
      "step": 2567
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0403087139129639,
      "learning_rate": 9.545482690854513e-06,
      "loss": 0.5738,
      "step": 2568
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0300084352493286,
      "learning_rate": 9.545055181062621e-06,
      "loss": 0.5697,
      "step": 2569
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0895107984542847,
      "learning_rate": 9.544627479894264e-06,
      "loss": 0.5992,
      "step": 2570
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9808819890022278,
      "learning_rate": 9.544199587367455e-06,
      "loss": 0.5548,
      "step": 2571
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9981071352958679,
      "learning_rate": 9.54377150350021e-06,
      "loss": 0.586,
      "step": 2572
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0077358484268188,
      "learning_rate": 9.543343228310551e-06,
      "loss": 0.5069,
      "step": 2573
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0661660432815552,
      "learning_rate": 9.542914761816518e-06,
      "loss": 0.5892,
      "step": 2574
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0031508207321167,
      "learning_rate": 9.542486104036143e-06,
      "loss": 0.5557,
      "step": 2575
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.002708077430725,
      "learning_rate": 9.542057254987485e-06,
      "loss": 0.5689,
      "step": 2576
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0708394050598145,
      "learning_rate": 9.541628214688595e-06,
      "loss": 0.5536,
      "step": 2577
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9945730566978455,
      "learning_rate": 9.541198983157538e-06,
      "loss": 0.5941,
      "step": 2578
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0138047933578491,
      "learning_rate": 9.54076956041239e-06,
      "loss": 0.5696,
      "step": 2579
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9891455769538879,
      "learning_rate": 9.540339946471235e-06,
      "loss": 0.5547,
      "step": 2580
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.969589114189148,
      "learning_rate": 9.539910141352156e-06,
      "loss": 0.5695,
      "step": 2581
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9683725833892822,
      "learning_rate": 9.539480145073257e-06,
      "loss": 0.534,
      "step": 2582
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0561916828155518,
      "learning_rate": 9.53904995765264e-06,
      "loss": 0.5461,
      "step": 2583
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.119350790977478,
      "learning_rate": 9.538619579108417e-06,
      "loss": 0.6128,
      "step": 2584
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0886105298995972,
      "learning_rate": 9.538189009458715e-06,
      "loss": 0.6064,
      "step": 2585
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.058983564376831,
      "learning_rate": 9.53775824872166e-06,
      "loss": 0.5322,
      "step": 2586
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0209020376205444,
      "learning_rate": 9.53732729691539e-06,
      "loss": 0.5659,
      "step": 2587
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0355514287948608,
      "learning_rate": 9.536896154058053e-06,
      "loss": 0.559,
      "step": 2588
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0672528743743896,
      "learning_rate": 9.536464820167804e-06,
      "loss": 0.5699,
      "step": 2589
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9848209619522095,
      "learning_rate": 9.536033295262799e-06,
      "loss": 0.5106,
      "step": 2590
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0069193840026855,
      "learning_rate": 9.535601579361214e-06,
      "loss": 0.5202,
      "step": 2591
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0515129566192627,
      "learning_rate": 9.535169672481222e-06,
      "loss": 0.5657,
      "step": 2592
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.049230933189392,
      "learning_rate": 9.534737574641014e-06,
      "loss": 0.5897,
      "step": 2593
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0274021625518799,
      "learning_rate": 9.53430528585878e-06,
      "loss": 0.5719,
      "step": 2594
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9516419172286987,
      "learning_rate": 9.533872806152727e-06,
      "loss": 0.5474,
      "step": 2595
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0155837535858154,
      "learning_rate": 9.533440135541059e-06,
      "loss": 0.5569,
      "step": 2596
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0431528091430664,
      "learning_rate": 9.533007274042e-06,
      "loss": 0.6354,
      "step": 2597
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1031321287155151,
      "learning_rate": 9.532574221673772e-06,
      "loss": 0.563,
      "step": 2598
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1353846788406372,
      "learning_rate": 9.532140978454614e-06,
      "loss": 0.5598,
      "step": 2599
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.140213131904602,
      "learning_rate": 9.531707544402762e-06,
      "loss": 0.5997,
      "step": 2600
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0954878330230713,
      "learning_rate": 9.531273919536473e-06,
      "loss": 0.5664,
      "step": 2601
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9809207916259766,
      "learning_rate": 9.530840103874001e-06,
      "loss": 0.5823,
      "step": 2602
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0627355575561523,
      "learning_rate": 9.530406097433615e-06,
      "loss": 0.514,
      "step": 2603
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0106245279312134,
      "learning_rate": 9.529971900233587e-06,
      "loss": 0.5331,
      "step": 2604
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0271943807601929,
      "learning_rate": 9.529537512292201e-06,
      "loss": 0.5586,
      "step": 2605
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0283169746398926,
      "learning_rate": 9.529102933627747e-06,
      "loss": 0.58,
      "step": 2606
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0217669010162354,
      "learning_rate": 9.528668164258525e-06,
      "loss": 0.572,
      "step": 2607
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9806956052780151,
      "learning_rate": 9.528233204202842e-06,
      "loss": 0.5579,
      "step": 2608
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9664952158927917,
      "learning_rate": 9.527798053479009e-06,
      "loss": 0.5279,
      "step": 2609
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0128159523010254,
      "learning_rate": 9.527362712105353e-06,
      "loss": 0.5462,
      "step": 2610
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0592056512832642,
      "learning_rate": 9.5269271801002e-06,
      "loss": 0.5338,
      "step": 2611
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0758192539215088,
      "learning_rate": 9.526491457481895e-06,
      "loss": 0.5544,
      "step": 2612
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9766221046447754,
      "learning_rate": 9.526055544268778e-06,
      "loss": 0.547,
      "step": 2613
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.972629964351654,
      "learning_rate": 9.525619440479209e-06,
      "loss": 0.5373,
      "step": 2614
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.006587028503418,
      "learning_rate": 9.525183146131549e-06,
      "loss": 0.6073,
      "step": 2615
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0332380533218384,
      "learning_rate": 9.524746661244166e-06,
      "loss": 0.5731,
      "step": 2616
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0720937252044678,
      "learning_rate": 9.524309985835444e-06,
      "loss": 0.5515,
      "step": 2617
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0120248794555664,
      "learning_rate": 9.523873119923768e-06,
      "loss": 0.5364,
      "step": 2618
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0338369607925415,
      "learning_rate": 9.523436063527531e-06,
      "loss": 0.532,
      "step": 2619
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.073209524154663,
      "learning_rate": 9.522998816665137e-06,
      "loss": 0.5833,
      "step": 2620
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9193693995475769,
      "learning_rate": 9.522561379354997e-06,
      "loss": 0.5559,
      "step": 2621
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.032959222793579,
      "learning_rate": 9.522123751615532e-06,
      "loss": 0.5919,
      "step": 2622
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9943687319755554,
      "learning_rate": 9.521685933465166e-06,
      "loss": 0.5811,
      "step": 2623
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0560085773468018,
      "learning_rate": 9.521247924922334e-06,
      "loss": 0.562,
      "step": 2624
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0780750513076782,
      "learning_rate": 9.520809726005481e-06,
      "loss": 0.5474,
      "step": 2625
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0348384380340576,
      "learning_rate": 9.52037133673306e-06,
      "loss": 0.5089,
      "step": 2626
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9498501420021057,
      "learning_rate": 9.519932757123523e-06,
      "loss": 0.5973,
      "step": 2627
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.070290446281433,
      "learning_rate": 9.519493987195343e-06,
      "loss": 0.5405,
      "step": 2628
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9983610510826111,
      "learning_rate": 9.519055026966995e-06,
      "loss": 0.5381,
      "step": 2629
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0487020015716553,
      "learning_rate": 9.518615876456958e-06,
      "loss": 0.5699,
      "step": 2630
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.965333878993988,
      "learning_rate": 9.518176535683727e-06,
      "loss": 0.5298,
      "step": 2631
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9889441728591919,
      "learning_rate": 9.5177370046658e-06,
      "loss": 0.5948,
      "step": 2632
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0800920724868774,
      "learning_rate": 9.517297283421681e-06,
      "loss": 0.5983,
      "step": 2633
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0012084245681763,
      "learning_rate": 9.51685737196989e-06,
      "loss": 0.56,
      "step": 2634
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9588029980659485,
      "learning_rate": 9.516417270328948e-06,
      "loss": 0.5488,
      "step": 2635
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.028010606765747,
      "learning_rate": 9.515976978517387e-06,
      "loss": 0.5601,
      "step": 2636
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9502065181732178,
      "learning_rate": 9.515536496553744e-06,
      "loss": 0.5276,
      "step": 2637
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0674821138381958,
      "learning_rate": 9.515095824456568e-06,
      "loss": 0.5827,
      "step": 2638
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9957113862037659,
      "learning_rate": 9.514654962244414e-06,
      "loss": 0.5788,
      "step": 2639
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9970918893814087,
      "learning_rate": 9.514213909935843e-06,
      "loss": 0.5072,
      "step": 2640
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0630422830581665,
      "learning_rate": 9.51377266754943e-06,
      "loss": 0.5901,
      "step": 2641
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.049082636833191,
      "learning_rate": 9.513331235103751e-06,
      "loss": 0.556,
      "step": 2642
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1107585430145264,
      "learning_rate": 9.512889612617397e-06,
      "loss": 0.5573,
      "step": 2643
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.990415096282959,
      "learning_rate": 9.512447800108958e-06,
      "loss": 0.5591,
      "step": 2644
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0263445377349854,
      "learning_rate": 9.512005797597042e-06,
      "loss": 0.5828,
      "step": 2645
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.073521375656128,
      "learning_rate": 9.511563605100255e-06,
      "loss": 0.5681,
      "step": 2646
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.095017671585083,
      "learning_rate": 9.511121222637222e-06,
      "loss": 0.5948,
      "step": 2647
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9400439858436584,
      "learning_rate": 9.510678650226567e-06,
      "loss": 0.5077,
      "step": 2648
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0273232460021973,
      "learning_rate": 9.510235887886923e-06,
      "loss": 0.5506,
      "step": 2649
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.04499089717865,
      "learning_rate": 9.509792935636939e-06,
      "loss": 0.5314,
      "step": 2650
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9724092483520508,
      "learning_rate": 9.50934979349526e-06,
      "loss": 0.5165,
      "step": 2651
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1168352365493774,
      "learning_rate": 9.508906461480549e-06,
      "loss": 0.6275,
      "step": 2652
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1057885885238647,
      "learning_rate": 9.508462939611473e-06,
      "loss": 0.5357,
      "step": 2653
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1065791845321655,
      "learning_rate": 9.508019227906706e-06,
      "loss": 0.5996,
      "step": 2654
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.092964768409729,
      "learning_rate": 9.507575326384932e-06,
      "loss": 0.5046,
      "step": 2655
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1059108972549438,
      "learning_rate": 9.507131235064842e-06,
      "loss": 0.5753,
      "step": 2656
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0257097482681274,
      "learning_rate": 9.506686953965134e-06,
      "loss": 0.5683,
      "step": 2657
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9293850064277649,
      "learning_rate": 9.506242483104517e-06,
      "loss": 0.4773,
      "step": 2658
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.008527159690857,
      "learning_rate": 9.505797822501704e-06,
      "loss": 0.5724,
      "step": 2659
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0008771419525146,
      "learning_rate": 9.505352972175419e-06,
      "loss": 0.5916,
      "step": 2660
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0681480169296265,
      "learning_rate": 9.504907932144394e-06,
      "loss": 0.5959,
      "step": 2661
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0716049671173096,
      "learning_rate": 9.504462702427369e-06,
      "loss": 0.5705,
      "step": 2662
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.114532470703125,
      "learning_rate": 9.504017283043087e-06,
      "loss": 0.5862,
      "step": 2663
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0168551206588745,
      "learning_rate": 9.503571674010305e-06,
      "loss": 0.5815,
      "step": 2664
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9963028430938721,
      "learning_rate": 9.503125875347789e-06,
      "loss": 0.5591,
      "step": 2665
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1091727018356323,
      "learning_rate": 9.502679887074306e-06,
      "loss": 0.5815,
      "step": 2666
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0106605291366577,
      "learning_rate": 9.502233709208637e-06,
      "loss": 0.6212,
      "step": 2667
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.087552547454834,
      "learning_rate": 9.50178734176957e-06,
      "loss": 0.5382,
      "step": 2668
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.003647804260254,
      "learning_rate": 9.501340784775896e-06,
      "loss": 0.5655,
      "step": 2669
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0045603513717651,
      "learning_rate": 9.500894038246424e-06,
      "loss": 0.5292,
      "step": 2670
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0269829034805298,
      "learning_rate": 9.50044710219996e-06,
      "loss": 0.5717,
      "step": 2671
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0510656833648682,
      "learning_rate": 9.499999976655324e-06,
      "loss": 0.5478,
      "step": 2672
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9509247541427612,
      "learning_rate": 9.499552661631342e-06,
      "loss": 0.547,
      "step": 2673
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1670129299163818,
      "learning_rate": 9.49910515714685e-06,
      "loss": 0.535,
      "step": 2674
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9437034726142883,
      "learning_rate": 9.498657463220694e-06,
      "loss": 0.5333,
      "step": 2675
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.115921139717102,
      "learning_rate": 9.49820957987172e-06,
      "loss": 0.5674,
      "step": 2676
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0253336429595947,
      "learning_rate": 9.49776150711879e-06,
      "loss": 0.5207,
      "step": 2677
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.922319769859314,
      "learning_rate": 9.497313244980768e-06,
      "loss": 0.5107,
      "step": 2678
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.033787727355957,
      "learning_rate": 9.496864793476532e-06,
      "loss": 0.5775,
      "step": 2679
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.230092167854309,
      "learning_rate": 9.49641615262496e-06,
      "loss": 0.5708,
      "step": 2680
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9727022051811218,
      "learning_rate": 9.49596732244495e-06,
      "loss": 0.5258,
      "step": 2681
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0040794610977173,
      "learning_rate": 9.495518302955393e-06,
      "loss": 0.5962,
      "step": 2682
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0190296173095703,
      "learning_rate": 9.4950690941752e-06,
      "loss": 0.5857,
      "step": 2683
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1475651264190674,
      "learning_rate": 9.494619696123286e-06,
      "loss": 0.5803,
      "step": 2684
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0474387407302856,
      "learning_rate": 9.49417010881857e-06,
      "loss": 0.5363,
      "step": 2685
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9686734080314636,
      "learning_rate": 9.493720332279987e-06,
      "loss": 0.5116,
      "step": 2686
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0977166891098022,
      "learning_rate": 9.493270366526471e-06,
      "loss": 0.5734,
      "step": 2687
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.140429139137268,
      "learning_rate": 9.492820211576971e-06,
      "loss": 0.552,
      "step": 2688
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.088618278503418,
      "learning_rate": 9.492369867450444e-06,
      "loss": 0.5414,
      "step": 2689
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1019575595855713,
      "learning_rate": 9.491919334165846e-06,
      "loss": 0.5658,
      "step": 2690
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0340977907180786,
      "learning_rate": 9.491468611742154e-06,
      "loss": 0.6331,
      "step": 2691
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.086870789527893,
      "learning_rate": 9.491017700198343e-06,
      "loss": 0.6129,
      "step": 2692
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0830391645431519,
      "learning_rate": 9.490566599553399e-06,
      "loss": 0.5659,
      "step": 2693
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0409371852874756,
      "learning_rate": 9.490115309826317e-06,
      "loss": 0.5554,
      "step": 2694
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0426156520843506,
      "learning_rate": 9.4896638310361e-06,
      "loss": 0.5665,
      "step": 2695
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0904855728149414,
      "learning_rate": 9.489212163201758e-06,
      "loss": 0.5656,
      "step": 2696
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0347099304199219,
      "learning_rate": 9.488760306342307e-06,
      "loss": 0.5764,
      "step": 2697
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.05708909034729,
      "learning_rate": 9.488308260476776e-06,
      "loss": 0.5354,
      "step": 2698
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0091209411621094,
      "learning_rate": 9.487856025624196e-06,
      "loss": 0.5512,
      "step": 2699
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0588990449905396,
      "learning_rate": 9.487403601803614e-06,
      "loss": 0.5239,
      "step": 2700
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0110063552856445,
      "learning_rate": 9.486950989034074e-06,
      "loss": 0.5683,
      "step": 2701
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9574355483055115,
      "learning_rate": 9.486498187334636e-06,
      "loss": 0.5331,
      "step": 2702
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.005077838897705,
      "learning_rate": 9.48604519672437e-06,
      "loss": 0.5988,
      "step": 2703
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0744143724441528,
      "learning_rate": 9.485592017222344e-06,
      "loss": 0.6271,
      "step": 2704
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.054773211479187,
      "learning_rate": 9.485138648847643e-06,
      "loss": 0.5351,
      "step": 2705
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9953201413154602,
      "learning_rate": 9.484685091619358e-06,
      "loss": 0.5487,
      "step": 2706
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0761560201644897,
      "learning_rate": 9.484231345556582e-06,
      "loss": 0.57,
      "step": 2707
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0151156187057495,
      "learning_rate": 9.483777410678427e-06,
      "loss": 0.5624,
      "step": 2708
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.977360188961029,
      "learning_rate": 9.483323287004001e-06,
      "loss": 0.5196,
      "step": 2709
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0837485790252686,
      "learning_rate": 9.482868974552427e-06,
      "loss": 0.5727,
      "step": 2710
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1661759614944458,
      "learning_rate": 9.482414473342835e-06,
      "loss": 0.6477,
      "step": 2711
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0656156539916992,
      "learning_rate": 9.481959783394365e-06,
      "loss": 0.6108,
      "step": 2712
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1635092496871948,
      "learning_rate": 9.48150490472616e-06,
      "loss": 0.5768,
      "step": 2713
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0615007877349854,
      "learning_rate": 9.481049837357371e-06,
      "loss": 0.5624,
      "step": 2714
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9256333112716675,
      "learning_rate": 9.480594581307164e-06,
      "loss": 0.5176,
      "step": 2715
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9532365202903748,
      "learning_rate": 9.480139136594706e-06,
      "loss": 0.5315,
      "step": 2716
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0971673727035522,
      "learning_rate": 9.479683503239172e-06,
      "loss": 0.5831,
      "step": 2717
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0594340562820435,
      "learning_rate": 9.479227681259751e-06,
      "loss": 0.5991,
      "step": 2718
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9971533417701721,
      "learning_rate": 9.478771670675635e-06,
      "loss": 0.5632,
      "step": 2719
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9794123768806458,
      "learning_rate": 9.478315471506023e-06,
      "loss": 0.5368,
      "step": 2720
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9903360605239868,
      "learning_rate": 9.477859083770126e-06,
      "loss": 0.5879,
      "step": 2721
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0230356454849243,
      "learning_rate": 9.477402507487162e-06,
      "loss": 0.5509,
      "step": 2722
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0338963270187378,
      "learning_rate": 9.476945742676352e-06,
      "loss": 0.5207,
      "step": 2723
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.096906304359436,
      "learning_rate": 9.476488789356933e-06,
      "loss": 0.5387,
      "step": 2724
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0951502323150635,
      "learning_rate": 9.47603164754814e-06,
      "loss": 0.6059,
      "step": 2725
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9961447715759277,
      "learning_rate": 9.47557431726923e-06,
      "loss": 0.5728,
      "step": 2726
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1282474994659424,
      "learning_rate": 9.475116798539451e-06,
      "loss": 0.6234,
      "step": 2727
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0593305826187134,
      "learning_rate": 9.474659091378074e-06,
      "loss": 0.5426,
      "step": 2728
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0862617492675781,
      "learning_rate": 9.474201195804367e-06,
      "loss": 0.6038,
      "step": 2729
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0345958471298218,
      "learning_rate": 9.473743111837612e-06,
      "loss": 0.577,
      "step": 2730
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9704797267913818,
      "learning_rate": 9.4732848394971e-06,
      "loss": 0.5234,
      "step": 2731
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9978111386299133,
      "learning_rate": 9.472826378802122e-06,
      "loss": 0.5623,
      "step": 2732
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0373756885528564,
      "learning_rate": 9.472367729771987e-06,
      "loss": 0.578,
      "step": 2733
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0830659866333008,
      "learning_rate": 9.471908892426005e-06,
      "loss": 0.5932,
      "step": 2734
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0066198110580444,
      "learning_rate": 9.471449866783495e-06,
      "loss": 0.5549,
      "step": 2735
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0600615739822388,
      "learning_rate": 9.470990652863787e-06,
      "loss": 0.6033,
      "step": 2736
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1532150506973267,
      "learning_rate": 9.470531250686216e-06,
      "loss": 0.5851,
      "step": 2737
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1124008893966675,
      "learning_rate": 9.470071660270126e-06,
      "loss": 0.5786,
      "step": 2738
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9894224405288696,
      "learning_rate": 9.469611881634868e-06,
      "loss": 0.5439,
      "step": 2739
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0182647705078125,
      "learning_rate": 9.469151914799803e-06,
      "loss": 0.5386,
      "step": 2740
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.978829562664032,
      "learning_rate": 9.468691759784298e-06,
      "loss": 0.5712,
      "step": 2741
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0401105880737305,
      "learning_rate": 9.468231416607727e-06,
      "loss": 0.5204,
      "step": 2742
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.068730115890503,
      "learning_rate": 9.467770885289477e-06,
      "loss": 0.5169,
      "step": 2743
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.023269772529602,
      "learning_rate": 9.467310165848935e-06,
      "loss": 0.5856,
      "step": 2744
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0364032983779907,
      "learning_rate": 9.466849258305504e-06,
      "loss": 0.586,
      "step": 2745
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0057346820831299,
      "learning_rate": 9.46638816267859e-06,
      "loss": 0.5624,
      "step": 2746
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.052634835243225,
      "learning_rate": 9.465926878987609e-06,
      "loss": 0.5905,
      "step": 2747
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9901856780052185,
      "learning_rate": 9.46546540725198e-06,
      "loss": 0.5594,
      "step": 2748
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0854971408843994,
      "learning_rate": 9.465003747491138e-06,
      "loss": 0.5894,
      "step": 2749
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0599004030227661,
      "learning_rate": 9.464541899724522e-06,
      "loss": 0.5523,
      "step": 2750
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1270490884780884,
      "learning_rate": 9.464079863971576e-06,
      "loss": 0.5561,
      "step": 2751
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1302123069763184,
      "learning_rate": 9.463617640251756e-06,
      "loss": 0.5472,
      "step": 2752
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.055801510810852,
      "learning_rate": 9.463155228584526e-06,
      "loss": 0.5548,
      "step": 2753
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.117521047592163,
      "learning_rate": 9.462692628989356e-06,
      "loss": 0.5113,
      "step": 2754
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0038899183273315,
      "learning_rate": 9.462229841485723e-06,
      "loss": 0.5977,
      "step": 2755
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.06837797164917,
      "learning_rate": 9.461766866093117e-06,
      "loss": 0.5819,
      "step": 2756
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0148512125015259,
      "learning_rate": 9.461303702831026e-06,
      "loss": 0.5387,
      "step": 2757
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.059126377105713,
      "learning_rate": 9.460840351718958e-06,
      "loss": 0.5427,
      "step": 2758
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0758804082870483,
      "learning_rate": 9.46037681277642e-06,
      "loss": 0.5507,
      "step": 2759
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0124850273132324,
      "learning_rate": 9.459913086022931e-06,
      "loss": 0.5496,
      "step": 2760
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9992918968200684,
      "learning_rate": 9.459449171478017e-06,
      "loss": 0.4975,
      "step": 2761
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9648446440696716,
      "learning_rate": 9.458985069161212e-06,
      "loss": 0.5103,
      "step": 2762
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1720576286315918,
      "learning_rate": 9.458520779092057e-06,
      "loss": 0.5341,
      "step": 2763
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0725833177566528,
      "learning_rate": 9.458056301290102e-06,
      "loss": 0.6193,
      "step": 2764
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9970990419387817,
      "learning_rate": 9.457591635774905e-06,
      "loss": 0.6286,
      "step": 2765
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0215911865234375,
      "learning_rate": 9.457126782566031e-06,
      "loss": 0.6008,
      "step": 2766
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0754748582839966,
      "learning_rate": 9.456661741683054e-06,
      "loss": 0.5929,
      "step": 2767
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.061965823173523,
      "learning_rate": 9.456196513145553e-06,
      "loss": 0.5575,
      "step": 2768
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0222315788269043,
      "learning_rate": 9.455731096973119e-06,
      "loss": 0.566,
      "step": 2769
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.079120397567749,
      "learning_rate": 9.455265493185349e-06,
      "loss": 0.5987,
      "step": 2770
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.013046383857727,
      "learning_rate": 9.454799701801849e-06,
      "loss": 0.5493,
      "step": 2771
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0657896995544434,
      "learning_rate": 9.45433372284223e-06,
      "loss": 0.5631,
      "step": 2772
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9840913414955139,
      "learning_rate": 9.453867556326113e-06,
      "loss": 0.5552,
      "step": 2773
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.123887062072754,
      "learning_rate": 9.453401202273127e-06,
      "loss": 0.5443,
      "step": 2774
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0064541101455688,
      "learning_rate": 9.45293466070291e-06,
      "loss": 0.5392,
      "step": 2775
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1079277992248535,
      "learning_rate": 9.452467931635104e-06,
      "loss": 0.595,
      "step": 2776
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9934383034706116,
      "learning_rate": 9.452001015089363e-06,
      "loss": 0.5407,
      "step": 2777
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.04298996925354,
      "learning_rate": 9.451533911085346e-06,
      "loss": 0.5536,
      "step": 2778
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1786850690841675,
      "learning_rate": 9.451066619642721e-06,
      "loss": 0.5682,
      "step": 2779
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0571482181549072,
      "learning_rate": 9.450599140781166e-06,
      "loss": 0.5803,
      "step": 2780
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9489169716835022,
      "learning_rate": 9.450131474520364e-06,
      "loss": 0.5458,
      "step": 2781
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9808425903320312,
      "learning_rate": 9.449663620880006e-06,
      "loss": 0.5421,
      "step": 2782
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9377076625823975,
      "learning_rate": 9.449195579879793e-06,
      "loss": 0.5443,
      "step": 2783
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0947315692901611,
      "learning_rate": 9.448727351539431e-06,
      "loss": 0.5892,
      "step": 2784
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0158424377441406,
      "learning_rate": 9.448258935878635e-06,
      "loss": 0.5569,
      "step": 2785
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0767693519592285,
      "learning_rate": 9.44779033291713e-06,
      "loss": 0.5791,
      "step": 2786
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0848816633224487,
      "learning_rate": 9.447321542674647e-06,
      "loss": 0.5863,
      "step": 2787
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.111331582069397,
      "learning_rate": 9.446852565170928e-06,
      "loss": 0.5767,
      "step": 2788
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0502716302871704,
      "learning_rate": 9.446383400425713e-06,
      "loss": 0.6005,
      "step": 2789
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0394223928451538,
      "learning_rate": 9.445914048458764e-06,
      "loss": 0.5051,
      "step": 2790
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9824061393737793,
      "learning_rate": 9.445444509289838e-06,
      "loss": 0.5472,
      "step": 2791
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0843420028686523,
      "learning_rate": 9.44497478293871e-06,
      "loss": 0.5515,
      "step": 2792
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0395381450653076,
      "learning_rate": 9.444504869425154e-06,
      "loss": 0.577,
      "step": 2793
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0165070295333862,
      "learning_rate": 9.44403476876896e-06,
      "loss": 0.5736,
      "step": 2794
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9988610148429871,
      "learning_rate": 9.443564480989924e-06,
      "loss": 0.55,
      "step": 2795
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9660263061523438,
      "learning_rate": 9.443094006107844e-06,
      "loss": 0.5494,
      "step": 2796
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9972324967384338,
      "learning_rate": 9.442623344142534e-06,
      "loss": 0.545,
      "step": 2797
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.038231611251831,
      "learning_rate": 9.442152495113808e-06,
      "loss": 0.5573,
      "step": 2798
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9843300580978394,
      "learning_rate": 9.441681459041494e-06,
      "loss": 0.5879,
      "step": 2799
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1102025508880615,
      "learning_rate": 9.441210235945425e-06,
      "loss": 0.5593,
      "step": 2800
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9363767504692078,
      "learning_rate": 9.440738825845441e-06,
      "loss": 0.5394,
      "step": 2801
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1123851537704468,
      "learning_rate": 9.440267228761395e-06,
      "loss": 0.5985,
      "step": 2802
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0247012376785278,
      "learning_rate": 9.439795444713143e-06,
      "loss": 0.5928,
      "step": 2803
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0402313470840454,
      "learning_rate": 9.43932347372055e-06,
      "loss": 0.5614,
      "step": 2804
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.025115728378296,
      "learning_rate": 9.438851315803488e-06,
      "loss": 0.5049,
      "step": 2805
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9654614329338074,
      "learning_rate": 9.438378970981839e-06,
      "loss": 0.5418,
      "step": 2806
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0685309171676636,
      "learning_rate": 9.43790643927549e-06,
      "loss": 0.6401,
      "step": 2807
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0241663455963135,
      "learning_rate": 9.437433720704342e-06,
      "loss": 0.5294,
      "step": 2808
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1073213815689087,
      "learning_rate": 9.436960815288294e-06,
      "loss": 0.5465,
      "step": 2809
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0069595575332642,
      "learning_rate": 9.436487723047263e-06,
      "loss": 0.5578,
      "step": 2810
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0204834938049316,
      "learning_rate": 9.436014444001167e-06,
      "loss": 0.5714,
      "step": 2811
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0144668817520142,
      "learning_rate": 9.435540978169933e-06,
      "loss": 0.5524,
      "step": 2812
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0032633543014526,
      "learning_rate": 9.435067325573499e-06,
      "loss": 0.5966,
      "step": 2813
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9835605621337891,
      "learning_rate": 9.43459348623181e-06,
      "loss": 0.6052,
      "step": 2814
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0459119081497192,
      "learning_rate": 9.434119460164816e-06,
      "loss": 0.5217,
      "step": 2815
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0917023420333862,
      "learning_rate": 9.433645247392476e-06,
      "loss": 0.543,
      "step": 2816
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0550017356872559,
      "learning_rate": 9.433170847934759e-06,
      "loss": 0.6093,
      "step": 2817
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.098501443862915,
      "learning_rate": 9.432696261811637e-06,
      "loss": 0.5914,
      "step": 2818
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0482356548309326,
      "learning_rate": 9.432221489043097e-06,
      "loss": 0.5857,
      "step": 2819
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0572627782821655,
      "learning_rate": 9.43174652964913e-06,
      "loss": 0.5746,
      "step": 2820
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1536352634429932,
      "learning_rate": 9.431271383649731e-06,
      "loss": 0.5707,
      "step": 2821
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.993578314781189,
      "learning_rate": 9.430796051064913e-06,
      "loss": 0.5792,
      "step": 2822
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0542477369308472,
      "learning_rate": 9.430320531914683e-06,
      "loss": 0.5783,
      "step": 2823
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0634816884994507,
      "learning_rate": 9.42984482621907e-06,
      "loss": 0.5851,
      "step": 2824
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9990572333335876,
      "learning_rate": 9.4293689339981e-06,
      "loss": 0.5078,
      "step": 2825
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.158766269683838,
      "learning_rate": 9.428892855271813e-06,
      "loss": 0.6321,
      "step": 2826
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0206190347671509,
      "learning_rate": 9.428416590060256e-06,
      "loss": 0.5622,
      "step": 2827
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0901415348052979,
      "learning_rate": 9.427940138383482e-06,
      "loss": 0.6017,
      "step": 2828
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0550830364227295,
      "learning_rate": 9.427463500261551e-06,
      "loss": 0.6087,
      "step": 2829
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0264674425125122,
      "learning_rate": 9.426986675714535e-06,
      "loss": 0.523,
      "step": 2830
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9756189584732056,
      "learning_rate": 9.426509664762509e-06,
      "loss": 0.4892,
      "step": 2831
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1184206008911133,
      "learning_rate": 9.42603246742556e-06,
      "loss": 0.5733,
      "step": 2832
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1206648349761963,
      "learning_rate": 9.425555083723783e-06,
      "loss": 0.5927,
      "step": 2833
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0198521614074707,
      "learning_rate": 9.425077513677276e-06,
      "loss": 0.5436,
      "step": 2834
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9669837355613708,
      "learning_rate": 9.424599757306148e-06,
      "loss": 0.5645,
      "step": 2835
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0391610860824585,
      "learning_rate": 9.424121814630516e-06,
      "loss": 0.5046,
      "step": 2836
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0172545909881592,
      "learning_rate": 9.423643685670504e-06,
      "loss": 0.5923,
      "step": 2837
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0235159397125244,
      "learning_rate": 9.423165370446249e-06,
      "loss": 0.5294,
      "step": 2838
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0332740545272827,
      "learning_rate": 9.422686868977884e-06,
      "loss": 0.5247,
      "step": 2839
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1540343761444092,
      "learning_rate": 9.42220818128556e-06,
      "loss": 0.5839,
      "step": 2840
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0458693504333496,
      "learning_rate": 9.421729307389435e-06,
      "loss": 0.5583,
      "step": 2841
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0047495365142822,
      "learning_rate": 9.42125024730967e-06,
      "loss": 0.5742,
      "step": 2842
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1194878816604614,
      "learning_rate": 9.420771001066439e-06,
      "loss": 0.6002,
      "step": 2843
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.068681240081787,
      "learning_rate": 9.420291568679917e-06,
      "loss": 0.6148,
      "step": 2844
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0517293214797974,
      "learning_rate": 9.419811950170294e-06,
      "loss": 0.5744,
      "step": 2845
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.161916971206665,
      "learning_rate": 9.419332145557768e-06,
      "loss": 0.5964,
      "step": 2846
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.14397394657135,
      "learning_rate": 9.418852154862538e-06,
      "loss": 0.5905,
      "step": 2847
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0074042081832886,
      "learning_rate": 9.418371978104816e-06,
      "loss": 0.5582,
      "step": 2848
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.015149474143982,
      "learning_rate": 9.41789161530482e-06,
      "loss": 0.5931,
      "step": 2849
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0828512907028198,
      "learning_rate": 9.417411066482777e-06,
      "loss": 0.5581,
      "step": 2850
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0007314682006836,
      "learning_rate": 9.41693033165892e-06,
      "loss": 0.5666,
      "step": 2851
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0738846063613892,
      "learning_rate": 9.416449410853495e-06,
      "loss": 0.5609,
      "step": 2852
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1264728307724,
      "learning_rate": 9.415968304086746e-06,
      "loss": 0.5753,
      "step": 2853
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0523388385772705,
      "learning_rate": 9.415487011378935e-06,
      "loss": 0.5625,
      "step": 2854
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.051973581314087,
      "learning_rate": 9.415005532750326e-06,
      "loss": 0.5322,
      "step": 2855
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1547439098358154,
      "learning_rate": 9.414523868221192e-06,
      "loss": 0.623,
      "step": 2856
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9445652365684509,
      "learning_rate": 9.414042017811817e-06,
      "loss": 0.5596,
      "step": 2857
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9419575333595276,
      "learning_rate": 9.413559981542486e-06,
      "loss": 0.5771,
      "step": 2858
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0547152757644653,
      "learning_rate": 9.413077759433498e-06,
      "loss": 0.5862,
      "step": 2859
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1181983947753906,
      "learning_rate": 9.412595351505158e-06,
      "loss": 0.5518,
      "step": 2860
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9860160946846008,
      "learning_rate": 9.412112757777777e-06,
      "loss": 0.5856,
      "step": 2861
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0584853887557983,
      "learning_rate": 9.411629978271679e-06,
      "loss": 0.5884,
      "step": 2862
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.132488489151001,
      "learning_rate": 9.411147013007188e-06,
      "loss": 0.5753,
      "step": 2863
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0454785823822021,
      "learning_rate": 9.41066386200464e-06,
      "loss": 0.5617,
      "step": 2864
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9679178595542908,
      "learning_rate": 9.410180525284384e-06,
      "loss": 0.5144,
      "step": 2865
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9169051051139832,
      "learning_rate": 9.409697002866765e-06,
      "loss": 0.5357,
      "step": 2866
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.045176386833191,
      "learning_rate": 9.409213294772147e-06,
      "loss": 0.5865,
      "step": 2867
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0670503377914429,
      "learning_rate": 9.408729401020896e-06,
      "loss": 0.5824,
      "step": 2868
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.10625422000885,
      "learning_rate": 9.408245321633385e-06,
      "loss": 0.5535,
      "step": 2869
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9990093111991882,
      "learning_rate": 9.407761056629999e-06,
      "loss": 0.5219,
      "step": 2870
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0798190832138062,
      "learning_rate": 9.40727660603113e-06,
      "loss": 0.5537,
      "step": 2871
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.19645094871521,
      "learning_rate": 9.406791969857173e-06,
      "loss": 0.6066,
      "step": 2872
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.039581537246704,
      "learning_rate": 9.406307148128537e-06,
      "loss": 0.5903,
      "step": 2873
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1147894859313965,
      "learning_rate": 9.405822140865636e-06,
      "loss": 0.5871,
      "step": 2874
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0248534679412842,
      "learning_rate": 9.40533694808889e-06,
      "loss": 0.5069,
      "step": 2875
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0174291133880615,
      "learning_rate": 9.404851569818731e-06,
      "loss": 0.5817,
      "step": 2876
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1911200284957886,
      "learning_rate": 9.404366006075596e-06,
      "loss": 0.645,
      "step": 2877
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1297792196273804,
      "learning_rate": 9.403880256879931e-06,
      "loss": 0.6082,
      "step": 2878
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1312628984451294,
      "learning_rate": 9.403394322252186e-06,
      "loss": 0.5478,
      "step": 2879
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9603383541107178,
      "learning_rate": 9.402908202212826e-06,
      "loss": 0.5066,
      "step": 2880
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9875933527946472,
      "learning_rate": 9.402421896782319e-06,
      "loss": 0.5505,
      "step": 2881
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9948558807373047,
      "learning_rate": 9.401935405981138e-06,
      "loss": 0.5821,
      "step": 2882
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1072025299072266,
      "learning_rate": 9.401448729829773e-06,
      "loss": 0.5781,
      "step": 2883
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.002524495124817,
      "learning_rate": 9.400961868348713e-06,
      "loss": 0.5846,
      "step": 2884
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.030574917793274,
      "learning_rate": 9.400474821558457e-06,
      "loss": 0.5437,
      "step": 2885
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0898710489273071,
      "learning_rate": 9.399987589479516e-06,
      "loss": 0.5841,
      "step": 2886
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0254002809524536,
      "learning_rate": 9.399500172132403e-06,
      "loss": 0.5442,
      "step": 2887
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0504629611968994,
      "learning_rate": 9.399012569537643e-06,
      "loss": 0.5413,
      "step": 2888
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.114540457725525,
      "learning_rate": 9.398524781715767e-06,
      "loss": 0.5611,
      "step": 2889
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0113818645477295,
      "learning_rate": 9.398036808687314e-06,
      "loss": 0.5544,
      "step": 2890
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0500532388687134,
      "learning_rate": 9.39754865047283e-06,
      "loss": 0.5893,
      "step": 2891
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9870203733444214,
      "learning_rate": 9.39706030709287e-06,
      "loss": 0.4995,
      "step": 2892
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0355092287063599,
      "learning_rate": 9.396571778567997e-06,
      "loss": 0.5499,
      "step": 2893
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0265161991119385,
      "learning_rate": 9.396083064918782e-06,
      "loss": 0.5549,
      "step": 2894
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.060455560684204,
      "learning_rate": 9.3955941661658e-06,
      "loss": 0.603,
      "step": 2895
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0171563625335693,
      "learning_rate": 9.39510508232964e-06,
      "loss": 0.5435,
      "step": 2896
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9958793520927429,
      "learning_rate": 9.394615813430895e-06,
      "loss": 0.5376,
      "step": 2897
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0014652013778687,
      "learning_rate": 9.394126359490166e-06,
      "loss": 0.5651,
      "step": 2898
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0729966163635254,
      "learning_rate": 9.393636720528061e-06,
      "loss": 0.5663,
      "step": 2899
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0536006689071655,
      "learning_rate": 9.393146896565197e-06,
      "loss": 0.5999,
      "step": 2900
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0769306421279907,
      "learning_rate": 9.392656887622202e-06,
      "loss": 0.5778,
      "step": 2901
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0698580741882324,
      "learning_rate": 9.392166693719706e-06,
      "loss": 0.5471,
      "step": 2902
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9738646745681763,
      "learning_rate": 9.391676314878348e-06,
      "loss": 0.5611,
      "step": 2903
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0991955995559692,
      "learning_rate": 9.391185751118782e-06,
      "loss": 0.6088,
      "step": 2904
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0504570007324219,
      "learning_rate": 9.390695002461657e-06,
      "loss": 0.5671,
      "step": 2905
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9739258885383606,
      "learning_rate": 9.390204068927638e-06,
      "loss": 0.5751,
      "step": 2906
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.022497296333313,
      "learning_rate": 9.389712950537399e-06,
      "loss": 0.5703,
      "step": 2907
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0213046073913574,
      "learning_rate": 9.38922164731162e-06,
      "loss": 0.5869,
      "step": 2908
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9645665287971497,
      "learning_rate": 9.388730159270984e-06,
      "loss": 0.4713,
      "step": 2909
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0854631662368774,
      "learning_rate": 9.38823848643619e-06,
      "loss": 0.5605,
      "step": 2910
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1496248245239258,
      "learning_rate": 9.38774662882794e-06,
      "loss": 0.676,
      "step": 2911
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.986517071723938,
      "learning_rate": 9.387254586466942e-06,
      "loss": 0.5374,
      "step": 2912
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9511660933494568,
      "learning_rate": 9.386762359373915e-06,
      "loss": 0.5468,
      "step": 2913
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9129959940910339,
      "learning_rate": 9.386269947569585e-06,
      "loss": 0.5195,
      "step": 2914
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0946681499481201,
      "learning_rate": 9.385777351074688e-06,
      "loss": 0.5783,
      "step": 2915
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0663743019104004,
      "learning_rate": 9.385284569909963e-06,
      "loss": 0.5628,
      "step": 2916
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9548754096031189,
      "learning_rate": 9.38479160409616e-06,
      "loss": 0.5574,
      "step": 2917
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0848325490951538,
      "learning_rate": 9.384298453654037e-06,
      "loss": 0.5916,
      "step": 2918
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.934111475944519,
      "learning_rate": 9.383805118604357e-06,
      "loss": 0.5702,
      "step": 2919
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.025260090827942,
      "learning_rate": 9.383311598967892e-06,
      "loss": 0.5516,
      "step": 2920
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0210864543914795,
      "learning_rate": 9.382817894765426e-06,
      "loss": 0.6516,
      "step": 2921
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0006023645401,
      "learning_rate": 9.382324006017746e-06,
      "loss": 0.5987,
      "step": 2922
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0651023387908936,
      "learning_rate": 9.381829932745646e-06,
      "loss": 0.5347,
      "step": 2923
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.056115746498108,
      "learning_rate": 9.38133567496993e-06,
      "loss": 0.5696,
      "step": 2924
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0541669130325317,
      "learning_rate": 9.380841232711412e-06,
      "loss": 0.5691,
      "step": 2925
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1002367734909058,
      "learning_rate": 9.38034660599091e-06,
      "loss": 0.5373,
      "step": 2926
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0014314651489258,
      "learning_rate": 9.379851794829247e-06,
      "loss": 0.5326,
      "step": 2927
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0438270568847656,
      "learning_rate": 9.379356799247263e-06,
      "loss": 0.6348,
      "step": 2928
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0377657413482666,
      "learning_rate": 9.3788616192658e-06,
      "loss": 0.5942,
      "step": 2929
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.029208779335022,
      "learning_rate": 9.378366254905706e-06,
      "loss": 0.5666,
      "step": 2930
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0012551546096802,
      "learning_rate": 9.377870706187841e-06,
      "loss": 0.521,
      "step": 2931
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.096712589263916,
      "learning_rate": 9.37737497313307e-06,
      "loss": 0.5228,
      "step": 2932
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.932421863079071,
      "learning_rate": 9.376879055762267e-06,
      "loss": 0.4983,
      "step": 2933
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.041609287261963,
      "learning_rate": 9.376382954096312e-06,
      "loss": 0.5816,
      "step": 2934
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9358463287353516,
      "learning_rate": 9.375886668156095e-06,
      "loss": 0.5758,
      "step": 2935
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.065076231956482,
      "learning_rate": 9.375390197962514e-06,
      "loss": 0.5325,
      "step": 2936
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0218020677566528,
      "learning_rate": 9.374893543536471e-06,
      "loss": 0.538,
      "step": 2937
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0039360523223877,
      "learning_rate": 9.374396704898883e-06,
      "loss": 0.5959,
      "step": 2938
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0219570398330688,
      "learning_rate": 9.373899682070664e-06,
      "loss": 0.5818,
      "step": 2939
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.084692358970642,
      "learning_rate": 9.373402475072746e-06,
      "loss": 0.5754,
      "step": 2940
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9906976819038391,
      "learning_rate": 9.372905083926064e-06,
      "loss": 0.5624,
      "step": 2941
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.009021520614624,
      "learning_rate": 9.372407508651561e-06,
      "loss": 0.5557,
      "step": 2942
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9522027373313904,
      "learning_rate": 9.371909749270189e-06,
      "loss": 0.5369,
      "step": 2943
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0123964548110962,
      "learning_rate": 9.371411805802907e-06,
      "loss": 0.5701,
      "step": 2944
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0369518995285034,
      "learning_rate": 9.370913678270678e-06,
      "loss": 0.5329,
      "step": 2945
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0261355638504028,
      "learning_rate": 9.370415366694483e-06,
      "loss": 0.5628,
      "step": 2946
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0193971395492554,
      "learning_rate": 9.369916871095299e-06,
      "loss": 0.5684,
      "step": 2947
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.948062002658844,
      "learning_rate": 9.369418191494117e-06,
      "loss": 0.5665,
      "step": 2948
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.036242961883545,
      "learning_rate": 9.368919327911934e-06,
      "loss": 0.5659,
      "step": 2949
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.047235131263733,
      "learning_rate": 9.368420280369759e-06,
      "loss": 0.5915,
      "step": 2950
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0826328992843628,
      "learning_rate": 9.3679210488886e-06,
      "loss": 0.567,
      "step": 2951
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9877567887306213,
      "learning_rate": 9.367421633489482e-06,
      "loss": 0.6275,
      "step": 2952
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9576612710952759,
      "learning_rate": 9.366922034193431e-06,
      "loss": 0.5483,
      "step": 2953
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0604145526885986,
      "learning_rate": 9.366422251021486e-06,
      "loss": 0.562,
      "step": 2954
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.986191987991333,
      "learning_rate": 9.365922283994689e-06,
      "loss": 0.5494,
      "step": 2955
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1650367975234985,
      "learning_rate": 9.365422133134093e-06,
      "loss": 0.5687,
      "step": 2956
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0348550081253052,
      "learning_rate": 9.364921798460756e-06,
      "loss": 0.558,
      "step": 2957
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.052424430847168,
      "learning_rate": 9.364421279995747e-06,
      "loss": 0.593,
      "step": 2958
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9829449653625488,
      "learning_rate": 9.36392057776014e-06,
      "loss": 0.5745,
      "step": 2959
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0334904193878174,
      "learning_rate": 9.36341969177502e-06,
      "loss": 0.51,
      "step": 2960
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0335619449615479,
      "learning_rate": 9.362918622061475e-06,
      "loss": 0.535,
      "step": 2961
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9990548491477966,
      "learning_rate": 9.362417368640604e-06,
      "loss": 0.5253,
      "step": 2962
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9685349464416504,
      "learning_rate": 9.361915931533513e-06,
      "loss": 0.5622,
      "step": 2963
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9835405349731445,
      "learning_rate": 9.36141431076132e-06,
      "loss": 0.5842,
      "step": 2964
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0301471948623657,
      "learning_rate": 9.360912506345139e-06,
      "loss": 0.5497,
      "step": 2965
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9751991033554077,
      "learning_rate": 9.360410518306103e-06,
      "loss": 0.5084,
      "step": 2966
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9731338024139404,
      "learning_rate": 9.359908346665349e-06,
      "loss": 0.5161,
      "step": 2967
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9955142140388489,
      "learning_rate": 9.359405991444022e-06,
      "loss": 0.6021,
      "step": 2968
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.101694107055664,
      "learning_rate": 9.358903452663273e-06,
      "loss": 0.6029,
      "step": 2969
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1504160165786743,
      "learning_rate": 9.358400730344265e-06,
      "loss": 0.5574,
      "step": 2970
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9698602557182312,
      "learning_rate": 9.357897824508163e-06,
      "loss": 0.5278,
      "step": 2971
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0260837078094482,
      "learning_rate": 9.357394735176144e-06,
      "loss": 0.5825,
      "step": 2972
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9875388741493225,
      "learning_rate": 9.356891462369391e-06,
      "loss": 0.545,
      "step": 2973
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0932499170303345,
      "learning_rate": 9.356388006109094e-06,
      "loss": 0.5889,
      "step": 2974
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8903037309646606,
      "learning_rate": 9.355884366416454e-06,
      "loss": 0.542,
      "step": 2975
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0137654542922974,
      "learning_rate": 9.355380543312676e-06,
      "loss": 0.6136,
      "step": 2976
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9530428051948547,
      "learning_rate": 9.354876536818974e-06,
      "loss": 0.5744,
      "step": 2977
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9858264923095703,
      "learning_rate": 9.35437234695657e-06,
      "loss": 0.564,
      "step": 2978
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0544084310531616,
      "learning_rate": 9.353867973746696e-06,
      "loss": 0.5591,
      "step": 2979
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.011245846748352,
      "learning_rate": 9.353363417210587e-06,
      "loss": 0.5045,
      "step": 2980
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0920368432998657,
      "learning_rate": 9.352858677369488e-06,
      "loss": 0.5875,
      "step": 2981
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9856042861938477,
      "learning_rate": 9.352353754244654e-06,
      "loss": 0.5399,
      "step": 2982
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.958921492099762,
      "learning_rate": 9.351848647857343e-06,
      "loss": 0.5416,
      "step": 2983
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9829522967338562,
      "learning_rate": 9.351343358228825e-06,
      "loss": 0.5362,
      "step": 2984
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0200073719024658,
      "learning_rate": 9.350837885380375e-06,
      "loss": 0.5788,
      "step": 2985
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0267620086669922,
      "learning_rate": 9.350332229333276e-06,
      "loss": 0.55,
      "step": 2986
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9330299496650696,
      "learning_rate": 9.349826390108823e-06,
      "loss": 0.5397,
      "step": 2987
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.913159191608429,
      "learning_rate": 9.349320367728312e-06,
      "loss": 0.5045,
      "step": 2988
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0245423316955566,
      "learning_rate": 9.34881416221305e-06,
      "loss": 0.5595,
      "step": 2989
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9669381976127625,
      "learning_rate": 9.348307773584351e-06,
      "loss": 0.5488,
      "step": 2990
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0299818515777588,
      "learning_rate": 9.34780120186354e-06,
      "loss": 0.5729,
      "step": 2991
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9824283719062805,
      "learning_rate": 9.347294447071945e-06,
      "loss": 0.5487,
      "step": 2992
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0039525032043457,
      "learning_rate": 9.346787509230903e-06,
      "loss": 0.5602,
      "step": 2993
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9912124276161194,
      "learning_rate": 9.346280388361761e-06,
      "loss": 0.5508,
      "step": 2994
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.016176462173462,
      "learning_rate": 9.34577308448587e-06,
      "loss": 0.5502,
      "step": 2995
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0755479335784912,
      "learning_rate": 9.345265597624595e-06,
      "loss": 0.5636,
      "step": 2996
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.038004994392395,
      "learning_rate": 9.344757927799299e-06,
      "loss": 0.5851,
      "step": 2997
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0409660339355469,
      "learning_rate": 9.344250075031362e-06,
      "loss": 0.5517,
      "step": 2998
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9538517594337463,
      "learning_rate": 9.343742039342168e-06,
      "loss": 0.5182,
      "step": 2999
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.941325306892395,
      "learning_rate": 9.343233820753107e-06,
      "loss": 0.5663,
      "step": 3000
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.998930811882019,
      "learning_rate": 9.34272541928558e-06,
      "loss": 0.5677,
      "step": 3001
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9965593814849854,
      "learning_rate": 9.34221683496099e-06,
      "loss": 0.5179,
      "step": 3002
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0460489988327026,
      "learning_rate": 9.341708067800757e-06,
      "loss": 0.55,
      "step": 3003
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0359423160552979,
      "learning_rate": 9.341199117826298e-06,
      "loss": 0.5749,
      "step": 3004
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9712229371070862,
      "learning_rate": 9.340689985059048e-06,
      "loss": 0.5765,
      "step": 3005
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0309653282165527,
      "learning_rate": 9.340180669520443e-06,
      "loss": 0.5637,
      "step": 3006
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0827995538711548,
      "learning_rate": 9.339671171231929e-06,
      "loss": 0.5617,
      "step": 3007
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9527615904808044,
      "learning_rate": 9.339161490214957e-06,
      "loss": 0.5781,
      "step": 3008
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.983928918838501,
      "learning_rate": 9.33865162649099e-06,
      "loss": 0.534,
      "step": 3009
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9210485816001892,
      "learning_rate": 9.338141580081496e-06,
      "loss": 0.509,
      "step": 3010
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.131502389907837,
      "learning_rate": 9.337631351007953e-06,
      "loss": 0.523,
      "step": 3011
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0279563665390015,
      "learning_rate": 9.337120939291842e-06,
      "loss": 0.5423,
      "step": 3012
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0946483612060547,
      "learning_rate": 9.336610344954656e-06,
      "loss": 0.5313,
      "step": 3013
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0433611869812012,
      "learning_rate": 9.336099568017895e-06,
      "loss": 0.5194,
      "step": 3014
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0965982675552368,
      "learning_rate": 9.335588608503065e-06,
      "loss": 0.5643,
      "step": 3015
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0462230443954468,
      "learning_rate": 9.33507746643168e-06,
      "loss": 0.545,
      "step": 3016
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0247931480407715,
      "learning_rate": 9.334566141825266e-06,
      "loss": 0.5885,
      "step": 3017
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9968416094779968,
      "learning_rate": 9.334054634705347e-06,
      "loss": 0.5317,
      "step": 3018
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.109590768814087,
      "learning_rate": 9.333542945093468e-06,
      "loss": 0.5622,
      "step": 3019
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0385175943374634,
      "learning_rate": 9.333031073011169e-06,
      "loss": 0.5883,
      "step": 3020
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0782296657562256,
      "learning_rate": 9.332519018480005e-06,
      "loss": 0.6121,
      "step": 3021
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1267576217651367,
      "learning_rate": 9.332006781521537e-06,
      "loss": 0.5382,
      "step": 3022
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0112955570220947,
      "learning_rate": 9.331494362157335e-06,
      "loss": 0.5186,
      "step": 3023
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0713708400726318,
      "learning_rate": 9.330981760408972e-06,
      "loss": 0.55,
      "step": 3024
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.981164813041687,
      "learning_rate": 9.330468976298033e-06,
      "loss": 0.5468,
      "step": 3025
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0624793767929077,
      "learning_rate": 9.329956009846111e-06,
      "loss": 0.573,
      "step": 3026
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0366548299789429,
      "learning_rate": 9.329442861074803e-06,
      "loss": 0.5816,
      "step": 3027
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0300683975219727,
      "learning_rate": 9.328929530005717e-06,
      "loss": 0.565,
      "step": 3028
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0933263301849365,
      "learning_rate": 9.328416016660471e-06,
      "loss": 0.5633,
      "step": 3029
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.003050684928894,
      "learning_rate": 9.327902321060681e-06,
      "loss": 0.5039,
      "step": 3030
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0441280603408813,
      "learning_rate": 9.327388443227981e-06,
      "loss": 0.5935,
      "step": 3031
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.143961787223816,
      "learning_rate": 9.326874383184006e-06,
      "loss": 0.5781,
      "step": 3032
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0665271282196045,
      "learning_rate": 9.326360140950406e-06,
      "loss": 0.5428,
      "step": 3033
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0226956605911255,
      "learning_rate": 9.325845716548827e-06,
      "loss": 0.5855,
      "step": 3034
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1042506694793701,
      "learning_rate": 9.325331110000937e-06,
      "loss": 0.5883,
      "step": 3035
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0352132320404053,
      "learning_rate": 9.324816321328398e-06,
      "loss": 0.5992,
      "step": 3036
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9326235055923462,
      "learning_rate": 9.324301350552889e-06,
      "loss": 0.566,
      "step": 3037
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.04964280128479,
      "learning_rate": 9.323786197696094e-06,
      "loss": 0.5552,
      "step": 3038
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1173388957977295,
      "learning_rate": 9.323270862779704e-06,
      "loss": 0.5909,
      "step": 3039
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0002803802490234,
      "learning_rate": 9.322755345825418e-06,
      "loss": 0.4963,
      "step": 3040
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9863665699958801,
      "learning_rate": 9.32223964685494e-06,
      "loss": 0.5456,
      "step": 3041
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0843548774719238,
      "learning_rate": 9.321723765889987e-06,
      "loss": 0.5545,
      "step": 3042
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0747976303100586,
      "learning_rate": 9.321207702952281e-06,
      "loss": 0.5945,
      "step": 3043
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2692376375198364,
      "learning_rate": 9.320691458063552e-06,
      "loss": 0.5809,
      "step": 3044
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9482994079589844,
      "learning_rate": 9.320175031245535e-06,
      "loss": 0.5497,
      "step": 3045
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.094792366027832,
      "learning_rate": 9.319658422519977e-06,
      "loss": 0.6102,
      "step": 3046
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2238280773162842,
      "learning_rate": 9.319141631908628e-06,
      "loss": 0.6002,
      "step": 3047
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9731894135475159,
      "learning_rate": 9.318624659433254e-06,
      "loss": 0.5756,
      "step": 3048
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0275579690933228,
      "learning_rate": 9.318107505115615e-06,
      "loss": 0.5615,
      "step": 3049
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0008189678192139,
      "learning_rate": 9.317590168977492e-06,
      "loss": 0.5839,
      "step": 3050
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0680350065231323,
      "learning_rate": 9.317072651040666e-06,
      "loss": 0.5497,
      "step": 3051
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.161730170249939,
      "learning_rate": 9.31655495132693e-06,
      "loss": 0.5489,
      "step": 3052
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0339173078536987,
      "learning_rate": 9.31603706985808e-06,
      "loss": 0.5424,
      "step": 3053
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.064933180809021,
      "learning_rate": 9.315519006655925e-06,
      "loss": 0.6057,
      "step": 3054
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0022574663162231,
      "learning_rate": 9.315000761742276e-06,
      "loss": 0.5506,
      "step": 3055
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0306645631790161,
      "learning_rate": 9.314482335138954e-06,
      "loss": 0.5416,
      "step": 3056
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0025889873504639,
      "learning_rate": 9.313963726867793e-06,
      "loss": 0.5841,
      "step": 3057
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0571597814559937,
      "learning_rate": 9.313444936950626e-06,
      "loss": 0.5199,
      "step": 3058
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.082176923751831,
      "learning_rate": 9.312925965409297e-06,
      "loss": 0.5846,
      "step": 3059
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1681592464447021,
      "learning_rate": 9.312406812265659e-06,
      "loss": 0.5624,
      "step": 3060
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0278209447860718,
      "learning_rate": 9.311887477541574e-06,
      "loss": 0.553,
      "step": 3061
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9465845227241516,
      "learning_rate": 9.311367961258906e-06,
      "loss": 0.5733,
      "step": 3062
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9811533093452454,
      "learning_rate": 9.31084826343953e-06,
      "loss": 0.541,
      "step": 3063
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.119778037071228,
      "learning_rate": 9.310328384105331e-06,
      "loss": 0.5966,
      "step": 3064
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0411384105682373,
      "learning_rate": 9.309808323278199e-06,
      "loss": 0.5322,
      "step": 3065
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9605517387390137,
      "learning_rate": 9.30928808098003e-06,
      "loss": 0.5559,
      "step": 3066
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.074902057647705,
      "learning_rate": 9.308767657232733e-06,
      "loss": 0.6172,
      "step": 3067
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.038602590560913,
      "learning_rate": 9.308247052058217e-06,
      "loss": 0.5566,
      "step": 3068
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.026792287826538,
      "learning_rate": 9.307726265478405e-06,
      "loss": 0.5894,
      "step": 3069
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0388686656951904,
      "learning_rate": 9.307205297515225e-06,
      "loss": 0.6314,
      "step": 3070
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.082331657409668,
      "learning_rate": 9.306684148190616e-06,
      "loss": 0.607,
      "step": 3071
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.00520920753479,
      "learning_rate": 9.306162817526519e-06,
      "loss": 0.6214,
      "step": 3072
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9240366220474243,
      "learning_rate": 9.305641305544884e-06,
      "loss": 0.5031,
      "step": 3073
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9548230171203613,
      "learning_rate": 9.305119612267673e-06,
      "loss": 0.555,
      "step": 3074
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0587652921676636,
      "learning_rate": 9.30459773771685e-06,
      "loss": 0.588,
      "step": 3075
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0245496034622192,
      "learning_rate": 9.304075681914392e-06,
      "loss": 0.5027,
      "step": 3076
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0220906734466553,
      "learning_rate": 9.30355344488228e-06,
      "loss": 0.5349,
      "step": 3077
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0830450057983398,
      "learning_rate": 9.303031026642504e-06,
      "loss": 0.5247,
      "step": 3078
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9544553756713867,
      "learning_rate": 9.302508427217059e-06,
      "loss": 0.5561,
      "step": 3079
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0785131454467773,
      "learning_rate": 9.301985646627953e-06,
      "loss": 0.587,
      "step": 3080
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9493861198425293,
      "learning_rate": 9.301462684897195e-06,
      "loss": 0.5444,
      "step": 3081
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9934033751487732,
      "learning_rate": 9.300939542046808e-06,
      "loss": 0.5378,
      "step": 3082
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0883805751800537,
      "learning_rate": 9.30041621809882e-06,
      "loss": 0.5908,
      "step": 3083
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0969899892807007,
      "learning_rate": 9.299892713075263e-06,
      "loss": 0.6121,
      "step": 3084
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0503705739974976,
      "learning_rate": 9.299369026998184e-06,
      "loss": 0.5734,
      "step": 3085
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.048270344734192,
      "learning_rate": 9.298845159889632e-06,
      "loss": 0.5618,
      "step": 3086
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0619757175445557,
      "learning_rate": 9.298321111771664e-06,
      "loss": 0.5845,
      "step": 3087
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0658220052719116,
      "learning_rate": 9.297796882666346e-06,
      "loss": 0.5824,
      "step": 3088
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0340075492858887,
      "learning_rate": 9.297272472595753e-06,
      "loss": 0.5979,
      "step": 3089
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0165023803710938,
      "learning_rate": 9.296747881581965e-06,
      "loss": 0.56,
      "step": 3090
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9769580364227295,
      "learning_rate": 9.29622310964707e-06,
      "loss": 0.559,
      "step": 3091
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1017225980758667,
      "learning_rate": 9.295698156813167e-06,
      "loss": 0.5867,
      "step": 3092
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.227968454360962,
      "learning_rate": 9.295173023102358e-06,
      "loss": 0.5888,
      "step": 3093
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.072351336479187,
      "learning_rate": 9.294647708536754e-06,
      "loss": 0.5908,
      "step": 3094
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.027206540107727,
      "learning_rate": 9.294122213138475e-06,
      "loss": 0.5276,
      "step": 3095
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1481643915176392,
      "learning_rate": 9.29359653692965e-06,
      "loss": 0.5594,
      "step": 3096
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9869410991668701,
      "learning_rate": 9.293070679932407e-06,
      "loss": 0.6166,
      "step": 3097
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.051023244857788,
      "learning_rate": 9.292544642168896e-06,
      "loss": 0.523,
      "step": 3098
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0125713348388672,
      "learning_rate": 9.292018423661261e-06,
      "loss": 0.5428,
      "step": 3099
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1149213314056396,
      "learning_rate": 9.291492024431661e-06,
      "loss": 0.5854,
      "step": 3100
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1189168691635132,
      "learning_rate": 9.290965444502263e-06,
      "loss": 0.6149,
      "step": 3101
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9649505019187927,
      "learning_rate": 9.290438683895236e-06,
      "loss": 0.5618,
      "step": 3102
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0438565015792847,
      "learning_rate": 9.28991174263276e-06,
      "loss": 0.5917,
      "step": 3103
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2185555696487427,
      "learning_rate": 9.289384620737025e-06,
      "loss": 0.6123,
      "step": 3104
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0851901769638062,
      "learning_rate": 9.288857318230225e-06,
      "loss": 0.5576,
      "step": 3105
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0335770845413208,
      "learning_rate": 9.288329835134563e-06,
      "loss": 0.6028,
      "step": 3106
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0511239767074585,
      "learning_rate": 9.28780217147225e-06,
      "loss": 0.5304,
      "step": 3107
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0801007747650146,
      "learning_rate": 9.287274327265505e-06,
      "loss": 0.5758,
      "step": 3108
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0751490592956543,
      "learning_rate": 9.286746302536551e-06,
      "loss": 0.5577,
      "step": 3109
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0887144804000854,
      "learning_rate": 9.286218097307623e-06,
      "loss": 0.5569,
      "step": 3110
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0135072469711304,
      "learning_rate": 9.285689711600961e-06,
      "loss": 0.5706,
      "step": 3111
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0304912328720093,
      "learning_rate": 9.285161145438815e-06,
      "loss": 0.5775,
      "step": 3112
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9967204332351685,
      "learning_rate": 9.284632398843439e-06,
      "loss": 0.5224,
      "step": 3113
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1232625246047974,
      "learning_rate": 9.284103471837097e-06,
      "loss": 0.577,
      "step": 3114
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0335229635238647,
      "learning_rate": 9.283574364442066e-06,
      "loss": 0.5093,
      "step": 3115
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.074714183807373,
      "learning_rate": 9.283045076680614e-06,
      "loss": 0.6136,
      "step": 3116
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0689712762832642,
      "learning_rate": 9.282515608575038e-06,
      "loss": 0.5831,
      "step": 3117
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9729591608047485,
      "learning_rate": 9.281985960147625e-06,
      "loss": 0.5608,
      "step": 3118
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0159952640533447,
      "learning_rate": 9.28145613142068e-06,
      "loss": 0.5936,
      "step": 3119
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0457758903503418,
      "learning_rate": 9.28092612241651e-06,
      "loss": 0.5497,
      "step": 3120
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0730384588241577,
      "learning_rate": 9.280395933157436e-06,
      "loss": 0.5271,
      "step": 3121
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9636267423629761,
      "learning_rate": 9.279865563665778e-06,
      "loss": 0.5305,
      "step": 3122
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9577395915985107,
      "learning_rate": 9.27933501396387e-06,
      "loss": 0.5282,
      "step": 3123
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9931013584136963,
      "learning_rate": 9.27880428407405e-06,
      "loss": 0.578,
      "step": 3124
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9747065305709839,
      "learning_rate": 9.278273374018669e-06,
      "loss": 0.5649,
      "step": 3125
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.026253342628479,
      "learning_rate": 9.277742283820077e-06,
      "loss": 0.5885,
      "step": 3126
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0813730955123901,
      "learning_rate": 9.27721101350064e-06,
      "loss": 0.598,
      "step": 3127
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0212422609329224,
      "learning_rate": 9.276679563082726e-06,
      "loss": 0.5821,
      "step": 3128
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9639314413070679,
      "learning_rate": 9.276147932588712e-06,
      "loss": 0.5182,
      "step": 3129
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0123671293258667,
      "learning_rate": 9.275616122040985e-06,
      "loss": 0.5395,
      "step": 3130
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0312772989273071,
      "learning_rate": 9.275084131461938e-06,
      "loss": 0.5851,
      "step": 3131
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.109800100326538,
      "learning_rate": 9.27455196087397e-06,
      "loss": 0.5442,
      "step": 3132
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0117939710617065,
      "learning_rate": 9.274019610299487e-06,
      "loss": 0.5628,
      "step": 3133
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.981764018535614,
      "learning_rate": 9.273487079760908e-06,
      "loss": 0.5369,
      "step": 3134
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0255862474441528,
      "learning_rate": 9.272954369280654e-06,
      "loss": 0.5556,
      "step": 3135
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9613683819770813,
      "learning_rate": 9.272421478881158e-06,
      "loss": 0.5672,
      "step": 3136
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2361464500427246,
      "learning_rate": 9.271888408584852e-06,
      "loss": 0.5917,
      "step": 3137
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9689004421234131,
      "learning_rate": 9.27135515841419e-06,
      "loss": 0.5511,
      "step": 3138
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0080313682556152,
      "learning_rate": 9.27082172839162e-06,
      "loss": 0.548,
      "step": 3139
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0074330568313599,
      "learning_rate": 9.270288118539603e-06,
      "loss": 0.5812,
      "step": 3140
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0355607271194458,
      "learning_rate": 9.26975432888061e-06,
      "loss": 0.5253,
      "step": 3141
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9730377793312073,
      "learning_rate": 9.269220359437114e-06,
      "loss": 0.5238,
      "step": 3142
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0539954900741577,
      "learning_rate": 9.2686862102316e-06,
      "loss": 0.589,
      "step": 3143
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0077118873596191,
      "learning_rate": 9.268151881286561e-06,
      "loss": 0.5713,
      "step": 3144
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9483577013015747,
      "learning_rate": 9.267617372624494e-06,
      "loss": 0.5718,
      "step": 3145
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.081387996673584,
      "learning_rate": 9.267082684267905e-06,
      "loss": 0.5745,
      "step": 3146
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1054128408432007,
      "learning_rate": 9.266547816239309e-06,
      "loss": 0.602,
      "step": 3147
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0518232583999634,
      "learning_rate": 9.266012768561225e-06,
      "loss": 0.62,
      "step": 3148
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9769362211227417,
      "learning_rate": 9.265477541256184e-06,
      "loss": 0.5942,
      "step": 3149
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9312898516654968,
      "learning_rate": 9.264942134346723e-06,
      "loss": 0.5428,
      "step": 3150
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9657699465751648,
      "learning_rate": 9.264406547855386e-06,
      "loss": 0.5389,
      "step": 3151
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0395054817199707,
      "learning_rate": 9.263870781804723e-06,
      "loss": 0.5474,
      "step": 3152
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0954946279525757,
      "learning_rate": 9.263334836217295e-06,
      "loss": 0.5394,
      "step": 3153
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9262494444847107,
      "learning_rate": 9.262798711115667e-06,
      "loss": 0.5598,
      "step": 3154
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9422882199287415,
      "learning_rate": 9.262262406522415e-06,
      "loss": 0.4883,
      "step": 3155
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.115454912185669,
      "learning_rate": 9.261725922460121e-06,
      "loss": 0.5713,
      "step": 3156
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1434146165847778,
      "learning_rate": 9.261189258951372e-06,
      "loss": 0.5763,
      "step": 3157
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9961076974868774,
      "learning_rate": 9.26065241601877e-06,
      "loss": 0.6126,
      "step": 3158
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9965333938598633,
      "learning_rate": 9.260115393684914e-06,
      "loss": 0.5771,
      "step": 3159
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9651138186454773,
      "learning_rate": 9.25957819197242e-06,
      "loss": 0.5508,
      "step": 3160
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0352110862731934,
      "learning_rate": 9.259040810903906e-06,
      "loss": 0.5425,
      "step": 3161
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0079376697540283,
      "learning_rate": 9.258503250501998e-06,
      "loss": 0.5814,
      "step": 3162
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0181200504302979,
      "learning_rate": 9.257965510789334e-06,
      "loss": 0.5294,
      "step": 3163
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0838521718978882,
      "learning_rate": 9.257427591788555e-06,
      "loss": 0.6009,
      "step": 3164
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9461649060249329,
      "learning_rate": 9.25688949352231e-06,
      "loss": 0.5694,
      "step": 3165
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8991617560386658,
      "learning_rate": 9.256351216013257e-06,
      "loss": 0.5699,
      "step": 3166
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9943413138389587,
      "learning_rate": 9.255812759284062e-06,
      "loss": 0.5118,
      "step": 3167
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.92888343334198,
      "learning_rate": 9.255274123357396e-06,
      "loss": 0.5091,
      "step": 3168
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0817584991455078,
      "learning_rate": 9.254735308255937e-06,
      "loss": 0.6395,
      "step": 3169
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0437732934951782,
      "learning_rate": 9.254196314002379e-06,
      "loss": 0.5639,
      "step": 3170
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1044567823410034,
      "learning_rate": 9.253657140619412e-06,
      "loss": 0.5903,
      "step": 3171
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1140618324279785,
      "learning_rate": 9.25311778812974e-06,
      "loss": 0.5645,
      "step": 3172
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1418890953063965,
      "learning_rate": 9.252578256556075e-06,
      "loss": 0.6061,
      "step": 3173
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.003599762916565,
      "learning_rate": 9.252038545921131e-06,
      "loss": 0.5037,
      "step": 3174
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2493435144424438,
      "learning_rate": 9.251498656247636e-06,
      "loss": 0.602,
      "step": 3175
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0258667469024658,
      "learning_rate": 9.250958587558326e-06,
      "loss": 0.5372,
      "step": 3176
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0604716539382935,
      "learning_rate": 9.250418339875934e-06,
      "loss": 0.5791,
      "step": 3177
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.565056562423706,
      "learning_rate": 9.249877913223213e-06,
      "loss": 0.5609,
      "step": 3178
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0567054748535156,
      "learning_rate": 9.249337307622916e-06,
      "loss": 0.5614,
      "step": 3179
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9943719506263733,
      "learning_rate": 9.24879652309781e-06,
      "loss": 0.576,
      "step": 3180
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.998900294303894,
      "learning_rate": 9.248255559670661e-06,
      "loss": 0.5364,
      "step": 3181
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9154987335205078,
      "learning_rate": 9.247714417364251e-06,
      "loss": 0.5405,
      "step": 3182
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9305417537689209,
      "learning_rate": 9.24717309620136e-06,
      "loss": 0.5161,
      "step": 3183
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.028682827949524,
      "learning_rate": 9.246631596204788e-06,
      "loss": 0.5365,
      "step": 3184
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.061288595199585,
      "learning_rate": 9.246089917397332e-06,
      "loss": 0.5778,
      "step": 3185
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.063146710395813,
      "learning_rate": 9.2455480598018e-06,
      "loss": 0.6103,
      "step": 3186
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1085253953933716,
      "learning_rate": 9.245006023441008e-06,
      "loss": 0.5487,
      "step": 3187
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0992125272750854,
      "learning_rate": 9.24446380833778e-06,
      "loss": 0.6294,
      "step": 3188
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9414675831794739,
      "learning_rate": 9.243921414514947e-06,
      "loss": 0.5078,
      "step": 3189
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9629780650138855,
      "learning_rate": 9.243378841995346e-06,
      "loss": 0.5164,
      "step": 3190
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.040055513381958,
      "learning_rate": 9.242836090801823e-06,
      "loss": 0.5511,
      "step": 3191
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8983215689659119,
      "learning_rate": 9.242293160957231e-06,
      "loss": 0.5451,
      "step": 3192
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0140423774719238,
      "learning_rate": 9.241750052484435e-06,
      "loss": 0.5477,
      "step": 3193
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0713225603103638,
      "learning_rate": 9.241206765406298e-06,
      "loss": 0.5152,
      "step": 3194
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0715563297271729,
      "learning_rate": 9.2406632997457e-06,
      "loss": 0.5458,
      "step": 3195
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0324233770370483,
      "learning_rate": 9.240119655525522e-06,
      "loss": 0.5821,
      "step": 3196
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0402965545654297,
      "learning_rate": 9.239575832768655e-06,
      "loss": 0.6,
      "step": 3197
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9679100513458252,
      "learning_rate": 9.239031831498e-06,
      "loss": 0.5253,
      "step": 3198
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.125067114830017,
      "learning_rate": 9.238487651736458e-06,
      "loss": 0.5982,
      "step": 3199
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.075700283050537,
      "learning_rate": 9.237943293506948e-06,
      "loss": 0.5908,
      "step": 3200
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9718443751335144,
      "learning_rate": 9.237398756832387e-06,
      "loss": 0.5514,
      "step": 3201
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0474004745483398,
      "learning_rate": 9.236854041735706e-06,
      "loss": 0.5362,
      "step": 3202
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0741773843765259,
      "learning_rate": 9.236309148239839e-06,
      "loss": 0.5573,
      "step": 3203
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0043758153915405,
      "learning_rate": 9.235764076367732e-06,
      "loss": 0.5263,
      "step": 3204
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0209715366363525,
      "learning_rate": 9.235218826142337e-06,
      "loss": 0.5497,
      "step": 3205
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9901759028434753,
      "learning_rate": 9.234673397586606e-06,
      "loss": 0.5489,
      "step": 3206
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.522571563720703,
      "learning_rate": 9.234127790723512e-06,
      "loss": 0.5618,
      "step": 3207
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.058837413787842,
      "learning_rate": 9.233582005576028e-06,
      "loss": 0.5912,
      "step": 3208
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.545076370239258,
      "learning_rate": 9.233036042167131e-06,
      "loss": 0.6748,
      "step": 3209
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.660017967224121,
      "learning_rate": 9.232489900519812e-06,
      "loss": 0.6371,
      "step": 3210
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4097462892532349,
      "learning_rate": 9.231943580657069e-06,
      "loss": 0.6299,
      "step": 3211
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1483733654022217,
      "learning_rate": 9.2313970826019e-06,
      "loss": 0.5256,
      "step": 3212
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.201183557510376,
      "learning_rate": 9.230850406377323e-06,
      "loss": 0.5777,
      "step": 3213
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2220267057418823,
      "learning_rate": 9.230303552006352e-06,
      "loss": 0.589,
      "step": 3214
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1205120086669922,
      "learning_rate": 9.229756519512014e-06,
      "loss": 0.5937,
      "step": 3215
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0527418851852417,
      "learning_rate": 9.229209308917343e-06,
      "loss": 0.5534,
      "step": 3216
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3884944915771484,
      "learning_rate": 9.228661920245383e-06,
      "loss": 0.5374,
      "step": 3217
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0388891696929932,
      "learning_rate": 9.22811435351918e-06,
      "loss": 0.5414,
      "step": 3218
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.263106107711792,
      "learning_rate": 9.227566608761786e-06,
      "loss": 0.6034,
      "step": 3219
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2488805055618286,
      "learning_rate": 9.227018685996272e-06,
      "loss": 0.5911,
      "step": 3220
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.839228391647339,
      "learning_rate": 9.226470585245706e-06,
      "loss": 0.599,
      "step": 3221
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4514201879501343,
      "learning_rate": 9.225922306533164e-06,
      "loss": 0.6333,
      "step": 3222
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2319753170013428,
      "learning_rate": 9.225373849881739e-06,
      "loss": 0.5535,
      "step": 3223
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9874230027198792,
      "learning_rate": 9.224825215314515e-06,
      "loss": 0.56,
      "step": 3224
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7198013067245483,
      "learning_rate": 9.224276402854601e-06,
      "loss": 0.5781,
      "step": 3225
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7840505838394165,
      "learning_rate": 9.223727412525103e-06,
      "loss": 0.5775,
      "step": 3226
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.298826813697815,
      "learning_rate": 9.223178244349135e-06,
      "loss": 0.5766,
      "step": 3227
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0350885391235352,
      "learning_rate": 9.222628898349825e-06,
      "loss": 0.5495,
      "step": 3228
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9768742918968201,
      "learning_rate": 9.2220793745503e-06,
      "loss": 0.5383,
      "step": 3229
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9744670987129211,
      "learning_rate": 9.221529672973701e-06,
      "loss": 0.5271,
      "step": 3230
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1338422298431396,
      "learning_rate": 9.220979793643173e-06,
      "loss": 0.5691,
      "step": 3231
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0255298614501953,
      "learning_rate": 9.220429736581869e-06,
      "loss": 0.5825,
      "step": 3232
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0402343273162842,
      "learning_rate": 9.219879501812952e-06,
      "loss": 0.5294,
      "step": 3233
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.087755560874939,
      "learning_rate": 9.219329089359588e-06,
      "loss": 0.5842,
      "step": 3234
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.099476933479309,
      "learning_rate": 9.218778499244953e-06,
      "loss": 0.6201,
      "step": 3235
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0949242115020752,
      "learning_rate": 9.218227731492234e-06,
      "loss": 0.6195,
      "step": 3236
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0557842254638672,
      "learning_rate": 9.217676786124616e-06,
      "loss": 0.5638,
      "step": 3237
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1988122463226318,
      "learning_rate": 9.217125663165303e-06,
      "loss": 0.5939,
      "step": 3238
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1216857433319092,
      "learning_rate": 9.216574362637498e-06,
      "loss": 0.6367,
      "step": 3239
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9827113151550293,
      "learning_rate": 9.216022884564414e-06,
      "loss": 0.5675,
      "step": 3240
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0841467380523682,
      "learning_rate": 9.215471228969275e-06,
      "loss": 0.574,
      "step": 3241
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0059484243392944,
      "learning_rate": 9.214919395875306e-06,
      "loss": 0.61,
      "step": 3242
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9875774383544922,
      "learning_rate": 9.214367385305744e-06,
      "loss": 0.4961,
      "step": 3243
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0952807664871216,
      "learning_rate": 9.213815197283834e-06,
      "loss": 0.5547,
      "step": 3244
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1362330913543701,
      "learning_rate": 9.21326283183282e-06,
      "loss": 0.5441,
      "step": 3245
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0097522735595703,
      "learning_rate": 9.21271028897597e-06,
      "loss": 0.5455,
      "step": 3246
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0766572952270508,
      "learning_rate": 9.212157568736542e-06,
      "loss": 0.5419,
      "step": 3247
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9383859038352966,
      "learning_rate": 9.211604671137812e-06,
      "loss": 0.5496,
      "step": 3248
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0516395568847656,
      "learning_rate": 9.211051596203061e-06,
      "loss": 0.6164,
      "step": 3249
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0035028457641602,
      "learning_rate": 9.210498343955576e-06,
      "loss": 0.5573,
      "step": 3250
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.065867304801941,
      "learning_rate": 9.209944914418653e-06,
      "loss": 0.6218,
      "step": 3251
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0126872062683105,
      "learning_rate": 9.209391307615596e-06,
      "loss": 0.5129,
      "step": 3252
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.238006591796875,
      "learning_rate": 9.208837523569713e-06,
      "loss": 0.5923,
      "step": 3253
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0148457288742065,
      "learning_rate": 9.208283562304326e-06,
      "loss": 0.5192,
      "step": 3254
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1629726886749268,
      "learning_rate": 9.207729423842755e-06,
      "loss": 0.5576,
      "step": 3255
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.10166335105896,
      "learning_rate": 9.207175108208334e-06,
      "loss": 0.5832,
      "step": 3256
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0380483865737915,
      "learning_rate": 9.20662061542441e-06,
      "loss": 0.5944,
      "step": 3257
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0519050359725952,
      "learning_rate": 9.206065945514321e-06,
      "loss": 0.5601,
      "step": 3258
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.071286678314209,
      "learning_rate": 9.20551109850143e-06,
      "loss": 0.5497,
      "step": 3259
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0530121326446533,
      "learning_rate": 9.204956074409095e-06,
      "loss": 0.5615,
      "step": 3260
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0857264995574951,
      "learning_rate": 9.204400873260688e-06,
      "loss": 0.5625,
      "step": 3261
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9711153507232666,
      "learning_rate": 9.203845495079587e-06,
      "loss": 0.5684,
      "step": 3262
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9729705452919006,
      "learning_rate": 9.203289939889175e-06,
      "loss": 0.5455,
      "step": 3263
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9644930958747864,
      "learning_rate": 9.202734207712847e-06,
      "loss": 0.55,
      "step": 3264
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0248082876205444,
      "learning_rate": 9.202178298574e-06,
      "loss": 0.5271,
      "step": 3265
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0923707485198975,
      "learning_rate": 9.201622212496043e-06,
      "loss": 0.5625,
      "step": 3266
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.037908673286438,
      "learning_rate": 9.201065949502394e-06,
      "loss": 0.6094,
      "step": 3267
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0656263828277588,
      "learning_rate": 9.20050950961647e-06,
      "loss": 0.5385,
      "step": 3268
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9770526885986328,
      "learning_rate": 9.199952892861706e-06,
      "loss": 0.5493,
      "step": 3269
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1895619630813599,
      "learning_rate": 9.199396099261532e-06,
      "loss": 0.5696,
      "step": 3270
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0683916807174683,
      "learning_rate": 9.198839128839399e-06,
      "loss": 0.5782,
      "step": 3271
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0089161396026611,
      "learning_rate": 9.198281981618757e-06,
      "loss": 0.5446,
      "step": 3272
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0737895965576172,
      "learning_rate": 9.197724657623066e-06,
      "loss": 0.5417,
      "step": 3273
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0453969240188599,
      "learning_rate": 9.197167156875793e-06,
      "loss": 0.5761,
      "step": 3274
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1445963382720947,
      "learning_rate": 9.19660947940041e-06,
      "loss": 0.5683,
      "step": 3275
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9256425499916077,
      "learning_rate": 9.196051625220401e-06,
      "loss": 0.5287,
      "step": 3276
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9852249622344971,
      "learning_rate": 9.195493594359254e-06,
      "loss": 0.5401,
      "step": 3277
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0960800647735596,
      "learning_rate": 9.19493538684047e-06,
      "loss": 0.5993,
      "step": 3278
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9759271740913391,
      "learning_rate": 9.194377002687547e-06,
      "loss": 0.5518,
      "step": 3279
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0800795555114746,
      "learning_rate": 9.193818441924003e-06,
      "loss": 0.5577,
      "step": 3280
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0303105115890503,
      "learning_rate": 9.19325970457335e-06,
      "loss": 0.5192,
      "step": 3281
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.977532684803009,
      "learning_rate": 9.192700790659121e-06,
      "loss": 0.4839,
      "step": 3282
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.118839144706726,
      "learning_rate": 9.192141700204844e-06,
      "loss": 0.6311,
      "step": 3283
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0272563695907593,
      "learning_rate": 9.191582433234067e-06,
      "loss": 0.5859,
      "step": 3284
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9714250564575195,
      "learning_rate": 9.191022989770332e-06,
      "loss": 0.5556,
      "step": 3285
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1177606582641602,
      "learning_rate": 9.1904633698372e-06,
      "loss": 0.5937,
      "step": 3286
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9881129860877991,
      "learning_rate": 9.189903573458234e-06,
      "loss": 0.5656,
      "step": 3287
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.031219244003296,
      "learning_rate": 9.189343600657002e-06,
      "loss": 0.5645,
      "step": 3288
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0255357027053833,
      "learning_rate": 9.188783451457086e-06,
      "loss": 0.5764,
      "step": 3289
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.020890235900879,
      "learning_rate": 9.18822312588207e-06,
      "loss": 0.5423,
      "step": 3290
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9295765161514282,
      "learning_rate": 9.187662623955548e-06,
      "loss": 0.4904,
      "step": 3291
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1230452060699463,
      "learning_rate": 9.18710194570112e-06,
      "loss": 0.557,
      "step": 3292
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9984371066093445,
      "learning_rate": 9.186541091142397e-06,
      "loss": 0.4955,
      "step": 3293
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1656453609466553,
      "learning_rate": 9.18598006030299e-06,
      "loss": 0.6019,
      "step": 3294
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0436992645263672,
      "learning_rate": 9.185418853206528e-06,
      "loss": 0.5177,
      "step": 3295
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.035557508468628,
      "learning_rate": 9.184857469876635e-06,
      "loss": 0.5584,
      "step": 3296
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0965100526809692,
      "learning_rate": 9.184295910336953e-06,
      "loss": 0.558,
      "step": 3297
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0783100128173828,
      "learning_rate": 9.183734174611125e-06,
      "loss": 0.6263,
      "step": 3298
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.103131651878357,
      "learning_rate": 9.183172262722807e-06,
      "loss": 0.6347,
      "step": 3299
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0050467252731323,
      "learning_rate": 9.182610174695656e-06,
      "loss": 0.5465,
      "step": 3300
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.054924726486206,
      "learning_rate": 9.182047910553342e-06,
      "loss": 0.5739,
      "step": 3301
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.973746120929718,
      "learning_rate": 9.181485470319537e-06,
      "loss": 0.5741,
      "step": 3302
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0047938823699951,
      "learning_rate": 9.180922854017927e-06,
      "loss": 0.5378,
      "step": 3303
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0483931303024292,
      "learning_rate": 9.1803600616722e-06,
      "loss": 0.5843,
      "step": 3304
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1777387857437134,
      "learning_rate": 9.179797093306053e-06,
      "loss": 0.6227,
      "step": 3305
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0783026218414307,
      "learning_rate": 9.17923394894319e-06,
      "loss": 0.5663,
      "step": 3306
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9983446598052979,
      "learning_rate": 9.178670628607325e-06,
      "loss": 0.563,
      "step": 3307
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9895806312561035,
      "learning_rate": 9.178107132322174e-06,
      "loss": 0.5261,
      "step": 3308
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0566648244857788,
      "learning_rate": 9.177543460111469e-06,
      "loss": 0.5581,
      "step": 3309
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0616939067840576,
      "learning_rate": 9.17697961199894e-06,
      "loss": 0.5691,
      "step": 3310
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.002230167388916,
      "learning_rate": 9.176415588008332e-06,
      "loss": 0.5919,
      "step": 3311
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.095953345298767,
      "learning_rate": 9.175851388163391e-06,
      "loss": 0.5681,
      "step": 3312
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.043747067451477,
      "learning_rate": 9.175287012487874e-06,
      "loss": 0.5646,
      "step": 3313
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9811042547225952,
      "learning_rate": 9.174722461005546e-06,
      "loss": 0.5461,
      "step": 3314
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.125097632408142,
      "learning_rate": 9.174157733740178e-06,
      "loss": 0.5883,
      "step": 3315
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0411174297332764,
      "learning_rate": 9.173592830715548e-06,
      "loss": 0.5333,
      "step": 3316
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0620931386947632,
      "learning_rate": 9.173027751955444e-06,
      "loss": 0.5858,
      "step": 3317
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.960599422454834,
      "learning_rate": 9.172462497483658e-06,
      "loss": 0.4991,
      "step": 3318
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.133169412612915,
      "learning_rate": 9.17189706732399e-06,
      "loss": 0.5543,
      "step": 3319
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2463678121566772,
      "learning_rate": 9.171331461500253e-06,
      "loss": 0.596,
      "step": 3320
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0226194858551025,
      "learning_rate": 9.170765680036256e-06,
      "loss": 0.5472,
      "step": 3321
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.02968168258667,
      "learning_rate": 9.170199722955825e-06,
      "loss": 0.5627,
      "step": 3322
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9317070245742798,
      "learning_rate": 9.169633590282793e-06,
      "loss": 0.549,
      "step": 3323
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1055186986923218,
      "learning_rate": 9.169067282040994e-06,
      "loss": 0.579,
      "step": 3324
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.012923002243042,
      "learning_rate": 9.168500798254275e-06,
      "loss": 0.5193,
      "step": 3325
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0518181324005127,
      "learning_rate": 9.167934138946489e-06,
      "loss": 0.5923,
      "step": 3326
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1636402606964111,
      "learning_rate": 9.167367304141494e-06,
      "loss": 0.5886,
      "step": 3327
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1672029495239258,
      "learning_rate": 9.166800293863161e-06,
      "loss": 0.6024,
      "step": 3328
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.037725806236267,
      "learning_rate": 9.166233108135362e-06,
      "loss": 0.5351,
      "step": 3329
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.094976544380188,
      "learning_rate": 9.165665746981982e-06,
      "loss": 0.5662,
      "step": 3330
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0267653465270996,
      "learning_rate": 9.165098210426905e-06,
      "loss": 0.5907,
      "step": 3331
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0563652515411377,
      "learning_rate": 9.164530498494035e-06,
      "loss": 0.5794,
      "step": 3332
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0362412929534912,
      "learning_rate": 9.163962611207272e-06,
      "loss": 0.573,
      "step": 3333
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0745573043823242,
      "learning_rate": 9.163394548590529e-06,
      "loss": 0.5796,
      "step": 3334
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.067790150642395,
      "learning_rate": 9.162826310667725e-06,
      "loss": 0.5735,
      "step": 3335
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0531738996505737,
      "learning_rate": 9.162257897462784e-06,
      "loss": 0.5827,
      "step": 3336
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0059359073638916,
      "learning_rate": 9.161689308999646e-06,
      "loss": 0.5369,
      "step": 3337
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.201866865158081,
      "learning_rate": 9.161120545302246e-06,
      "loss": 0.5747,
      "step": 3338
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0412216186523438,
      "learning_rate": 9.160551606394537e-06,
      "loss": 0.6078,
      "step": 3339
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0173043012619019,
      "learning_rate": 9.159982492300473e-06,
      "loss": 0.5753,
      "step": 3340
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0637174844741821,
      "learning_rate": 9.159413203044017e-06,
      "loss": 0.5532,
      "step": 3341
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.042886734008789,
      "learning_rate": 9.158843738649141e-06,
      "loss": 0.5398,
      "step": 3342
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0245264768600464,
      "learning_rate": 9.158274099139823e-06,
      "loss": 0.542,
      "step": 3343
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9581206440925598,
      "learning_rate": 9.157704284540047e-06,
      "loss": 0.5205,
      "step": 3344
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0562598705291748,
      "learning_rate": 9.15713429487381e-06,
      "loss": 0.574,
      "step": 3345
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0625674724578857,
      "learning_rate": 9.156564130165106e-06,
      "loss": 0.5444,
      "step": 3346
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0383031368255615,
      "learning_rate": 9.155993790437949e-06,
      "loss": 0.5663,
      "step": 3347
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.969848096370697,
      "learning_rate": 9.155423275716351e-06,
      "loss": 0.5429,
      "step": 3348
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.017959713935852,
      "learning_rate": 9.154852586024332e-06,
      "loss": 0.5594,
      "step": 3349
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9905710816383362,
      "learning_rate": 9.154281721385928e-06,
      "loss": 0.5361,
      "step": 3350
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1172434091567993,
      "learning_rate": 9.153710681825169e-06,
      "loss": 0.5811,
      "step": 3351
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0339463949203491,
      "learning_rate": 9.153139467366103e-06,
      "loss": 0.5715,
      "step": 3352
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9711489081382751,
      "learning_rate": 9.152568078032783e-06,
      "loss": 0.579,
      "step": 3353
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.065482258796692,
      "learning_rate": 9.151996513849267e-06,
      "loss": 0.5532,
      "step": 3354
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8725598454475403,
      "learning_rate": 9.151424774839622e-06,
      "loss": 0.508,
      "step": 3355
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0380140542984009,
      "learning_rate": 9.15085286102792e-06,
      "loss": 0.5271,
      "step": 3356
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9468619227409363,
      "learning_rate": 9.150280772438245e-06,
      "loss": 0.5565,
      "step": 3357
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9295430779457092,
      "learning_rate": 9.149708509094684e-06,
      "loss": 0.5414,
      "step": 3358
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0121490955352783,
      "learning_rate": 9.149136071021333e-06,
      "loss": 0.5603,
      "step": 3359
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0209137201309204,
      "learning_rate": 9.148563458242296e-06,
      "loss": 0.5791,
      "step": 3360
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.954666018486023,
      "learning_rate": 9.147990670781683e-06,
      "loss": 0.5656,
      "step": 3361
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9398199319839478,
      "learning_rate": 9.147417708663615e-06,
      "loss": 0.4706,
      "step": 3362
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1586277484893799,
      "learning_rate": 9.146844571912213e-06,
      "loss": 0.561,
      "step": 3363
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.058401107788086,
      "learning_rate": 9.146271260551614e-06,
      "loss": 0.5768,
      "step": 3364
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0301518440246582,
      "learning_rate": 9.145697774605953e-06,
      "loss": 0.5582,
      "step": 3365
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0140354633331299,
      "learning_rate": 9.145124114099382e-06,
      "loss": 0.5525,
      "step": 3366
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1071993112564087,
      "learning_rate": 9.144550279056055e-06,
      "loss": 0.5199,
      "step": 3367
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0806834697723389,
      "learning_rate": 9.143976269500133e-06,
      "loss": 0.5089,
      "step": 3368
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0306366682052612,
      "learning_rate": 9.143402085455785e-06,
      "loss": 0.5495,
      "step": 3369
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9952037930488586,
      "learning_rate": 9.142827726947193e-06,
      "loss": 0.5691,
      "step": 3370
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0687406063079834,
      "learning_rate": 9.142253193998533e-06,
      "loss": 0.5451,
      "step": 3371
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9995650053024292,
      "learning_rate": 9.141678486634002e-06,
      "loss": 0.5679,
      "step": 3372
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9620756506919861,
      "learning_rate": 9.1411036048778e-06,
      "loss": 0.5671,
      "step": 3373
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0601290464401245,
      "learning_rate": 9.140528548754128e-06,
      "loss": 0.5516,
      "step": 3374
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9143772721290588,
      "learning_rate": 9.139953318287204e-06,
      "loss": 0.4883,
      "step": 3375
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1627060174942017,
      "learning_rate": 9.139377913501247e-06,
      "loss": 0.563,
      "step": 3376
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0104249715805054,
      "learning_rate": 9.138802334420486e-06,
      "loss": 0.5254,
      "step": 3377
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9357181191444397,
      "learning_rate": 9.138226581069158e-06,
      "loss": 0.5523,
      "step": 3378
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0746586322784424,
      "learning_rate": 9.137650653471505e-06,
      "loss": 0.6169,
      "step": 3379
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0098421573638916,
      "learning_rate": 9.137074551651774e-06,
      "loss": 0.5593,
      "step": 3380
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.048406720161438,
      "learning_rate": 9.136498275634226e-06,
      "loss": 0.5816,
      "step": 3381
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9700804948806763,
      "learning_rate": 9.135921825443125e-06,
      "loss": 0.58,
      "step": 3382
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0831280946731567,
      "learning_rate": 9.135345201102745e-06,
      "loss": 0.5414,
      "step": 3383
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.141015648841858,
      "learning_rate": 9.134768402637366e-06,
      "loss": 0.6002,
      "step": 3384
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9244864583015442,
      "learning_rate": 9.13419143007127e-06,
      "loss": 0.5431,
      "step": 3385
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9787561297416687,
      "learning_rate": 9.133614283428757e-06,
      "loss": 0.5202,
      "step": 3386
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0705188512802124,
      "learning_rate": 9.133036962734127e-06,
      "loss": 0.5641,
      "step": 3387
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9717293381690979,
      "learning_rate": 9.132459468011686e-06,
      "loss": 0.5757,
      "step": 3388
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0418078899383545,
      "learning_rate": 9.131881799285754e-06,
      "loss": 0.5758,
      "step": 3389
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.106235384941101,
      "learning_rate": 9.131303956580653e-06,
      "loss": 0.6181,
      "step": 3390
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9898834824562073,
      "learning_rate": 9.130725939920712e-06,
      "loss": 0.5413,
      "step": 3391
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0746660232543945,
      "learning_rate": 9.130147749330275e-06,
      "loss": 0.5748,
      "step": 3392
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9885543584823608,
      "learning_rate": 9.129569384833682e-06,
      "loss": 0.5702,
      "step": 3393
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0890642404556274,
      "learning_rate": 9.128990846455287e-06,
      "loss": 0.5958,
      "step": 3394
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0893391370773315,
      "learning_rate": 9.128412134219453e-06,
      "loss": 0.5275,
      "step": 3395
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0430580377578735,
      "learning_rate": 9.127833248150546e-06,
      "loss": 0.5993,
      "step": 3396
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.997788667678833,
      "learning_rate": 9.12725418827294e-06,
      "loss": 0.5298,
      "step": 3397
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9546692371368408,
      "learning_rate": 9.126674954611016e-06,
      "loss": 0.5516,
      "step": 3398
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0463684797286987,
      "learning_rate": 9.12609554718917e-06,
      "loss": 0.5576,
      "step": 3399
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9936715960502625,
      "learning_rate": 9.12551596603179e-06,
      "loss": 0.5558,
      "step": 3400
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0408931970596313,
      "learning_rate": 9.124936211163284e-06,
      "loss": 0.5593,
      "step": 3401
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0101039409637451,
      "learning_rate": 9.124356282608065e-06,
      "loss": 0.5265,
      "step": 3402
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1082333326339722,
      "learning_rate": 9.123776180390552e-06,
      "loss": 0.5835,
      "step": 3403
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0752975940704346,
      "learning_rate": 9.123195904535167e-06,
      "loss": 0.5165,
      "step": 3404
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0898516178131104,
      "learning_rate": 9.122615455066348e-06,
      "loss": 0.5818,
      "step": 3405
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1162127256393433,
      "learning_rate": 9.122034832008532e-06,
      "loss": 0.5463,
      "step": 3406
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0598623752593994,
      "learning_rate": 9.12145403538617e-06,
      "loss": 0.5618,
      "step": 3407
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0100927352905273,
      "learning_rate": 9.120873065223716e-06,
      "loss": 0.5602,
      "step": 3408
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0553600788116455,
      "learning_rate": 9.120291921545633e-06,
      "loss": 0.5251,
      "step": 3409
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0603712797164917,
      "learning_rate": 9.11971060437639e-06,
      "loss": 0.6051,
      "step": 3410
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.015459656715393,
      "learning_rate": 9.119129113740463e-06,
      "loss": 0.5398,
      "step": 3411
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9911149144172668,
      "learning_rate": 9.118547449662342e-06,
      "loss": 0.534,
      "step": 3412
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0005937814712524,
      "learning_rate": 9.117965612166514e-06,
      "loss": 0.5691,
      "step": 3413
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.947547435760498,
      "learning_rate": 9.117383601277478e-06,
      "loss": 0.5596,
      "step": 3414
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0150891542434692,
      "learning_rate": 9.116801417019744e-06,
      "loss": 0.5266,
      "step": 3415
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.959476113319397,
      "learning_rate": 9.116219059417821e-06,
      "loss": 0.5825,
      "step": 3416
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0116640329360962,
      "learning_rate": 9.115636528496236e-06,
      "loss": 0.5495,
      "step": 3417
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9909448027610779,
      "learning_rate": 9.115053824279511e-06,
      "loss": 0.5555,
      "step": 3418
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9822895526885986,
      "learning_rate": 9.114470946792187e-06,
      "loss": 0.5671,
      "step": 3419
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9034964442253113,
      "learning_rate": 9.113887896058805e-06,
      "loss": 0.5463,
      "step": 3420
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0540194511413574,
      "learning_rate": 9.11330467210391e-06,
      "loss": 0.5678,
      "step": 3421
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0437064170837402,
      "learning_rate": 9.11272127495207e-06,
      "loss": 0.5398,
      "step": 3422
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0747253894805908,
      "learning_rate": 9.112137704627842e-06,
      "loss": 0.6188,
      "step": 3423
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9856347441673279,
      "learning_rate": 9.1115539611558e-06,
      "loss": 0.544,
      "step": 3424
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0398938655853271,
      "learning_rate": 9.110970044560524e-06,
      "loss": 0.5679,
      "step": 3425
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0096843242645264,
      "learning_rate": 9.1103859548666e-06,
      "loss": 0.5964,
      "step": 3426
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0516611337661743,
      "learning_rate": 9.109801692098624e-06,
      "loss": 0.5566,
      "step": 3427
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9619464874267578,
      "learning_rate": 9.109217256281196e-06,
      "loss": 0.5298,
      "step": 3428
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.08042311668396,
      "learning_rate": 9.108632647438922e-06,
      "loss": 0.585,
      "step": 3429
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.047071933746338,
      "learning_rate": 9.108047865596421e-06,
      "loss": 0.5438,
      "step": 3430
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0152941942214966,
      "learning_rate": 9.107462910778316e-06,
      "loss": 0.5251,
      "step": 3431
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9794025421142578,
      "learning_rate": 9.106877783009236e-06,
      "loss": 0.5746,
      "step": 3432
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0107927322387695,
      "learning_rate": 9.106292482313819e-06,
      "loss": 0.557,
      "step": 3433
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.936278760433197,
      "learning_rate": 9.105707008716712e-06,
      "loss": 0.5274,
      "step": 3434
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0086160898208618,
      "learning_rate": 9.105121362242564e-06,
      "loss": 0.5522,
      "step": 3435
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0340970754623413,
      "learning_rate": 9.104535542916035e-06,
      "loss": 0.5361,
      "step": 3436
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9634861946105957,
      "learning_rate": 9.103949550761795e-06,
      "loss": 0.5679,
      "step": 3437
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0549044609069824,
      "learning_rate": 9.103363385804516e-06,
      "loss": 0.5772,
      "step": 3438
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0148675441741943,
      "learning_rate": 9.102777048068878e-06,
      "loss": 0.5776,
      "step": 3439
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0089242458343506,
      "learning_rate": 9.102190537579572e-06,
      "loss": 0.5738,
      "step": 3440
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9703475832939148,
      "learning_rate": 9.101603854361291e-06,
      "loss": 0.5424,
      "step": 3441
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0789226293563843,
      "learning_rate": 9.101016998438743e-06,
      "loss": 0.5257,
      "step": 3442
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1241742372512817,
      "learning_rate": 9.100429969836636e-06,
      "loss": 0.5992,
      "step": 3443
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0380760431289673,
      "learning_rate": 9.099842768579685e-06,
      "loss": 0.5483,
      "step": 3444
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.958708643913269,
      "learning_rate": 9.099255394692618e-06,
      "loss": 0.4942,
      "step": 3445
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9943519830703735,
      "learning_rate": 9.098667848200167e-06,
      "loss": 0.521,
      "step": 3446
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0717390775680542,
      "learning_rate": 9.09808012912707e-06,
      "loss": 0.58,
      "step": 3447
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9994247555732727,
      "learning_rate": 9.097492237498076e-06,
      "loss": 0.5667,
      "step": 3448
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.003158450126648,
      "learning_rate": 9.096904173337937e-06,
      "loss": 0.5413,
      "step": 3449
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1830018758773804,
      "learning_rate": 9.096315936671416e-06,
      "loss": 0.5561,
      "step": 3450
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.113922119140625,
      "learning_rate": 9.095727527523282e-06,
      "loss": 0.5983,
      "step": 3451
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1049774885177612,
      "learning_rate": 9.095138945918309e-06,
      "loss": 0.5698,
      "step": 3452
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.987943172454834,
      "learning_rate": 9.094550191881281e-06,
      "loss": 0.511,
      "step": 3453
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9769154787063599,
      "learning_rate": 9.093961265436988e-06,
      "loss": 0.5826,
      "step": 3454
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9816447496414185,
      "learning_rate": 9.093372166610229e-06,
      "loss": 0.5603,
      "step": 3455
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0525661706924438,
      "learning_rate": 9.092782895425806e-06,
      "loss": 0.5196,
      "step": 3456
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0274732112884521,
      "learning_rate": 9.092193451908533e-06,
      "loss": 0.5814,
      "step": 3457
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0459325313568115,
      "learning_rate": 9.091603836083231e-06,
      "loss": 0.6087,
      "step": 3458
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.008176565170288,
      "learning_rate": 9.091014047974725e-06,
      "loss": 0.4993,
      "step": 3459
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0283236503601074,
      "learning_rate": 9.090424087607848e-06,
      "loss": 0.5641,
      "step": 3460
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9741441011428833,
      "learning_rate": 9.089833955007443e-06,
      "loss": 0.5211,
      "step": 3461
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0388422012329102,
      "learning_rate": 9.089243650198359e-06,
      "loss": 0.5607,
      "step": 3462
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9902390837669373,
      "learning_rate": 9.088653173205449e-06,
      "loss": 0.5419,
      "step": 3463
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1023180484771729,
      "learning_rate": 9.088062524053575e-06,
      "loss": 0.5474,
      "step": 3464
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0138156414031982,
      "learning_rate": 9.087471702767612e-06,
      "loss": 0.5287,
      "step": 3465
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0068479776382446,
      "learning_rate": 9.086880709372434e-06,
      "loss": 0.5632,
      "step": 3466
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9584867358207703,
      "learning_rate": 9.086289543892928e-06,
      "loss": 0.507,
      "step": 3467
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9820325970649719,
      "learning_rate": 9.085698206353983e-06,
      "loss": 0.5424,
      "step": 3468
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0293880701065063,
      "learning_rate": 9.085106696780499e-06,
      "loss": 0.5286,
      "step": 3469
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1403424739837646,
      "learning_rate": 9.084515015197384e-06,
      "loss": 0.6048,
      "step": 3470
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.075882911682129,
      "learning_rate": 9.08392316162955e-06,
      "loss": 0.5827,
      "step": 3471
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9643628001213074,
      "learning_rate": 9.083331136101921e-06,
      "loss": 0.549,
      "step": 3472
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.06292724609375,
      "learning_rate": 9.08273893863942e-06,
      "loss": 0.5346,
      "step": 3473
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9991394281387329,
      "learning_rate": 9.082146569266988e-06,
      "loss": 0.5242,
      "step": 3474
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.058954119682312,
      "learning_rate": 9.081554028009562e-06,
      "loss": 0.5945,
      "step": 3475
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9722981452941895,
      "learning_rate": 9.080961314892096e-06,
      "loss": 0.554,
      "step": 3476
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.054129719734192,
      "learning_rate": 9.080368429939546e-06,
      "loss": 0.5414,
      "step": 3477
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.144458293914795,
      "learning_rate": 9.079775373176874e-06,
      "loss": 0.5698,
      "step": 3478
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.02845299243927,
      "learning_rate": 9.079182144629055e-06,
      "loss": 0.5586,
      "step": 3479
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9779374599456787,
      "learning_rate": 9.078588744321067e-06,
      "loss": 0.5392,
      "step": 3480
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0148526430130005,
      "learning_rate": 9.077995172277894e-06,
      "loss": 0.5343,
      "step": 3481
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9910771250724792,
      "learning_rate": 9.07740142852453e-06,
      "loss": 0.5702,
      "step": 3482
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0904290676116943,
      "learning_rate": 9.076807513085976e-06,
      "loss": 0.5666,
      "step": 3483
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.087215781211853,
      "learning_rate": 9.076213425987242e-06,
      "loss": 0.5903,
      "step": 3484
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0257717370986938,
      "learning_rate": 9.07561916725334e-06,
      "loss": 0.5993,
      "step": 3485
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1014600992202759,
      "learning_rate": 9.075024736909292e-06,
      "loss": 0.5377,
      "step": 3486
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0270522832870483,
      "learning_rate": 9.074430134980129e-06,
      "loss": 0.567,
      "step": 3487
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1104378700256348,
      "learning_rate": 9.073835361490885e-06,
      "loss": 0.6008,
      "step": 3488
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0529365539550781,
      "learning_rate": 9.073240416466609e-06,
      "loss": 0.577,
      "step": 3489
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0253465175628662,
      "learning_rate": 9.072645299932347e-06,
      "loss": 0.5496,
      "step": 3490
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.056884527206421,
      "learning_rate": 9.07205001191316e-06,
      "loss": 0.5844,
      "step": 3491
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9805004000663757,
      "learning_rate": 9.071454552434111e-06,
      "loss": 0.5795,
      "step": 3492
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0552654266357422,
      "learning_rate": 9.070858921520276e-06,
      "loss": 0.5736,
      "step": 3493
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9912766814231873,
      "learning_rate": 9.070263119196734e-06,
      "loss": 0.5463,
      "step": 3494
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0589691400527954,
      "learning_rate": 9.06966714548857e-06,
      "loss": 0.5687,
      "step": 3495
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0722336769104004,
      "learning_rate": 9.069071000420879e-06,
      "loss": 0.5335,
      "step": 3496
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0283418893814087,
      "learning_rate": 9.068474684018765e-06,
      "loss": 0.5064,
      "step": 3497
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0223890542984009,
      "learning_rate": 9.067878196307334e-06,
      "loss": 0.5515,
      "step": 3498
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.053024172782898,
      "learning_rate": 9.067281537311705e-06,
      "loss": 0.5376,
      "step": 3499
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0045981407165527,
      "learning_rate": 9.066684707056999e-06,
      "loss": 0.5207,
      "step": 3500
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0248801708221436,
      "learning_rate": 9.066087705568346e-06,
      "loss": 0.5377,
      "step": 3501
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0563328266143799,
      "learning_rate": 9.065490532870884e-06,
      "loss": 0.5073,
      "step": 3502
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0533092021942139,
      "learning_rate": 9.06489318898976e-06,
      "loss": 0.5845,
      "step": 3503
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1314165592193604,
      "learning_rate": 9.064295673950125e-06,
      "loss": 0.5801,
      "step": 3504
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0564184188842773,
      "learning_rate": 9.063697987777136e-06,
      "loss": 0.5964,
      "step": 3505
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0110223293304443,
      "learning_rate": 9.063100130495962e-06,
      "loss": 0.5694,
      "step": 3506
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9970470666885376,
      "learning_rate": 9.062502102131777e-06,
      "loss": 0.5601,
      "step": 3507
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0728727579116821,
      "learning_rate": 9.06190390270976e-06,
      "loss": 0.5879,
      "step": 3508
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9721595048904419,
      "learning_rate": 9.0613055322551e-06,
      "loss": 0.5765,
      "step": 3509
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0315375328063965,
      "learning_rate": 9.060706990792993e-06,
      "loss": 0.5413,
      "step": 3510
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.981389045715332,
      "learning_rate": 9.06010827834864e-06,
      "loss": 0.5462,
      "step": 3511
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0459198951721191,
      "learning_rate": 9.059509394947252e-06,
      "loss": 0.546,
      "step": 3512
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0360572338104248,
      "learning_rate": 9.058910340614045e-06,
      "loss": 0.5462,
      "step": 3513
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0255624055862427,
      "learning_rate": 9.058311115374244e-06,
      "loss": 0.5757,
      "step": 3514
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.991475522518158,
      "learning_rate": 9.057711719253077e-06,
      "loss": 0.5359,
      "step": 3515
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0928494930267334,
      "learning_rate": 9.057112152275788e-06,
      "loss": 0.5771,
      "step": 3516
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9886402487754822,
      "learning_rate": 9.05651241446762e-06,
      "loss": 0.5433,
      "step": 3517
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9913406372070312,
      "learning_rate": 9.055912505853826e-06,
      "loss": 0.5207,
      "step": 3518
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0736424922943115,
      "learning_rate": 9.055312426459663e-06,
      "loss": 0.5671,
      "step": 3519
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0118083953857422,
      "learning_rate": 9.054712176310405e-06,
      "loss": 0.6284,
      "step": 3520
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0888111591339111,
      "learning_rate": 9.05411175543132e-06,
      "loss": 0.5334,
      "step": 3521
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0824137926101685,
      "learning_rate": 9.053511163847694e-06,
      "loss": 0.5147,
      "step": 3522
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1179592609405518,
      "learning_rate": 9.052910401584812e-06,
      "loss": 0.6117,
      "step": 3523
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9232814311981201,
      "learning_rate": 9.052309468667974e-06,
      "loss": 0.5319,
      "step": 3524
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9559396505355835,
      "learning_rate": 9.05170836512248e-06,
      "loss": 0.505,
      "step": 3525
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0154615640640259,
      "learning_rate": 9.051107090973642e-06,
      "loss": 0.5432,
      "step": 3526
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0286099910736084,
      "learning_rate": 9.050505646246777e-06,
      "loss": 0.6052,
      "step": 3527
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9501967430114746,
      "learning_rate": 9.04990403096721e-06,
      "loss": 0.524,
      "step": 3528
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0195550918579102,
      "learning_rate": 9.049302245160273e-06,
      "loss": 0.5972,
      "step": 3529
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0528881549835205,
      "learning_rate": 9.048700288851305e-06,
      "loss": 0.5714,
      "step": 3530
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9923539161682129,
      "learning_rate": 9.048098162065652e-06,
      "loss": 0.5882,
      "step": 3531
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0214570760726929,
      "learning_rate": 9.047495864828668e-06,
      "loss": 0.5554,
      "step": 3532
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0018694400787354,
      "learning_rate": 9.046893397165713e-06,
      "loss": 0.5691,
      "step": 3533
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0348128080368042,
      "learning_rate": 9.046290759102155e-06,
      "loss": 0.5482,
      "step": 3534
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9383018612861633,
      "learning_rate": 9.04568795066337e-06,
      "loss": 0.5379,
      "step": 3535
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0389997959136963,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.5788,
      "step": 3536
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0366233587265015,
      "learning_rate": 9.044481822761651e-06,
      "loss": 0.5706,
      "step": 3537
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0714468955993652,
      "learning_rate": 9.043878503349503e-06,
      "loss": 0.555,
      "step": 3538
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9625380635261536,
      "learning_rate": 9.043275013663699e-06,
      "loss": 0.5278,
      "step": 3539
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0158138275146484,
      "learning_rate": 9.04267135372965e-06,
      "loss": 0.5828,
      "step": 3540
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1385537385940552,
      "learning_rate": 9.042067523572775e-06,
      "loss": 0.539,
      "step": 3541
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0848937034606934,
      "learning_rate": 9.041463523218496e-06,
      "loss": 0.6276,
      "step": 3542
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0919116735458374,
      "learning_rate": 9.040859352692249e-06,
      "loss": 0.612,
      "step": 3543
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0339999198913574,
      "learning_rate": 9.04025501201947e-06,
      "loss": 0.5891,
      "step": 3544
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9081238508224487,
      "learning_rate": 9.039650501225608e-06,
      "loss": 0.5463,
      "step": 3545
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.077070951461792,
      "learning_rate": 9.039045820336116e-06,
      "loss": 0.5707,
      "step": 3546
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9889419078826904,
      "learning_rate": 9.038440969376456e-06,
      "loss": 0.5492,
      "step": 3547
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0376169681549072,
      "learning_rate": 9.037835948372095e-06,
      "loss": 0.5193,
      "step": 3548
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0179705619812012,
      "learning_rate": 9.03723075734851e-06,
      "loss": 0.5445,
      "step": 3549
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0555214881896973,
      "learning_rate": 9.03662539633118e-06,
      "loss": 0.5367,
      "step": 3550
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.040494680404663,
      "learning_rate": 9.0360198653456e-06,
      "loss": 0.5805,
      "step": 3551
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.985089898109436,
      "learning_rate": 9.035414164417262e-06,
      "loss": 0.5218,
      "step": 3552
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0762962102890015,
      "learning_rate": 9.034808293571672e-06,
      "loss": 0.5925,
      "step": 3553
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0686595439910889,
      "learning_rate": 9.03420225283434e-06,
      "loss": 0.5909,
      "step": 3554
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9033122658729553,
      "learning_rate": 9.033596042230788e-06,
      "loss": 0.4947,
      "step": 3555
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.055325984954834,
      "learning_rate": 9.032989661786535e-06,
      "loss": 0.5935,
      "step": 3556
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9803203344345093,
      "learning_rate": 9.032383111527119e-06,
      "loss": 0.5634,
      "step": 3557
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0444809198379517,
      "learning_rate": 9.031776391478077e-06,
      "loss": 0.5677,
      "step": 3558
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0093274116516113,
      "learning_rate": 9.031169501664958e-06,
      "loss": 0.5633,
      "step": 3559
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0077446699142456,
      "learning_rate": 9.030562442113313e-06,
      "loss": 0.5624,
      "step": 3560
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9775265455245972,
      "learning_rate": 9.029955212848706e-06,
      "loss": 0.5783,
      "step": 3561
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0331493616104126,
      "learning_rate": 9.029347813896704e-06,
      "loss": 0.5234,
      "step": 3562
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9706992506980896,
      "learning_rate": 9.028740245282881e-06,
      "loss": 0.5356,
      "step": 3563
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9969547986984253,
      "learning_rate": 9.028132507032823e-06,
      "loss": 0.5537,
      "step": 3564
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9665756225585938,
      "learning_rate": 9.027524599172117e-06,
      "loss": 0.5453,
      "step": 3565
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1190989017486572,
      "learning_rate": 9.026916521726361e-06,
      "loss": 0.6168,
      "step": 3566
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0341004133224487,
      "learning_rate": 9.026308274721161e-06,
      "loss": 0.528,
      "step": 3567
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0381500720977783,
      "learning_rate": 9.025699858182125e-06,
      "loss": 0.5352,
      "step": 3568
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9730229377746582,
      "learning_rate": 9.02509127213487e-06,
      "loss": 0.5532,
      "step": 3569
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9915220737457275,
      "learning_rate": 9.024482516605026e-06,
      "loss": 0.5211,
      "step": 3570
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0444742441177368,
      "learning_rate": 9.023873591618224e-06,
      "loss": 0.5641,
      "step": 3571
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0250283479690552,
      "learning_rate": 9.023264497200102e-06,
      "loss": 0.5628,
      "step": 3572
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9876275062561035,
      "learning_rate": 9.022655233376308e-06,
      "loss": 0.5419,
      "step": 3573
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9815468788146973,
      "learning_rate": 9.022045800172493e-06,
      "loss": 0.5366,
      "step": 3574
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9828575849533081,
      "learning_rate": 9.021436197614326e-06,
      "loss": 0.5414,
      "step": 3575
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1150814294815063,
      "learning_rate": 9.020826425727468e-06,
      "loss": 0.5584,
      "step": 3576
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0118638277053833,
      "learning_rate": 9.020216484537595e-06,
      "loss": 0.5998,
      "step": 3577
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1020475625991821,
      "learning_rate": 9.019606374070394e-06,
      "loss": 0.5849,
      "step": 3578
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0934542417526245,
      "learning_rate": 9.01899609435155e-06,
      "loss": 0.5789,
      "step": 3579
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0679789781570435,
      "learning_rate": 9.018385645406765e-06,
      "loss": 0.5612,
      "step": 3580
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9859147071838379,
      "learning_rate": 9.017775027261735e-06,
      "loss": 0.5722,
      "step": 3581
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.018585205078125,
      "learning_rate": 9.017164239942178e-06,
      "loss": 0.5516,
      "step": 3582
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.986737847328186,
      "learning_rate": 9.016553283473808e-06,
      "loss": 0.576,
      "step": 3583
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.925209105014801,
      "learning_rate": 9.015942157882353e-06,
      "loss": 0.5147,
      "step": 3584
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9873782992362976,
      "learning_rate": 9.015330863193543e-06,
      "loss": 0.5421,
      "step": 3585
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.02249014377594,
      "learning_rate": 9.01471939943312e-06,
      "loss": 0.5569,
      "step": 3586
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0699585676193237,
      "learning_rate": 9.014107766626828e-06,
      "loss": 0.5106,
      "step": 3587
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9971165060997009,
      "learning_rate": 9.013495964800423e-06,
      "loss": 0.5997,
      "step": 3588
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9946675896644592,
      "learning_rate": 9.012883993979663e-06,
      "loss": 0.6373,
      "step": 3589
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9921069145202637,
      "learning_rate": 9.01227185419032e-06,
      "loss": 0.5683,
      "step": 3590
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1467154026031494,
      "learning_rate": 9.011659545458167e-06,
      "loss": 0.5316,
      "step": 3591
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0683609247207642,
      "learning_rate": 9.011047067808985e-06,
      "loss": 0.5763,
      "step": 3592
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9084008932113647,
      "learning_rate": 9.010434421268564e-06,
      "loss": 0.5203,
      "step": 3593
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1025091409683228,
      "learning_rate": 9.009821605862701e-06,
      "loss": 0.5671,
      "step": 3594
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9929082989692688,
      "learning_rate": 9.0092086216172e-06,
      "loss": 0.5385,
      "step": 3595
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9512858986854553,
      "learning_rate": 9.00859546855787e-06,
      "loss": 0.5357,
      "step": 3596
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9521564841270447,
      "learning_rate": 9.007982146710533e-06,
      "loss": 0.5221,
      "step": 3597
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0225049257278442,
      "learning_rate": 9.007368656101006e-06,
      "loss": 0.5682,
      "step": 3598
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9886544942855835,
      "learning_rate": 9.006754996755129e-06,
      "loss": 0.5341,
      "step": 3599
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0694276094436646,
      "learning_rate": 9.006141168698735e-06,
      "loss": 0.5978,
      "step": 3600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9691435694694519,
      "learning_rate": 9.005527171957676e-06,
      "loss": 0.5321,
      "step": 3601
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0941452980041504,
      "learning_rate": 9.004913006557798e-06,
      "loss": 0.5851,
      "step": 3602
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9894156455993652,
      "learning_rate": 9.004298672524967e-06,
      "loss": 0.5904,
      "step": 3603
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9922523498535156,
      "learning_rate": 9.003684169885049e-06,
      "loss": 0.5586,
      "step": 3604
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9479037523269653,
      "learning_rate": 9.00306949866392e-06,
      "loss": 0.5359,
      "step": 3605
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9888592958450317,
      "learning_rate": 9.002454658887458e-06,
      "loss": 0.5346,
      "step": 3606
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0730727910995483,
      "learning_rate": 9.001839650581554e-06,
      "loss": 0.5736,
      "step": 3607
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0897572040557861,
      "learning_rate": 9.001224473772104e-06,
      "loss": 0.5993,
      "step": 3608
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0272395610809326,
      "learning_rate": 9.000609128485011e-06,
      "loss": 0.501,
      "step": 3609
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.076352596282959,
      "learning_rate": 8.999993614746184e-06,
      "loss": 0.6239,
      "step": 3610
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9634479880332947,
      "learning_rate": 8.999377932581541e-06,
      "loss": 0.5195,
      "step": 3611
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0180588960647583,
      "learning_rate": 8.998762082017006e-06,
      "loss": 0.5528,
      "step": 3612
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1297459602355957,
      "learning_rate": 8.998146063078512e-06,
      "loss": 0.6002,
      "step": 3613
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0049798488616943,
      "learning_rate": 8.997529875791993e-06,
      "loss": 0.5378,
      "step": 3614
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9840946793556213,
      "learning_rate": 8.9969135201834e-06,
      "loss": 0.5592,
      "step": 3615
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0649902820587158,
      "learning_rate": 8.996296996278682e-06,
      "loss": 0.6086,
      "step": 3616
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9876604080200195,
      "learning_rate": 8.9956803041038e-06,
      "loss": 0.5474,
      "step": 3617
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9918978214263916,
      "learning_rate": 8.99506344368472e-06,
      "loss": 0.4996,
      "step": 3618
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9975392818450928,
      "learning_rate": 8.994446415047415e-06,
      "loss": 0.5084,
      "step": 3619
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0040109157562256,
      "learning_rate": 8.993829218217867e-06,
      "loss": 0.5604,
      "step": 3620
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0073280334472656,
      "learning_rate": 8.993211853222065e-06,
      "loss": 0.5751,
      "step": 3621
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0646840333938599,
      "learning_rate": 8.992594320086005e-06,
      "loss": 0.5922,
      "step": 3622
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9566529989242554,
      "learning_rate": 8.991976618835685e-06,
      "loss": 0.556,
      "step": 3623
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0741969347000122,
      "learning_rate": 8.991358749497117e-06,
      "loss": 0.5727,
      "step": 3624
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0362106561660767,
      "learning_rate": 8.990740712096317e-06,
      "loss": 0.5696,
      "step": 3625
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9603510499000549,
      "learning_rate": 8.99012250665931e-06,
      "loss": 0.5189,
      "step": 3626
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1036278009414673,
      "learning_rate": 8.989504133212123e-06,
      "loss": 0.6535,
      "step": 3627
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.18490469455719,
      "learning_rate": 8.988885591780795e-06,
      "loss": 0.5566,
      "step": 3628
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0526707172393799,
      "learning_rate": 8.988266882391374e-06,
      "loss": 0.5357,
      "step": 3629
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0246859788894653,
      "learning_rate": 8.987648005069907e-06,
      "loss": 0.5803,
      "step": 3630
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0607661008834839,
      "learning_rate": 8.987028959842454e-06,
      "loss": 0.5744,
      "step": 3631
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.073825478553772,
      "learning_rate": 8.986409746735084e-06,
      "loss": 0.5299,
      "step": 3632
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0737742185592651,
      "learning_rate": 8.985790365773864e-06,
      "loss": 0.5551,
      "step": 3633
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0232653617858887,
      "learning_rate": 8.985170816984878e-06,
      "loss": 0.5318,
      "step": 3634
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9758453369140625,
      "learning_rate": 8.984551100394212e-06,
      "loss": 0.5652,
      "step": 3635
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0398926734924316,
      "learning_rate": 8.98393121602796e-06,
      "loss": 0.5543,
      "step": 3636
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9773332476615906,
      "learning_rate": 8.983311163912227e-06,
      "loss": 0.5269,
      "step": 3637
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0076240301132202,
      "learning_rate": 8.982690944073113e-06,
      "loss": 0.5352,
      "step": 3638
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0406144857406616,
      "learning_rate": 8.982070556536741e-06,
      "loss": 0.5671,
      "step": 3639
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0525169372558594,
      "learning_rate": 8.98145000132923e-06,
      "loss": 0.5772,
      "step": 3640
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0494791269302368,
      "learning_rate": 8.980829278476711e-06,
      "loss": 0.5476,
      "step": 3641
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0154716968536377,
      "learning_rate": 8.980208388005318e-06,
      "loss": 0.5468,
      "step": 3642
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9966602325439453,
      "learning_rate": 8.979587329941197e-06,
      "loss": 0.5753,
      "step": 3643
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0978286266326904,
      "learning_rate": 8.978966104310497e-06,
      "loss": 0.6039,
      "step": 3644
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9729987978935242,
      "learning_rate": 8.978344711139374e-06,
      "loss": 0.5369,
      "step": 3645
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1959797143936157,
      "learning_rate": 8.977723150453999e-06,
      "loss": 0.5857,
      "step": 3646
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0096715688705444,
      "learning_rate": 8.977101422280536e-06,
      "loss": 0.5511,
      "step": 3647
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9775328040122986,
      "learning_rate": 8.97647952664517e-06,
      "loss": 0.517,
      "step": 3648
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.944109320640564,
      "learning_rate": 8.975857463574082e-06,
      "loss": 0.4827,
      "step": 3649
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0979822874069214,
      "learning_rate": 8.97523523309347e-06,
      "loss": 0.5146,
      "step": 3650
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0472910404205322,
      "learning_rate": 8.974612835229528e-06,
      "loss": 0.5592,
      "step": 3651
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9543821811676025,
      "learning_rate": 8.973990270008467e-06,
      "loss": 0.5554,
      "step": 3652
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9596877694129944,
      "learning_rate": 8.973367537456502e-06,
      "loss": 0.5711,
      "step": 3653
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1064952611923218,
      "learning_rate": 8.97274463759985e-06,
      "loss": 0.6165,
      "step": 3654
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1261231899261475,
      "learning_rate": 8.972121570464744e-06,
      "loss": 0.5903,
      "step": 3655
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0051484107971191,
      "learning_rate": 8.971498336077415e-06,
      "loss": 0.5356,
      "step": 3656
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0241944789886475,
      "learning_rate": 8.970874934464108e-06,
      "loss": 0.532,
      "step": 3657
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0307494401931763,
      "learning_rate": 8.970251365651071e-06,
      "loss": 0.5779,
      "step": 3658
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0298916101455688,
      "learning_rate": 8.969627629664559e-06,
      "loss": 0.4963,
      "step": 3659
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.992840051651001,
      "learning_rate": 8.969003726530838e-06,
      "loss": 0.5462,
      "step": 3660
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9768746495246887,
      "learning_rate": 8.968379656276177e-06,
      "loss": 0.5891,
      "step": 3661
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.003993034362793,
      "learning_rate": 8.967755418926854e-06,
      "loss": 0.5489,
      "step": 3662
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.04116690158844,
      "learning_rate": 8.967131014509152e-06,
      "loss": 0.5331,
      "step": 3663
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.098099708557129,
      "learning_rate": 8.966506443049366e-06,
      "loss": 0.5863,
      "step": 3664
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0058609247207642,
      "learning_rate": 8.965881704573789e-06,
      "loss": 0.5659,
      "step": 3665
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.147666573524475,
      "learning_rate": 8.965256799108733e-06,
      "loss": 0.5475,
      "step": 3666
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0394258499145508,
      "learning_rate": 8.964631726680504e-06,
      "loss": 0.591,
      "step": 3667
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0518441200256348,
      "learning_rate": 8.964006487315426e-06,
      "loss": 0.6134,
      "step": 3668
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.028096318244934,
      "learning_rate": 8.963381081039826e-06,
      "loss": 0.5475,
      "step": 3669
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9712651371955872,
      "learning_rate": 8.962755507880036e-06,
      "loss": 0.5356,
      "step": 3670
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.001176357269287,
      "learning_rate": 8.962129767862395e-06,
      "loss": 0.5402,
      "step": 3671
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9954277276992798,
      "learning_rate": 8.961503861013255e-06,
      "loss": 0.5845,
      "step": 3672
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0858553647994995,
      "learning_rate": 8.960877787358968e-06,
      "loss": 0.547,
      "step": 3673
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.068434476852417,
      "learning_rate": 8.960251546925895e-06,
      "loss": 0.572,
      "step": 3674
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.898658812046051,
      "learning_rate": 8.959625139740407e-06,
      "loss": 0.4826,
      "step": 3675
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0096508264541626,
      "learning_rate": 8.95899856582888e-06,
      "loss": 0.5776,
      "step": 3676
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.932982325553894,
      "learning_rate": 8.958371825217693e-06,
      "loss": 0.5157,
      "step": 3677
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0382206439971924,
      "learning_rate": 8.957744917933241e-06,
      "loss": 0.5287,
      "step": 3678
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0119495391845703,
      "learning_rate": 8.957117844001919e-06,
      "loss": 0.5399,
      "step": 3679
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.061253547668457,
      "learning_rate": 8.956490603450128e-06,
      "loss": 0.5403,
      "step": 3680
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0558768510818481,
      "learning_rate": 8.955863196304282e-06,
      "loss": 0.5512,
      "step": 3681
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0816651582717896,
      "learning_rate": 8.9552356225908e-06,
      "loss": 0.5424,
      "step": 3682
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.955443799495697,
      "learning_rate": 8.954607882336105e-06,
      "loss": 0.5196,
      "step": 3683
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.108539342880249,
      "learning_rate": 8.953979975566626e-06,
      "loss": 0.5846,
      "step": 3684
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.037209391593933,
      "learning_rate": 8.953351902308807e-06,
      "loss": 0.5725,
      "step": 3685
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0951324701309204,
      "learning_rate": 8.952723662589093e-06,
      "loss": 0.5924,
      "step": 3686
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0157463550567627,
      "learning_rate": 8.952095256433934e-06,
      "loss": 0.5943,
      "step": 3687
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.954087495803833,
      "learning_rate": 8.951466683869795e-06,
      "loss": 0.4879,
      "step": 3688
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0344793796539307,
      "learning_rate": 8.950837944923138e-06,
      "loss": 0.525,
      "step": 3689
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.050633192062378,
      "learning_rate": 8.95020903962044e-06,
      "loss": 0.5889,
      "step": 3690
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2160776853561401,
      "learning_rate": 8.94957996798818e-06,
      "loss": 0.5498,
      "step": 3691
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.988061249256134,
      "learning_rate": 8.948950730052847e-06,
      "loss": 0.5514,
      "step": 3692
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0714434385299683,
      "learning_rate": 8.948321325840937e-06,
      "loss": 0.6054,
      "step": 3693
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0225694179534912,
      "learning_rate": 8.94769175537895e-06,
      "loss": 0.5472,
      "step": 3694
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0245158672332764,
      "learning_rate": 8.9470620186934e-06,
      "loss": 0.5868,
      "step": 3695
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0072945356369019,
      "learning_rate": 8.946432115810795e-06,
      "loss": 0.5886,
      "step": 3696
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1089059114456177,
      "learning_rate": 8.945802046757666e-06,
      "loss": 0.6025,
      "step": 3697
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.119722604751587,
      "learning_rate": 8.945171811560535e-06,
      "loss": 0.5642,
      "step": 3698
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1201279163360596,
      "learning_rate": 8.944541410245947e-06,
      "loss": 0.5244,
      "step": 3699
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0516458749771118,
      "learning_rate": 8.943910842840439e-06,
      "loss": 0.5755,
      "step": 3700
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0389517545700073,
      "learning_rate": 8.943280109370568e-06,
      "loss": 0.5597,
      "step": 3701
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9651356935501099,
      "learning_rate": 8.942649209862888e-06,
      "loss": 0.5182,
      "step": 3702
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9806023240089417,
      "learning_rate": 8.942018144343965e-06,
      "loss": 0.5615,
      "step": 3703
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.080400824546814,
      "learning_rate": 8.941386912840372e-06,
      "loss": 0.5771,
      "step": 3704
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9524903893470764,
      "learning_rate": 8.940755515378687e-06,
      "loss": 0.5383,
      "step": 3705
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0646077394485474,
      "learning_rate": 8.940123951985495e-06,
      "loss": 0.547,
      "step": 3706
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1073490381240845,
      "learning_rate": 8.939492222687392e-06,
      "loss": 0.5689,
      "step": 3707
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.01206374168396,
      "learning_rate": 8.938860327510975e-06,
      "loss": 0.5544,
      "step": 3708
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0279171466827393,
      "learning_rate": 8.938228266482852e-06,
      "loss": 0.5748,
      "step": 3709
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0886095762252808,
      "learning_rate": 8.937596039629637e-06,
      "loss": 0.5399,
      "step": 3710
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0480284690856934,
      "learning_rate": 8.93696364697795e-06,
      "loss": 0.566,
      "step": 3711
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9844214916229248,
      "learning_rate": 8.936331088554419e-06,
      "loss": 0.5417,
      "step": 3712
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9495744705200195,
      "learning_rate": 8.93569836438568e-06,
      "loss": 0.5301,
      "step": 3713
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0315167903900146,
      "learning_rate": 8.935065474498375e-06,
      "loss": 0.543,
      "step": 3714
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0174928903579712,
      "learning_rate": 8.934432418919153e-06,
      "loss": 0.5741,
      "step": 3715
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0040485858917236,
      "learning_rate": 8.933799197674667e-06,
      "loss": 0.537,
      "step": 3716
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.072627305984497,
      "learning_rate": 8.933165810791579e-06,
      "loss": 0.5555,
      "step": 3717
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9701638221740723,
      "learning_rate": 8.932532258296565e-06,
      "loss": 0.5739,
      "step": 3718
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0700533390045166,
      "learning_rate": 8.931898540216297e-06,
      "loss": 0.5862,
      "step": 3719
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0565009117126465,
      "learning_rate": 8.931264656577459e-06,
      "loss": 0.5621,
      "step": 3720
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0032659769058228,
      "learning_rate": 8.930630607406743e-06,
      "loss": 0.5661,
      "step": 3721
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0463674068450928,
      "learning_rate": 8.929996392730844e-06,
      "loss": 0.5696,
      "step": 3722
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0808651447296143,
      "learning_rate": 8.92936201257647e-06,
      "loss": 0.5766,
      "step": 3723
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0453593730926514,
      "learning_rate": 8.928727466970331e-06,
      "loss": 0.5321,
      "step": 3724
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1240564584732056,
      "learning_rate": 8.928092755939145e-06,
      "loss": 0.5606,
      "step": 3725
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1166630983352661,
      "learning_rate": 8.927457879509638e-06,
      "loss": 0.5597,
      "step": 3726
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0657453536987305,
      "learning_rate": 8.926822837708542e-06,
      "loss": 0.5721,
      "step": 3727
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9846172332763672,
      "learning_rate": 8.926187630562597e-06,
      "loss": 0.5524,
      "step": 3728
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0242836475372314,
      "learning_rate": 8.925552258098549e-06,
      "loss": 0.5224,
      "step": 3729
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0484217405319214,
      "learning_rate": 8.924916720343151e-06,
      "loss": 0.5579,
      "step": 3730
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1069062948226929,
      "learning_rate": 8.924281017323164e-06,
      "loss": 0.5332,
      "step": 3731
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9859890341758728,
      "learning_rate": 8.923645149065354e-06,
      "loss": 0.5214,
      "step": 3732
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0303205251693726,
      "learning_rate": 8.923009115596498e-06,
      "loss": 0.5266,
      "step": 3733
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9743627309799194,
      "learning_rate": 8.922372916943374e-06,
      "loss": 0.5929,
      "step": 3734
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0966832637786865,
      "learning_rate": 8.921736553132772e-06,
      "loss": 0.5618,
      "step": 3735
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0996084213256836,
      "learning_rate": 8.921100024191486e-06,
      "loss": 0.5385,
      "step": 3736
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0357041358947754,
      "learning_rate": 8.920463330146318e-06,
      "loss": 0.5574,
      "step": 3737
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0976433753967285,
      "learning_rate": 8.919826471024078e-06,
      "loss": 0.5603,
      "step": 3738
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.055237054824829,
      "learning_rate": 8.919189446851583e-06,
      "loss": 0.5319,
      "step": 3739
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0910391807556152,
      "learning_rate": 8.918552257655652e-06,
      "loss": 0.5432,
      "step": 3740
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0958404541015625,
      "learning_rate": 8.917914903463119e-06,
      "loss": 0.6082,
      "step": 3741
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0359296798706055,
      "learning_rate": 8.917277384300817e-06,
      "loss": 0.5747,
      "step": 3742
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9910179376602173,
      "learning_rate": 8.916639700195593e-06,
      "loss": 0.5327,
      "step": 3743
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0079916715621948,
      "learning_rate": 8.916001851174296e-06,
      "loss": 0.5484,
      "step": 3744
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.047631859779358,
      "learning_rate": 8.915363837263782e-06,
      "loss": 0.5795,
      "step": 3745
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9888326525688171,
      "learning_rate": 8.91472565849092e-06,
      "loss": 0.5323,
      "step": 3746
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9907726049423218,
      "learning_rate": 8.914087314882578e-06,
      "loss": 0.5223,
      "step": 3747
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0509141683578491,
      "learning_rate": 8.913448806465634e-06,
      "loss": 0.5716,
      "step": 3748
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9748091697692871,
      "learning_rate": 8.912810133266976e-06,
      "loss": 0.5324,
      "step": 3749
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0037689208984375,
      "learning_rate": 8.912171295313493e-06,
      "loss": 0.5359,
      "step": 3750
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.970689058303833,
      "learning_rate": 8.911532292632089e-06,
      "loss": 0.5731,
      "step": 3751
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1158560514450073,
      "learning_rate": 8.910893125249666e-06,
      "loss": 0.5589,
      "step": 3752
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.082456350326538,
      "learning_rate": 8.91025379319314e-06,
      "loss": 0.5883,
      "step": 3753
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9528515338897705,
      "learning_rate": 8.909614296489428e-06,
      "loss": 0.5179,
      "step": 3754
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0401341915130615,
      "learning_rate": 8.908974635165458e-06,
      "loss": 0.5784,
      "step": 3755
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9220054745674133,
      "learning_rate": 8.908334809248165e-06,
      "loss": 0.5363,
      "step": 3756
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0673452615737915,
      "learning_rate": 8.90769481876449e-06,
      "loss": 0.6093,
      "step": 3757
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0971558094024658,
      "learning_rate": 8.90705466374138e-06,
      "loss": 0.5634,
      "step": 3758
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1336888074874878,
      "learning_rate": 8.906414344205789e-06,
      "loss": 0.5662,
      "step": 3759
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0179674625396729,
      "learning_rate": 8.905773860184679e-06,
      "loss": 0.5693,
      "step": 3760
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9385787844657898,
      "learning_rate": 8.905133211705019e-06,
      "loss": 0.5391,
      "step": 3761
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9712759256362915,
      "learning_rate": 8.904492398793785e-06,
      "loss": 0.5767,
      "step": 3762
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9476292729377747,
      "learning_rate": 8.903851421477959e-06,
      "loss": 0.5587,
      "step": 3763
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9753716588020325,
      "learning_rate": 8.90321027978453e-06,
      "loss": 0.5619,
      "step": 3764
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0432862043380737,
      "learning_rate": 8.902568973740495e-06,
      "loss": 0.5288,
      "step": 3765
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0975617170333862,
      "learning_rate": 8.901927503372855e-06,
      "loss": 0.5406,
      "step": 3766
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0205985307693481,
      "learning_rate": 8.901285868708622e-06,
      "loss": 0.5532,
      "step": 3767
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9458399415016174,
      "learning_rate": 8.900644069774815e-06,
      "loss": 0.5218,
      "step": 3768
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0280425548553467,
      "learning_rate": 8.900002106598453e-06,
      "loss": 0.5456,
      "step": 3769
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0532450675964355,
      "learning_rate": 8.89935997920657e-06,
      "loss": 0.5718,
      "step": 3770
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1915584802627563,
      "learning_rate": 8.898717687626203e-06,
      "loss": 0.5831,
      "step": 3771
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0443774461746216,
      "learning_rate": 8.898075231884397e-06,
      "loss": 0.5566,
      "step": 3772
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9577083587646484,
      "learning_rate": 8.897432612008206e-06,
      "loss": 0.5656,
      "step": 3773
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9909464120864868,
      "learning_rate": 8.896789828024682e-06,
      "loss": 0.5585,
      "step": 3774
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0283769369125366,
      "learning_rate": 8.896146879960896e-06,
      "loss": 0.5992,
      "step": 3775
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1171537637710571,
      "learning_rate": 8.895503767843918e-06,
      "loss": 0.5796,
      "step": 3776
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1655315160751343,
      "learning_rate": 8.89486049170083e-06,
      "loss": 0.5953,
      "step": 3777
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9457029104232788,
      "learning_rate": 8.894217051558713e-06,
      "loss": 0.5288,
      "step": 3778
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9799229502677917,
      "learning_rate": 8.893573447444663e-06,
      "loss": 0.5732,
      "step": 3779
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0082273483276367,
      "learning_rate": 8.892929679385783e-06,
      "loss": 0.5022,
      "step": 3780
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0606393814086914,
      "learning_rate": 8.892285747409172e-06,
      "loss": 0.5212,
      "step": 3781
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0303373336791992,
      "learning_rate": 8.891641651541953e-06,
      "loss": 0.5262,
      "step": 3782
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.126741647720337,
      "learning_rate": 8.89099739181124e-06,
      "loss": 0.5396,
      "step": 3783
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0784149169921875,
      "learning_rate": 8.890352968244162e-06,
      "loss": 0.5377,
      "step": 3784
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.032558798789978,
      "learning_rate": 8.889708380867856e-06,
      "loss": 0.5923,
      "step": 3785
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9988136291503906,
      "learning_rate": 8.88906362970946e-06,
      "loss": 0.5248,
      "step": 3786
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9523390531539917,
      "learning_rate": 8.888418714796124e-06,
      "loss": 0.556,
      "step": 3787
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0088428258895874,
      "learning_rate": 8.887773636155002e-06,
      "loss": 0.5189,
      "step": 3788
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0588726997375488,
      "learning_rate": 8.887128393813257e-06,
      "loss": 0.5795,
      "step": 3789
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0953381061553955,
      "learning_rate": 8.886482987798059e-06,
      "loss": 0.5686,
      "step": 3790
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1218312978744507,
      "learning_rate": 8.885837418136581e-06,
      "loss": 0.534,
      "step": 3791
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.00978422164917,
      "learning_rate": 8.885191684856007e-06,
      "loss": 0.4912,
      "step": 3792
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0210072994232178,
      "learning_rate": 8.884545787983528e-06,
      "loss": 0.5406,
      "step": 3793
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0807887315750122,
      "learning_rate": 8.88389972754634e-06,
      "loss": 0.5797,
      "step": 3794
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.090825080871582,
      "learning_rate": 8.883253503571643e-06,
      "loss": 0.5606,
      "step": 3795
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0245369672775269,
      "learning_rate": 8.882607116086651e-06,
      "loss": 0.5506,
      "step": 3796
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9361253380775452,
      "learning_rate": 8.881960565118581e-06,
      "loss": 0.4825,
      "step": 3797
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1194905042648315,
      "learning_rate": 8.881313850694653e-06,
      "loss": 0.5873,
      "step": 3798
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1902475357055664,
      "learning_rate": 8.880666972842105e-06,
      "loss": 0.5689,
      "step": 3799
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.98819500207901,
      "learning_rate": 8.880019931588167e-06,
      "loss": 0.5874,
      "step": 3800
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.106116771697998,
      "learning_rate": 8.87937272696009e-06,
      "loss": 0.5685,
      "step": 3801
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.007765531539917,
      "learning_rate": 8.878725358985121e-06,
      "loss": 0.5097,
      "step": 3802
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0698480606079102,
      "learning_rate": 8.87807782769052e-06,
      "loss": 0.6116,
      "step": 3803
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0733602046966553,
      "learning_rate": 8.877430133103555e-06,
      "loss": 0.5521,
      "step": 3804
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0455069541931152,
      "learning_rate": 8.876782275251491e-06,
      "loss": 0.5661,
      "step": 3805
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0989000797271729,
      "learning_rate": 8.876134254161617e-06,
      "loss": 0.5475,
      "step": 3806
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0597692728042603,
      "learning_rate": 8.87548606986121e-06,
      "loss": 0.5844,
      "step": 3807
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.148749589920044,
      "learning_rate": 8.874837722377568e-06,
      "loss": 0.5301,
      "step": 3808
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0871633291244507,
      "learning_rate": 8.87418921173799e-06,
      "loss": 0.5459,
      "step": 3809
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9634197354316711,
      "learning_rate": 8.87354053796978e-06,
      "loss": 0.5407,
      "step": 3810
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9858682155609131,
      "learning_rate": 8.872891701100253e-06,
      "loss": 0.5547,
      "step": 3811
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.064706563949585,
      "learning_rate": 8.872242701156731e-06,
      "loss": 0.5557,
      "step": 3812
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0465726852416992,
      "learning_rate": 8.871593538166538e-06,
      "loss": 0.5593,
      "step": 3813
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.292360544204712,
      "learning_rate": 8.870944212157008e-06,
      "loss": 0.5751,
      "step": 3814
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0987080335617065,
      "learning_rate": 8.870294723155486e-06,
      "loss": 0.5797,
      "step": 3815
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0372346639633179,
      "learning_rate": 8.869645071189316e-06,
      "loss": 0.5913,
      "step": 3816
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.109615445137024,
      "learning_rate": 8.868995256285853e-06,
      "loss": 0.5662,
      "step": 3817
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.04058837890625,
      "learning_rate": 8.868345278472458e-06,
      "loss": 0.5612,
      "step": 3818
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0476531982421875,
      "learning_rate": 8.867695137776503e-06,
      "loss": 0.5716,
      "step": 3819
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0954654216766357,
      "learning_rate": 8.86704483422536e-06,
      "loss": 0.5752,
      "step": 3820
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0084110498428345,
      "learning_rate": 8.86639436784641e-06,
      "loss": 0.5679,
      "step": 3821
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9866928458213806,
      "learning_rate": 8.865743738667045e-06,
      "loss": 0.4968,
      "step": 3822
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9854780435562134,
      "learning_rate": 8.865092946714657e-06,
      "loss": 0.5099,
      "step": 3823
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.969935953617096,
      "learning_rate": 8.864441992016653e-06,
      "loss": 0.5269,
      "step": 3824
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0676237344741821,
      "learning_rate": 8.863790874600438e-06,
      "loss": 0.547,
      "step": 3825
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9443878531455994,
      "learning_rate": 8.863139594493432e-06,
      "loss": 0.5228,
      "step": 3826
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0739343166351318,
      "learning_rate": 8.862488151723055e-06,
      "loss": 0.5478,
      "step": 3827
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0631316900253296,
      "learning_rate": 8.86183654631674e-06,
      "loss": 0.5672,
      "step": 3828
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9959419965744019,
      "learning_rate": 8.861184778301921e-06,
      "loss": 0.5609,
      "step": 3829
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0151301622390747,
      "learning_rate": 8.860532847706046e-06,
      "loss": 0.5751,
      "step": 3830
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0595030784606934,
      "learning_rate": 8.85988075455656e-06,
      "loss": 0.5408,
      "step": 3831
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0776829719543457,
      "learning_rate": 8.859228498880923e-06,
      "loss": 0.5859,
      "step": 3832
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0213021039962769,
      "learning_rate": 8.8585760807066e-06,
      "loss": 0.5698,
      "step": 3833
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.074037790298462,
      "learning_rate": 8.85792350006106e-06,
      "loss": 0.6023,
      "step": 3834
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9719785451889038,
      "learning_rate": 8.857270756971785e-06,
      "loss": 0.54,
      "step": 3835
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0435043573379517,
      "learning_rate": 8.856617851466254e-06,
      "loss": 0.5853,
      "step": 3836
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9525250792503357,
      "learning_rate": 8.855964783571963e-06,
      "loss": 0.5082,
      "step": 3837
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0572185516357422,
      "learning_rate": 8.855311553316409e-06,
      "loss": 0.5296,
      "step": 3838
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.005979061126709,
      "learning_rate": 8.854658160727096e-06,
      "loss": 0.5716,
      "step": 3839
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9383775591850281,
      "learning_rate": 8.854004605831536e-06,
      "loss": 0.4758,
      "step": 3840
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0103214979171753,
      "learning_rate": 8.853350888657251e-06,
      "loss": 0.5514,
      "step": 3841
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0398743152618408,
      "learning_rate": 8.852697009231766e-06,
      "loss": 0.5314,
      "step": 3842
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0169912576675415,
      "learning_rate": 8.852042967582611e-06,
      "loss": 0.5853,
      "step": 3843
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0007593631744385,
      "learning_rate": 8.851388763737328e-06,
      "loss": 0.4835,
      "step": 3844
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0083773136138916,
      "learning_rate": 8.850734397723461e-06,
      "loss": 0.5283,
      "step": 3845
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2991793155670166,
      "learning_rate": 8.850079869568565e-06,
      "loss": 0.594,
      "step": 3846
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9768350720405579,
      "learning_rate": 8.849425179300197e-06,
      "loss": 0.5486,
      "step": 3847
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0022505521774292,
      "learning_rate": 8.848770326945927e-06,
      "loss": 0.5329,
      "step": 3848
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0250089168548584,
      "learning_rate": 8.84811531253333e-06,
      "loss": 0.5185,
      "step": 3849
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1248204708099365,
      "learning_rate": 8.847460136089982e-06,
      "loss": 0.5634,
      "step": 3850
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0407178401947021,
      "learning_rate": 8.846804797643472e-06,
      "loss": 0.5423,
      "step": 3851
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9828975796699524,
      "learning_rate": 8.846149297221394e-06,
      "loss": 0.5327,
      "step": 3852
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0683937072753906,
      "learning_rate": 8.845493634851348e-06,
      "loss": 0.6222,
      "step": 3853
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9499751925468445,
      "learning_rate": 8.844837810560943e-06,
      "loss": 0.5243,
      "step": 3854
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0025689601898193,
      "learning_rate": 8.844181824377793e-06,
      "loss": 0.5128,
      "step": 3855
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9878833293914795,
      "learning_rate": 8.843525676329521e-06,
      "loss": 0.5791,
      "step": 3856
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9180898666381836,
      "learning_rate": 8.842869366443751e-06,
      "loss": 0.5351,
      "step": 3857
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9682977795600891,
      "learning_rate": 8.842212894748122e-06,
      "loss": 0.5325,
      "step": 3858
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9068602919578552,
      "learning_rate": 8.841556261270272e-06,
      "loss": 0.5159,
      "step": 3859
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0087989568710327,
      "learning_rate": 8.840899466037854e-06,
      "loss": 0.5818,
      "step": 3860
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9543938040733337,
      "learning_rate": 8.840242509078521e-06,
      "loss": 0.5371,
      "step": 3861
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9779382944107056,
      "learning_rate": 8.839585390419933e-06,
      "loss": 0.5227,
      "step": 3862
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9689169526100159,
      "learning_rate": 8.838928110089763e-06,
      "loss": 0.5444,
      "step": 3863
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.093464970588684,
      "learning_rate": 8.838270668115685e-06,
      "loss": 0.584,
      "step": 3864
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9167064428329468,
      "learning_rate": 8.837613064525381e-06,
      "loss": 0.5368,
      "step": 3865
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0776047706604004,
      "learning_rate": 8.83695529934654e-06,
      "loss": 0.5563,
      "step": 3866
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9500616788864136,
      "learning_rate": 8.83629737260686e-06,
      "loss": 0.5357,
      "step": 3867
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.008630633354187,
      "learning_rate": 8.835639284334043e-06,
      "loss": 0.5135,
      "step": 3868
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0061575174331665,
      "learning_rate": 8.834981034555799e-06,
      "loss": 0.5408,
      "step": 3869
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9785071611404419,
      "learning_rate": 8.834322623299844e-06,
      "loss": 0.5372,
      "step": 3870
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.997992217540741,
      "learning_rate": 8.833664050593904e-06,
      "loss": 0.5252,
      "step": 3871
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0702146291732788,
      "learning_rate": 8.833005316465706e-06,
      "loss": 0.5157,
      "step": 3872
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.964113712310791,
      "learning_rate": 8.832346420942987e-06,
      "loss": 0.5289,
      "step": 3873
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9863677620887756,
      "learning_rate": 8.831687364053493e-06,
      "loss": 0.5532,
      "step": 3874
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1146938800811768,
      "learning_rate": 8.831028145824974e-06,
      "loss": 0.5715,
      "step": 3875
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.073601484298706,
      "learning_rate": 8.830368766285186e-06,
      "loss": 0.5765,
      "step": 3876
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0261064767837524,
      "learning_rate": 8.829709225461894e-06,
      "loss": 0.5395,
      "step": 3877
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.164169192314148,
      "learning_rate": 8.829049523382871e-06,
      "loss": 0.5879,
      "step": 3878
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9541839361190796,
      "learning_rate": 8.828389660075891e-06,
      "loss": 0.5518,
      "step": 3879
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9897672533988953,
      "learning_rate": 8.82772963556874e-06,
      "loss": 0.5619,
      "step": 3880
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0605300664901733,
      "learning_rate": 8.827069449889211e-06,
      "loss": 0.5179,
      "step": 3881
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0233609676361084,
      "learning_rate": 8.8264091030651e-06,
      "loss": 0.555,
      "step": 3882
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.999971866607666,
      "learning_rate": 8.825748595124214e-06,
      "loss": 0.5514,
      "step": 3883
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0390762090682983,
      "learning_rate": 8.825087926094363e-06,
      "loss": 0.52,
      "step": 3884
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9346777200698853,
      "learning_rate": 8.824427096003367e-06,
      "loss": 0.5183,
      "step": 3885
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0920510292053223,
      "learning_rate": 8.823766104879047e-06,
      "loss": 0.5254,
      "step": 3886
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9973437190055847,
      "learning_rate": 8.823104952749242e-06,
      "loss": 0.505,
      "step": 3887
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9639909267425537,
      "learning_rate": 8.822443639641785e-06,
      "loss": 0.4865,
      "step": 3888
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0465211868286133,
      "learning_rate": 8.821782165584524e-06,
      "loss": 0.5365,
      "step": 3889
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0439425706863403,
      "learning_rate": 8.82112053060531e-06,
      "loss": 0.5753,
      "step": 3890
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.057376742362976,
      "learning_rate": 8.820458734732004e-06,
      "loss": 0.6169,
      "step": 3891
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.064401626586914,
      "learning_rate": 8.819796777992471e-06,
      "loss": 0.5196,
      "step": 3892
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0528103113174438,
      "learning_rate": 8.819134660414585e-06,
      "loss": 0.5504,
      "step": 3893
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.065921425819397,
      "learning_rate": 8.818472382026222e-06,
      "loss": 0.5123,
      "step": 3894
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.990153968334198,
      "learning_rate": 8.817809942855272e-06,
      "loss": 0.524,
      "step": 3895
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9567554593086243,
      "learning_rate": 8.817147342929626e-06,
      "loss": 0.5169,
      "step": 3896
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.169144630432129,
      "learning_rate": 8.816484582277184e-06,
      "loss": 0.5749,
      "step": 3897
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0559519529342651,
      "learning_rate": 8.815821660925853e-06,
      "loss": 0.5486,
      "step": 3898
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0077050924301147,
      "learning_rate": 8.815158578903548e-06,
      "loss": 0.5413,
      "step": 3899
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9848203659057617,
      "learning_rate": 8.814495336238185e-06,
      "loss": 0.5758,
      "step": 3900
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0258076190948486,
      "learning_rate": 8.813831932957696e-06,
      "loss": 0.5552,
      "step": 3901
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9313562512397766,
      "learning_rate": 8.813168369090007e-06,
      "loss": 0.4832,
      "step": 3902
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9199298620223999,
      "learning_rate": 8.812504644663066e-06,
      "loss": 0.5393,
      "step": 3903
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1476012468338013,
      "learning_rate": 8.811840759704816e-06,
      "loss": 0.5524,
      "step": 3904
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9469196200370789,
      "learning_rate": 8.811176714243213e-06,
      "loss": 0.5498,
      "step": 3905
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.152643084526062,
      "learning_rate": 8.810512508306216e-06,
      "loss": 0.5241,
      "step": 3906
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9760724902153015,
      "learning_rate": 8.809848141921793e-06,
      "loss": 0.5174,
      "step": 3907
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0510609149932861,
      "learning_rate": 8.809183615117919e-06,
      "loss": 0.5741,
      "step": 3908
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0153599977493286,
      "learning_rate": 8.808518927922574e-06,
      "loss": 0.5575,
      "step": 3909
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0817402601242065,
      "learning_rate": 8.807854080363745e-06,
      "loss": 0.5589,
      "step": 3910
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1702064275741577,
      "learning_rate": 8.807189072469428e-06,
      "loss": 0.5604,
      "step": 3911
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9715337753295898,
      "learning_rate": 8.806523904267623e-06,
      "loss": 0.5388,
      "step": 3912
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.10527765750885,
      "learning_rate": 8.80585857578634e-06,
      "loss": 0.588,
      "step": 3913
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9401335120201111,
      "learning_rate": 8.80519308705359e-06,
      "loss": 0.5018,
      "step": 3914
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0677719116210938,
      "learning_rate": 8.804527438097396e-06,
      "loss": 0.5637,
      "step": 3915
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0230120420455933,
      "learning_rate": 8.803861628945787e-06,
      "loss": 0.5323,
      "step": 3916
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1242260932922363,
      "learning_rate": 8.803195659626798e-06,
      "loss": 0.597,
      "step": 3917
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9725340604782104,
      "learning_rate": 8.802529530168469e-06,
      "loss": 0.5711,
      "step": 3918
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9721158742904663,
      "learning_rate": 8.801863240598851e-06,
      "loss": 0.532,
      "step": 3919
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0960687398910522,
      "learning_rate": 8.801196790945999e-06,
      "loss": 0.5299,
      "step": 3920
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0951040983200073,
      "learning_rate": 8.800530181237971e-06,
      "loss": 0.5936,
      "step": 3921
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0156009197235107,
      "learning_rate": 8.799863411502838e-06,
      "loss": 0.5214,
      "step": 3922
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9303261637687683,
      "learning_rate": 8.799196481768677e-06,
      "loss": 0.5371,
      "step": 3923
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.001246452331543,
      "learning_rate": 8.798529392063569e-06,
      "loss": 0.5324,
      "step": 3924
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9814727902412415,
      "learning_rate": 8.7978621424156e-06,
      "loss": 0.515,
      "step": 3925
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0274420976638794,
      "learning_rate": 8.79719473285287e-06,
      "loss": 0.5823,
      "step": 3926
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0317124128341675,
      "learning_rate": 8.796527163403479e-06,
      "loss": 0.5664,
      "step": 3927
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1095083951950073,
      "learning_rate": 8.795859434095535e-06,
      "loss": 0.5479,
      "step": 3928
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0397136211395264,
      "learning_rate": 8.795191544957156e-06,
      "loss": 0.6244,
      "step": 3929
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0844511985778809,
      "learning_rate": 8.794523496016465e-06,
      "loss": 0.6134,
      "step": 3930
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0013247728347778,
      "learning_rate": 8.793855287301588e-06,
      "loss": 0.5852,
      "step": 3931
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9349052906036377,
      "learning_rate": 8.793186918840661e-06,
      "loss": 0.4861,
      "step": 3932
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.006710171699524,
      "learning_rate": 8.792518390661831e-06,
      "loss": 0.552,
      "step": 3933
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0546201467514038,
      "learning_rate": 8.791849702793245e-06,
      "loss": 0.5596,
      "step": 3934
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0243421792984009,
      "learning_rate": 8.791180855263057e-06,
      "loss": 0.5695,
      "step": 3935
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0765653848648071,
      "learning_rate": 8.790511848099433e-06,
      "loss": 0.5773,
      "step": 3936
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9342508316040039,
      "learning_rate": 8.789842681330543e-06,
      "loss": 0.5546,
      "step": 3937
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0190470218658447,
      "learning_rate": 8.789173354984557e-06,
      "loss": 0.5331,
      "step": 3938
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0095438957214355,
      "learning_rate": 8.788503869089667e-06,
      "loss": 0.6117,
      "step": 3939
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0041183233261108,
      "learning_rate": 8.787834223674056e-06,
      "loss": 0.5361,
      "step": 3940
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0229700803756714,
      "learning_rate": 8.787164418765923e-06,
      "loss": 0.5201,
      "step": 3941
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0589138269424438,
      "learning_rate": 8.786494454393472e-06,
      "loss": 0.5675,
      "step": 3942
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0750982761383057,
      "learning_rate": 8.785824330584912e-06,
      "loss": 0.5354,
      "step": 3943
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0840270519256592,
      "learning_rate": 8.785154047368459e-06,
      "loss": 0.5619,
      "step": 3944
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.048685073852539,
      "learning_rate": 8.784483604772336e-06,
      "loss": 0.5506,
      "step": 3945
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.044411301612854,
      "learning_rate": 8.783813002824773e-06,
      "loss": 0.5588,
      "step": 3946
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0236300230026245,
      "learning_rate": 8.783142241554009e-06,
      "loss": 0.5374,
      "step": 3947
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0593035221099854,
      "learning_rate": 8.782471320988284e-06,
      "loss": 0.5684,
      "step": 3948
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.996816873550415,
      "learning_rate": 8.781800241155851e-06,
      "loss": 0.5102,
      "step": 3949
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0247293710708618,
      "learning_rate": 8.781129002084965e-06,
      "loss": 0.5328,
      "step": 3950
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0288833379745483,
      "learning_rate": 8.780457603803892e-06,
      "loss": 0.5485,
      "step": 3951
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0378539562225342,
      "learning_rate": 8.779786046340898e-06,
      "loss": 0.5774,
      "step": 3952
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0052213668823242,
      "learning_rate": 8.779114329724265e-06,
      "loss": 0.5665,
      "step": 3953
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9967714548110962,
      "learning_rate": 8.778442453982272e-06,
      "loss": 0.5205,
      "step": 3954
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.982059121131897,
      "learning_rate": 8.777770419143214e-06,
      "loss": 0.5836,
      "step": 3955
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0280338525772095,
      "learning_rate": 8.777098225235384e-06,
      "loss": 0.5844,
      "step": 3956
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0566469430923462,
      "learning_rate": 8.776425872287087e-06,
      "loss": 0.5494,
      "step": 3957
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.08198881149292,
      "learning_rate": 8.775753360326635e-06,
      "loss": 0.5845,
      "step": 3958
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0469781160354614,
      "learning_rate": 8.775080689382342e-06,
      "loss": 0.5745,
      "step": 3959
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9827989935874939,
      "learning_rate": 8.774407859482537e-06,
      "loss": 0.5656,
      "step": 3960
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9459227919578552,
      "learning_rate": 8.773734870655544e-06,
      "loss": 0.5435,
      "step": 3961
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0413237810134888,
      "learning_rate": 8.773061722929704e-06,
      "loss": 0.5485,
      "step": 3962
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9192708730697632,
      "learning_rate": 8.772388416333361e-06,
      "loss": 0.4988,
      "step": 3963
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0438575744628906,
      "learning_rate": 8.771714950894865e-06,
      "loss": 0.5377,
      "step": 3964
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0168156623840332,
      "learning_rate": 8.771041326642572e-06,
      "loss": 0.5476,
      "step": 3965
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9964485168457031,
      "learning_rate": 8.770367543604849e-06,
      "loss": 0.5807,
      "step": 3966
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0404568910598755,
      "learning_rate": 8.769693601810066e-06,
      "loss": 0.5739,
      "step": 3967
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0175347328186035,
      "learning_rate": 8.769019501286598e-06,
      "loss": 0.5346,
      "step": 3968
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9991718530654907,
      "learning_rate": 8.768345242062828e-06,
      "loss": 0.5996,
      "step": 3969
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0009833574295044,
      "learning_rate": 8.767670824167151e-06,
      "loss": 0.6081,
      "step": 3970
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9895312786102295,
      "learning_rate": 8.766996247627963e-06,
      "loss": 0.5379,
      "step": 3971
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9684005379676819,
      "learning_rate": 8.766321512473666e-06,
      "loss": 0.5624,
      "step": 3972
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9689050912857056,
      "learning_rate": 8.765646618732672e-06,
      "loss": 0.5462,
      "step": 3973
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0098981857299805,
      "learning_rate": 8.7649715664334e-06,
      "loss": 0.5476,
      "step": 3974
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0482535362243652,
      "learning_rate": 8.764296355604273e-06,
      "loss": 0.5777,
      "step": 3975
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9891552329063416,
      "learning_rate": 8.76362098627372e-06,
      "loss": 0.5396,
      "step": 3976
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1055810451507568,
      "learning_rate": 8.76294545847018e-06,
      "loss": 0.6089,
      "step": 3977
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0154608488082886,
      "learning_rate": 8.762269772222099e-06,
      "loss": 0.5323,
      "step": 3978
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0198290348052979,
      "learning_rate": 8.761593927557923e-06,
      "loss": 0.5662,
      "step": 3979
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.971450924873352,
      "learning_rate": 8.760917924506114e-06,
      "loss": 0.5111,
      "step": 3980
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1433659791946411,
      "learning_rate": 8.760241763095135e-06,
      "loss": 0.5972,
      "step": 3981
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1008565425872803,
      "learning_rate": 8.759565443353454e-06,
      "loss": 0.5831,
      "step": 3982
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2331944704055786,
      "learning_rate": 8.758888965309554e-06,
      "loss": 0.5825,
      "step": 3983
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0712262392044067,
      "learning_rate": 8.758212328991913e-06,
      "loss": 0.5521,
      "step": 3984
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0214992761611938,
      "learning_rate": 8.757535534429027e-06,
      "loss": 0.5681,
      "step": 3985
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.002253532409668,
      "learning_rate": 8.756858581649391e-06,
      "loss": 0.5614,
      "step": 3986
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9840101599693298,
      "learning_rate": 8.756181470681507e-06,
      "loss": 0.5775,
      "step": 3987
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0549638271331787,
      "learning_rate": 8.755504201553889e-06,
      "loss": 0.5666,
      "step": 3988
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0759743452072144,
      "learning_rate": 8.754826774295056e-06,
      "loss": 0.5767,
      "step": 3989
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0006356239318848,
      "learning_rate": 8.754149188933527e-06,
      "loss": 0.532,
      "step": 3990
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.030730128288269,
      "learning_rate": 8.753471445497837e-06,
      "loss": 0.4923,
      "step": 3991
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9614889025688171,
      "learning_rate": 8.752793544016519e-06,
      "loss": 0.5366,
      "step": 3992
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9809851050376892,
      "learning_rate": 8.752115484518123e-06,
      "loss": 0.5864,
      "step": 3993
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1068106889724731,
      "learning_rate": 8.751437267031194e-06,
      "loss": 0.5662,
      "step": 3994
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0743038654327393,
      "learning_rate": 8.750758891584293e-06,
      "loss": 0.5702,
      "step": 3995
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9942998290061951,
      "learning_rate": 8.750080358205983e-06,
      "loss": 0.5074,
      "step": 3996
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.923958957195282,
      "learning_rate": 8.749401666924834e-06,
      "loss": 0.4981,
      "step": 3997
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.925460934638977,
      "learning_rate": 8.748722817769426e-06,
      "loss": 0.5518,
      "step": 3998
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0104377269744873,
      "learning_rate": 8.74804381076834e-06,
      "loss": 0.5347,
      "step": 3999
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0946342945098877,
      "learning_rate": 8.747364645950168e-06,
      "loss": 0.5968,
      "step": 4000
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0404244661331177,
      "learning_rate": 8.746685323343507e-06,
      "loss": 0.5761,
      "step": 4001
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9306938648223877,
      "learning_rate": 8.74600584297696e-06,
      "loss": 0.5315,
      "step": 4002
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0647470951080322,
      "learning_rate": 8.745326204879139e-06,
      "loss": 0.5447,
      "step": 4003
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9752390384674072,
      "learning_rate": 8.74464640907866e-06,
      "loss": 0.5313,
      "step": 4004
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1666275262832642,
      "learning_rate": 8.743966455604147e-06,
      "loss": 0.5659,
      "step": 4005
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0324372053146362,
      "learning_rate": 8.743286344484232e-06,
      "loss": 0.4939,
      "step": 4006
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.951132595539093,
      "learning_rate": 8.74260607574755e-06,
      "loss": 0.507,
      "step": 4007
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0060588121414185,
      "learning_rate": 8.741925649422746e-06,
      "loss": 0.5316,
      "step": 4008
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0284314155578613,
      "learning_rate": 8.741245065538471e-06,
      "loss": 0.5578,
      "step": 4009
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.935452401638031,
      "learning_rate": 8.74056432412338e-06,
      "loss": 0.5039,
      "step": 4010
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9619759917259216,
      "learning_rate": 8.739883425206138e-06,
      "loss": 0.5432,
      "step": 4011
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0670369863510132,
      "learning_rate": 8.739202368815416e-06,
      "loss": 0.5651,
      "step": 4012
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0108201503753662,
      "learning_rate": 8.738521154979889e-06,
      "loss": 0.5562,
      "step": 4013
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9825449585914612,
      "learning_rate": 8.737839783728242e-06,
      "loss": 0.5479,
      "step": 4014
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9974722862243652,
      "learning_rate": 8.737158255089164e-06,
      "loss": 0.5501,
      "step": 4015
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9536095261573792,
      "learning_rate": 8.736476569091352e-06,
      "loss": 0.5288,
      "step": 4016
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9635430574417114,
      "learning_rate": 8.735794725763512e-06,
      "loss": 0.5658,
      "step": 4017
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.066859245300293,
      "learning_rate": 8.735112725134352e-06,
      "loss": 0.5349,
      "step": 4018
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0659446716308594,
      "learning_rate": 8.734430567232585e-06,
      "loss": 0.5434,
      "step": 4019
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9656541347503662,
      "learning_rate": 8.733748252086943e-06,
      "loss": 0.5587,
      "step": 4020
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0233099460601807,
      "learning_rate": 8.733065779726146e-06,
      "loss": 0.6173,
      "step": 4021
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9754229187965393,
      "learning_rate": 8.732383150178938e-06,
      "loss": 0.5712,
      "step": 4022
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0898586511611938,
      "learning_rate": 8.73170036347406e-06,
      "loss": 0.6084,
      "step": 4023
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0394037961959839,
      "learning_rate": 8.731017419640261e-06,
      "loss": 0.5765,
      "step": 4024
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0213582515716553,
      "learning_rate": 8.730334318706297e-06,
      "loss": 0.5323,
      "step": 4025
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0330605506896973,
      "learning_rate": 8.729651060700932e-06,
      "loss": 0.5812,
      "step": 4026
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9591134786605835,
      "learning_rate": 8.728967645652936e-06,
      "loss": 0.5151,
      "step": 4027
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9901994466781616,
      "learning_rate": 8.728284073591083e-06,
      "loss": 0.5721,
      "step": 4028
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.09280264377594,
      "learning_rate": 8.727600344544159e-06,
      "loss": 0.5826,
      "step": 4029
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9844223260879517,
      "learning_rate": 8.72691645854095e-06,
      "loss": 0.5586,
      "step": 4030
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.013307809829712,
      "learning_rate": 8.726232415610257e-06,
      "loss": 0.5962,
      "step": 4031
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.01223623752594,
      "learning_rate": 8.725548215780877e-06,
      "loss": 0.5316,
      "step": 4032
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.052016019821167,
      "learning_rate": 8.724863859081622e-06,
      "loss": 0.5531,
      "step": 4033
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9758538007736206,
      "learning_rate": 8.724179345541308e-06,
      "loss": 0.5258,
      "step": 4034
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9941028356552124,
      "learning_rate": 8.72349467518876e-06,
      "loss": 0.6022,
      "step": 4035
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9477452039718628,
      "learning_rate": 8.7228098480528e-06,
      "loss": 0.5549,
      "step": 4036
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0785270929336548,
      "learning_rate": 8.72212486416227e-06,
      "loss": 0.5849,
      "step": 4037
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1175118684768677,
      "learning_rate": 8.721439723546012e-06,
      "loss": 0.6015,
      "step": 4038
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0209541320800781,
      "learning_rate": 8.720754426232871e-06,
      "loss": 0.5323,
      "step": 4039
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0251224040985107,
      "learning_rate": 8.720068972251705e-06,
      "loss": 0.5495,
      "step": 4040
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9551291465759277,
      "learning_rate": 8.719383361631376e-06,
      "loss": 0.5908,
      "step": 4041
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0323892831802368,
      "learning_rate": 8.718697594400753e-06,
      "loss": 0.5788,
      "step": 4042
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9632778167724609,
      "learning_rate": 8.71801167058871e-06,
      "loss": 0.5492,
      "step": 4043
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.033238172531128,
      "learning_rate": 8.717325590224129e-06,
      "loss": 0.5496,
      "step": 4044
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1615618467330933,
      "learning_rate": 8.7166393533359e-06,
      "loss": 0.5668,
      "step": 4045
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0879848003387451,
      "learning_rate": 8.715952959952917e-06,
      "loss": 0.5658,
      "step": 4046
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9767739772796631,
      "learning_rate": 8.715266410104081e-06,
      "loss": 0.5787,
      "step": 4047
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.992115318775177,
      "learning_rate": 8.714579703818301e-06,
      "loss": 0.6004,
      "step": 4048
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0199464559555054,
      "learning_rate": 8.713892841124492e-06,
      "loss": 0.5387,
      "step": 4049
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0499255657196045,
      "learning_rate": 8.713205822051576e-06,
      "loss": 0.5862,
      "step": 4050
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0221071243286133,
      "learning_rate": 8.71251864662848e-06,
      "loss": 0.5519,
      "step": 4051
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.016858696937561,
      "learning_rate": 8.711831314884137e-06,
      "loss": 0.5548,
      "step": 4052
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9993886947631836,
      "learning_rate": 8.711143826847491e-06,
      "loss": 0.5586,
      "step": 4053
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9646019339561462,
      "learning_rate": 8.71045618254749e-06,
      "loss": 0.5213,
      "step": 4054
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0232441425323486,
      "learning_rate": 8.709768382013084e-06,
      "loss": 0.5256,
      "step": 4055
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1045476198196411,
      "learning_rate": 8.709080425273238e-06,
      "loss": 0.5761,
      "step": 4056
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0643750429153442,
      "learning_rate": 8.708392312356919e-06,
      "loss": 0.5942,
      "step": 4057
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1480286121368408,
      "learning_rate": 8.7077040432931e-06,
      "loss": 0.6403,
      "step": 4058
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0662150382995605,
      "learning_rate": 8.707015618110761e-06,
      "loss": 0.5633,
      "step": 4059
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0232387781143188,
      "learning_rate": 8.706327036838891e-06,
      "loss": 0.5391,
      "step": 4060
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0814896821975708,
      "learning_rate": 8.705638299506482e-06,
      "loss": 0.5135,
      "step": 4061
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1783430576324463,
      "learning_rate": 8.704949406142536e-06,
      "loss": 0.5899,
      "step": 4062
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0899817943572998,
      "learning_rate": 8.70426035677606e-06,
      "loss": 0.6027,
      "step": 4063
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9908663034439087,
      "learning_rate": 8.703571151436064e-06,
      "loss": 0.4875,
      "step": 4064
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1074823141098022,
      "learning_rate": 8.702881790151572e-06,
      "loss": 0.5187,
      "step": 4065
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0565532445907593,
      "learning_rate": 8.70219227295161e-06,
      "loss": 0.6422,
      "step": 4066
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9629092216491699,
      "learning_rate": 8.70150259986521e-06,
      "loss": 0.5202,
      "step": 4067
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0577901601791382,
      "learning_rate": 8.70081277092141e-06,
      "loss": 0.5438,
      "step": 4068
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9888391494750977,
      "learning_rate": 8.700122786149261e-06,
      "loss": 0.5352,
      "step": 4069
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0722532272338867,
      "learning_rate": 8.699432645577812e-06,
      "loss": 0.622,
      "step": 4070
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9681339263916016,
      "learning_rate": 8.698742349236124e-06,
      "loss": 0.5098,
      "step": 4071
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0090398788452148,
      "learning_rate": 8.698051897153264e-06,
      "loss": 0.5901,
      "step": 4072
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0440917015075684,
      "learning_rate": 8.697361289358302e-06,
      "loss": 0.5435,
      "step": 4073
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1044894456863403,
      "learning_rate": 8.696670525880318e-06,
      "loss": 0.5635,
      "step": 4074
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0052359104156494,
      "learning_rate": 8.695979606748398e-06,
      "loss": 0.5303,
      "step": 4075
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1398298740386963,
      "learning_rate": 8.695288531991633e-06,
      "loss": 0.5506,
      "step": 4076
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0224276781082153,
      "learning_rate": 8.694597301639125e-06,
      "loss": 0.5696,
      "step": 4077
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1038308143615723,
      "learning_rate": 8.693905915719976e-06,
      "loss": 0.5572,
      "step": 4078
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.01349937915802,
      "learning_rate": 8.693214374263298e-06,
      "loss": 0.5485,
      "step": 4079
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0743470191955566,
      "learning_rate": 8.692522677298213e-06,
      "loss": 0.5594,
      "step": 4080
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9773702621459961,
      "learning_rate": 8.691830824853843e-06,
      "loss": 0.5573,
      "step": 4081
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9347333312034607,
      "learning_rate": 8.691138816959318e-06,
      "loss": 0.566,
      "step": 4082
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0899261236190796,
      "learning_rate": 8.690446653643778e-06,
      "loss": 0.5562,
      "step": 4083
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.035556674003601,
      "learning_rate": 8.68975433493637e-06,
      "loss": 0.5625,
      "step": 4084
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.104180932044983,
      "learning_rate": 8.689061860866242e-06,
      "loss": 0.5346,
      "step": 4085
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0645705461502075,
      "learning_rate": 8.68836923146255e-06,
      "loss": 0.6061,
      "step": 4086
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0560028553009033,
      "learning_rate": 8.687676446754464e-06,
      "loss": 0.5779,
      "step": 4087
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.142768383026123,
      "learning_rate": 8.686983506771149e-06,
      "loss": 0.5514,
      "step": 4088
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9595299959182739,
      "learning_rate": 8.686290411541785e-06,
      "loss": 0.5195,
      "step": 4089
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0044152736663818,
      "learning_rate": 8.685597161095555e-06,
      "loss": 0.5095,
      "step": 4090
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9950119256973267,
      "learning_rate": 8.68490375546165e-06,
      "loss": 0.538,
      "step": 4091
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0089884996414185,
      "learning_rate": 8.684210194669269e-06,
      "loss": 0.5648,
      "step": 4092
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1214574575424194,
      "learning_rate": 8.68351647874761e-06,
      "loss": 0.5571,
      "step": 4093
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0534110069274902,
      "learning_rate": 8.682822607725887e-06,
      "loss": 0.5782,
      "step": 4094
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0555564165115356,
      "learning_rate": 8.682128581633316e-06,
      "loss": 0.5501,
      "step": 4095
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9699358344078064,
      "learning_rate": 8.68143440049912e-06,
      "loss": 0.552,
      "step": 4096
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.959537148475647,
      "learning_rate": 8.68074006435253e-06,
      "loss": 0.5549,
      "step": 4097
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.134029746055603,
      "learning_rate": 8.680045573222776e-06,
      "loss": 0.6056,
      "step": 4098
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0371557474136353,
      "learning_rate": 8.679350927139108e-06,
      "loss": 0.5999,
      "step": 4099
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9098154306411743,
      "learning_rate": 8.678656126130768e-06,
      "loss": 0.5404,
      "step": 4100
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.038533329963684,
      "learning_rate": 8.677961170227021e-06,
      "loss": 0.5811,
      "step": 4101
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9490874409675598,
      "learning_rate": 8.677266059457121e-06,
      "loss": 0.5481,
      "step": 4102
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.037155270576477,
      "learning_rate": 8.67657079385034e-06,
      "loss": 0.5693,
      "step": 4103
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.065422773361206,
      "learning_rate": 8.675875373435951e-06,
      "loss": 0.5286,
      "step": 4104
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0294665098190308,
      "learning_rate": 8.67517979824324e-06,
      "loss": 0.5112,
      "step": 4105
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0523234605789185,
      "learning_rate": 8.674484068301492e-06,
      "loss": 0.5605,
      "step": 4106
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.013358473777771,
      "learning_rate": 8.673788183640001e-06,
      "loss": 0.6047,
      "step": 4107
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1127411127090454,
      "learning_rate": 8.673092144288071e-06,
      "loss": 0.5797,
      "step": 4108
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.998127818107605,
      "learning_rate": 8.672395950275008e-06,
      "loss": 0.5816,
      "step": 4109
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.053147554397583,
      "learning_rate": 8.671699601630127e-06,
      "loss": 0.5526,
      "step": 4110
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9740100502967834,
      "learning_rate": 8.67100309838275e-06,
      "loss": 0.5408,
      "step": 4111
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.016816258430481,
      "learning_rate": 8.670306440562202e-06,
      "loss": 0.5728,
      "step": 4112
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9359219074249268,
      "learning_rate": 8.669609628197817e-06,
      "loss": 0.517,
      "step": 4113
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0418291091918945,
      "learning_rate": 8.668912661318938e-06,
      "loss": 0.5534,
      "step": 4114
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0207910537719727,
      "learning_rate": 8.66821553995491e-06,
      "loss": 0.5825,
      "step": 4115
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9650876522064209,
      "learning_rate": 8.667518264135085e-06,
      "loss": 0.5182,
      "step": 4116
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0291123390197754,
      "learning_rate": 8.666820833888825e-06,
      "loss": 0.5197,
      "step": 4117
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.034320592880249,
      "learning_rate": 8.666123249245495e-06,
      "loss": 0.5744,
      "step": 4118
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9790547490119934,
      "learning_rate": 8.665425510234469e-06,
      "loss": 0.5938,
      "step": 4119
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.018493413925171,
      "learning_rate": 8.664727616885126e-06,
      "loss": 0.5469,
      "step": 4120
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.008954405784607,
      "learning_rate": 8.66402956922685e-06,
      "loss": 0.5409,
      "step": 4121
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0612338781356812,
      "learning_rate": 8.663331367289038e-06,
      "loss": 0.6004,
      "step": 4122
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.946708083152771,
      "learning_rate": 8.662633011101084e-06,
      "loss": 0.5077,
      "step": 4123
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9556915760040283,
      "learning_rate": 8.661934500692395e-06,
      "loss": 0.5697,
      "step": 4124
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9655421376228333,
      "learning_rate": 8.661235836092385e-06,
      "loss": 0.6102,
      "step": 4125
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9389283061027527,
      "learning_rate": 8.660537017330468e-06,
      "loss": 0.5537,
      "step": 4126
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.082326054573059,
      "learning_rate": 8.659838044436074e-06,
      "loss": 0.5764,
      "step": 4127
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9879428744316101,
      "learning_rate": 8.65913891743863e-06,
      "loss": 0.4938,
      "step": 4128
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9940571784973145,
      "learning_rate": 8.658439636367574e-06,
      "loss": 0.5274,
      "step": 4129
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0283193588256836,
      "learning_rate": 8.657740201252353e-06,
      "loss": 0.6089,
      "step": 4130
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1298843622207642,
      "learning_rate": 8.657040612122418e-06,
      "loss": 0.5857,
      "step": 4131
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1527552604675293,
      "learning_rate": 8.656340869007225e-06,
      "loss": 0.6263,
      "step": 4132
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0959068536758423,
      "learning_rate": 8.655640971936236e-06,
      "loss": 0.5546,
      "step": 4133
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0256848335266113,
      "learning_rate": 8.654940920938922e-06,
      "loss": 0.5583,
      "step": 4134
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.949115514755249,
      "learning_rate": 8.654240716044762e-06,
      "loss": 0.5556,
      "step": 4135
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.077285885810852,
      "learning_rate": 8.653540357283236e-06,
      "loss": 0.5281,
      "step": 4136
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0292668342590332,
      "learning_rate": 8.652839844683836e-06,
      "loss": 0.5358,
      "step": 4137
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9717956185340881,
      "learning_rate": 8.652139178276058e-06,
      "loss": 0.5689,
      "step": 4138
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0240721702575684,
      "learning_rate": 8.651438358089403e-06,
      "loss": 0.5914,
      "step": 4139
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0816338062286377,
      "learning_rate": 8.650737384153382e-06,
      "loss": 0.6093,
      "step": 4140
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1217504739761353,
      "learning_rate": 8.65003625649751e-06,
      "loss": 0.5252,
      "step": 4141
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0407133102416992,
      "learning_rate": 8.649334975151307e-06,
      "loss": 0.567,
      "step": 4142
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0991817712783813,
      "learning_rate": 8.648633540144304e-06,
      "loss": 0.5795,
      "step": 4143
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9913834929466248,
      "learning_rate": 8.647931951506037e-06,
      "loss": 0.5412,
      "step": 4144
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0170899629592896,
      "learning_rate": 8.647230209266043e-06,
      "loss": 0.564,
      "step": 4145
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0658817291259766,
      "learning_rate": 8.646528313453876e-06,
      "loss": 0.5481,
      "step": 4146
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9929413795471191,
      "learning_rate": 8.645826264099085e-06,
      "loss": 0.5121,
      "step": 4147
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0202983617782593,
      "learning_rate": 8.645124061231234e-06,
      "loss": 0.5619,
      "step": 4148
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0099726915359497,
      "learning_rate": 8.644421704879889e-06,
      "loss": 0.4819,
      "step": 4149
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9748672842979431,
      "learning_rate": 8.643719195074622e-06,
      "loss": 0.535,
      "step": 4150
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0206111669540405,
      "learning_rate": 8.643016531845017e-06,
      "loss": 0.5487,
      "step": 4151
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2215276956558228,
      "learning_rate": 8.642313715220659e-06,
      "loss": 0.5484,
      "step": 4152
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9748323559761047,
      "learning_rate": 8.641610745231142e-06,
      "loss": 0.5357,
      "step": 4153
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0382490158081055,
      "learning_rate": 8.640907621906062e-06,
      "loss": 0.5182,
      "step": 4154
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0930920839309692,
      "learning_rate": 8.640204345275029e-06,
      "loss": 0.5927,
      "step": 4155
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.106478214263916,
      "learning_rate": 8.639500915367656e-06,
      "loss": 0.5741,
      "step": 4156
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0560954809188843,
      "learning_rate": 8.63879733221356e-06,
      "loss": 0.5941,
      "step": 4157
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0674949884414673,
      "learning_rate": 8.638093595842366e-06,
      "loss": 0.5958,
      "step": 4158
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.994944155216217,
      "learning_rate": 8.637389706283705e-06,
      "loss": 0.5292,
      "step": 4159
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.093483328819275,
      "learning_rate": 8.636685663567219e-06,
      "loss": 0.5908,
      "step": 4160
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9695919156074524,
      "learning_rate": 8.635981467722552e-06,
      "loss": 0.5486,
      "step": 4161
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9622272253036499,
      "learning_rate": 8.635277118779353e-06,
      "loss": 0.5281,
      "step": 4162
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9899247288703918,
      "learning_rate": 8.63457261676728e-06,
      "loss": 0.5503,
      "step": 4163
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0092194080352783,
      "learning_rate": 8.633867961715998e-06,
      "loss": 0.5343,
      "step": 4164
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0169821977615356,
      "learning_rate": 8.633163153655178e-06,
      "loss": 0.5738,
      "step": 4165
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0969207286834717,
      "learning_rate": 8.632458192614495e-06,
      "loss": 0.6267,
      "step": 4166
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0135194063186646,
      "learning_rate": 8.631753078623635e-06,
      "loss": 0.5283,
      "step": 4167
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0597193241119385,
      "learning_rate": 8.631047811712288e-06,
      "loss": 0.6067,
      "step": 4168
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0797663927078247,
      "learning_rate": 8.630342391910147e-06,
      "loss": 0.594,
      "step": 4169
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0816011428833008,
      "learning_rate": 8.629636819246919e-06,
      "loss": 0.5519,
      "step": 4170
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.047973871231079,
      "learning_rate": 8.628931093752308e-06,
      "loss": 0.5402,
      "step": 4171
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0002282857894897,
      "learning_rate": 8.628225215456037e-06,
      "loss": 0.5496,
      "step": 4172
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0012603998184204,
      "learning_rate": 8.627519184387821e-06,
      "loss": 0.5941,
      "step": 4173
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0786024332046509,
      "learning_rate": 8.626813000577393e-06,
      "loss": 0.5951,
      "step": 4174
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1156187057495117,
      "learning_rate": 8.626106664054483e-06,
      "loss": 0.5681,
      "step": 4175
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0088963508605957,
      "learning_rate": 8.62540017484884e-06,
      "loss": 0.5534,
      "step": 4176
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.043418526649475,
      "learning_rate": 8.624693532990205e-06,
      "loss": 0.5366,
      "step": 4177
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1141891479492188,
      "learning_rate": 8.623986738508334e-06,
      "loss": 0.5752,
      "step": 4178
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0138537883758545,
      "learning_rate": 8.62327979143299e-06,
      "loss": 0.537,
      "step": 4179
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9940417408943176,
      "learning_rate": 8.622572691793937e-06,
      "loss": 0.5016,
      "step": 4180
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0246286392211914,
      "learning_rate": 8.621865439620952e-06,
      "loss": 0.5365,
      "step": 4181
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9923083782196045,
      "learning_rate": 8.621158034943812e-06,
      "loss": 0.544,
      "step": 4182
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9740118384361267,
      "learning_rate": 8.620450477792303e-06,
      "loss": 0.5447,
      "step": 4183
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0033918619155884,
      "learning_rate": 8.619742768196221e-06,
      "loss": 0.522,
      "step": 4184
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0502849817276,
      "learning_rate": 8.619034906185362e-06,
      "loss": 0.6178,
      "step": 4185
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0749516487121582,
      "learning_rate": 8.618326891789534e-06,
      "loss": 0.5715,
      "step": 4186
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1323692798614502,
      "learning_rate": 8.617618725038545e-06,
      "loss": 0.5474,
      "step": 4187
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0600783824920654,
      "learning_rate": 8.61691040596222e-06,
      "loss": 0.5766,
      "step": 4188
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0678776502609253,
      "learning_rate": 8.616201934590379e-06,
      "loss": 0.5907,
      "step": 4189
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1253869533538818,
      "learning_rate": 8.615493310952852e-06,
      "loss": 0.5871,
      "step": 4190
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9809844493865967,
      "learning_rate": 8.614784535079482e-06,
      "loss": 0.5472,
      "step": 4191
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0507099628448486,
      "learning_rate": 8.614075607000108e-06,
      "loss": 0.5836,
      "step": 4192
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0916355848312378,
      "learning_rate": 8.613366526744584e-06,
      "loss": 0.5245,
      "step": 4193
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0502519607543945,
      "learning_rate": 8.612657294342765e-06,
      "loss": 0.5358,
      "step": 4194
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0014053583145142,
      "learning_rate": 8.611947909824514e-06,
      "loss": 0.5794,
      "step": 4195
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9775264263153076,
      "learning_rate": 8.611238373219703e-06,
      "loss": 0.5321,
      "step": 4196
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1208726167678833,
      "learning_rate": 8.610528684558206e-06,
      "loss": 0.5351,
      "step": 4197
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0999410152435303,
      "learning_rate": 8.609818843869907e-06,
      "loss": 0.5385,
      "step": 4198
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9840497374534607,
      "learning_rate": 8.609108851184693e-06,
      "loss": 0.5264,
      "step": 4199
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9287744760513306,
      "learning_rate": 8.608398706532462e-06,
      "loss": 0.5643,
      "step": 4200
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9510626792907715,
      "learning_rate": 8.607688409943112e-06,
      "loss": 0.5107,
      "step": 4201
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9852222204208374,
      "learning_rate": 8.606977961446554e-06,
      "loss": 0.5483,
      "step": 4202
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9964022636413574,
      "learning_rate": 8.606267361072704e-06,
      "loss": 0.5655,
      "step": 4203
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9797468781471252,
      "learning_rate": 8.605556608851478e-06,
      "loss": 0.5991,
      "step": 4204
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.094444990158081,
      "learning_rate": 8.604845704812808e-06,
      "loss": 0.5936,
      "step": 4205
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0043718814849854,
      "learning_rate": 8.604134648986625e-06,
      "loss": 0.5328,
      "step": 4206
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0017262697219849,
      "learning_rate": 8.603423441402868e-06,
      "loss": 0.558,
      "step": 4207
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0760914087295532,
      "learning_rate": 8.602712082091487e-06,
      "loss": 0.5093,
      "step": 4208
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0324655771255493,
      "learning_rate": 8.602000571082432e-06,
      "loss": 0.5664,
      "step": 4209
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9881264567375183,
      "learning_rate": 8.601288908405665e-06,
      "loss": 0.5565,
      "step": 4210
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9853984713554382,
      "learning_rate": 8.60057709409115e-06,
      "loss": 0.5614,
      "step": 4211
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9987500309944153,
      "learning_rate": 8.599865128168858e-06,
      "loss": 0.5252,
      "step": 4212
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9709988832473755,
      "learning_rate": 8.599153010668768e-06,
      "loss": 0.5274,
      "step": 4213
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.00737464427948,
      "learning_rate": 8.598440741620868e-06,
      "loss": 0.5745,
      "step": 4214
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9636495113372803,
      "learning_rate": 8.597728321055144e-06,
      "loss": 0.5178,
      "step": 4215
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9868621230125427,
      "learning_rate": 8.597015749001596e-06,
      "loss": 0.5453,
      "step": 4216
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9736589193344116,
      "learning_rate": 8.59630302549023e-06,
      "loss": 0.5534,
      "step": 4217
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.090067744255066,
      "learning_rate": 8.595590150551052e-06,
      "loss": 0.5462,
      "step": 4218
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1201854944229126,
      "learning_rate": 8.59487712421408e-06,
      "loss": 0.553,
      "step": 4219
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.06673264503479,
      "learning_rate": 8.594163946509339e-06,
      "loss": 0.5585,
      "step": 4220
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0092447996139526,
      "learning_rate": 8.593450617466859e-06,
      "loss": 0.5357,
      "step": 4221
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.069238305091858,
      "learning_rate": 8.592737137116673e-06,
      "loss": 0.5405,
      "step": 4222
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0110065937042236,
      "learning_rate": 8.592023505488825e-06,
      "loss": 0.5798,
      "step": 4223
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8846083283424377,
      "learning_rate": 8.591309722613362e-06,
      "loss": 0.5074,
      "step": 4224
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9890139102935791,
      "learning_rate": 8.590595788520342e-06,
      "loss": 0.517,
      "step": 4225
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9960560202598572,
      "learning_rate": 8.589881703239821e-06,
      "loss": 0.4919,
      "step": 4226
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9747819900512695,
      "learning_rate": 8.58916746680187e-06,
      "loss": 0.544,
      "step": 4227
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0688061714172363,
      "learning_rate": 8.588453079236565e-06,
      "loss": 0.6009,
      "step": 4228
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0313738584518433,
      "learning_rate": 8.587738540573984e-06,
      "loss": 0.5276,
      "step": 4229
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9333264231681824,
      "learning_rate": 8.587023850844212e-06,
      "loss": 0.5402,
      "step": 4230
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0281132459640503,
      "learning_rate": 8.586309010077345e-06,
      "loss": 0.6009,
      "step": 4231
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.923550009727478,
      "learning_rate": 8.585594018303482e-06,
      "loss": 0.5505,
      "step": 4232
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9499755501747131,
      "learning_rate": 8.584878875552727e-06,
      "loss": 0.5427,
      "step": 4233
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9870060086250305,
      "learning_rate": 8.584163581855194e-06,
      "loss": 0.5592,
      "step": 4234
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0823084115982056,
      "learning_rate": 8.583448137241002e-06,
      "loss": 0.5651,
      "step": 4235
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0290069580078125,
      "learning_rate": 8.582732541740273e-06,
      "loss": 0.5512,
      "step": 4236
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9916083812713623,
      "learning_rate": 8.582016795383142e-06,
      "loss": 0.5345,
      "step": 4237
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.997864305973053,
      "learning_rate": 8.581300898199743e-06,
      "loss": 0.5139,
      "step": 4238
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0194896459579468,
      "learning_rate": 8.580584850220222e-06,
      "loss": 0.5385,
      "step": 4239
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0112195014953613,
      "learning_rate": 8.57986865147473e-06,
      "loss": 0.5308,
      "step": 4240
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0058863162994385,
      "learning_rate": 8.57915230199342e-06,
      "loss": 0.5421,
      "step": 4241
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0418747663497925,
      "learning_rate": 8.578435801806461e-06,
      "loss": 0.5858,
      "step": 4242
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9853035807609558,
      "learning_rate": 8.577719150944017e-06,
      "loss": 0.5623,
      "step": 4243
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0543568134307861,
      "learning_rate": 8.577002349436264e-06,
      "loss": 0.5407,
      "step": 4244
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1742209196090698,
      "learning_rate": 8.57628539731339e-06,
      "loss": 0.6039,
      "step": 4245
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9908335208892822,
      "learning_rate": 8.575568294605574e-06,
      "loss": 0.589,
      "step": 4246
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0744788646697998,
      "learning_rate": 8.574851041343018e-06,
      "loss": 0.5337,
      "step": 4247
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9883583188056946,
      "learning_rate": 8.574133637555921e-06,
      "loss": 0.5498,
      "step": 4248
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9712103009223938,
      "learning_rate": 8.57341608327449e-06,
      "loss": 0.5297,
      "step": 4249
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.036982774734497,
      "learning_rate": 8.572698378528937e-06,
      "loss": 0.5845,
      "step": 4250
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9253586530685425,
      "learning_rate": 8.571980523349485e-06,
      "loss": 0.5515,
      "step": 4251
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1356269121170044,
      "learning_rate": 8.57126251776636e-06,
      "loss": 0.593,
      "step": 4252
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0142786502838135,
      "learning_rate": 8.570544361809792e-06,
      "loss": 0.6091,
      "step": 4253
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9978875517845154,
      "learning_rate": 8.569826055510025e-06,
      "loss": 0.5441,
      "step": 4254
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9625915288925171,
      "learning_rate": 8.569107598897296e-06,
      "loss": 0.4728,
      "step": 4255
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1040743589401245,
      "learning_rate": 8.568388992001868e-06,
      "loss": 0.5302,
      "step": 4256
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0358699560165405,
      "learning_rate": 8.56767023485399e-06,
      "loss": 0.5349,
      "step": 4257
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.031915545463562,
      "learning_rate": 8.56695132748393e-06,
      "loss": 0.591,
      "step": 4258
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0375105142593384,
      "learning_rate": 8.566232269921957e-06,
      "loss": 0.6325,
      "step": 4259
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9367465972900391,
      "learning_rate": 8.565513062198351e-06,
      "loss": 0.5572,
      "step": 4260
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9881364107131958,
      "learning_rate": 8.564793704343392e-06,
      "loss": 0.511,
      "step": 4261
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3143302202224731,
      "learning_rate": 8.564074196387371e-06,
      "loss": 0.6167,
      "step": 4262
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9849549531936646,
      "learning_rate": 8.563354538360585e-06,
      "loss": 0.5401,
      "step": 4263
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9980334043502808,
      "learning_rate": 8.562634730293335e-06,
      "loss": 0.537,
      "step": 4264
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.069611668586731,
      "learning_rate": 8.56191477221593e-06,
      "loss": 0.5925,
      "step": 4265
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0978347063064575,
      "learning_rate": 8.561194664158685e-06,
      "loss": 0.6103,
      "step": 4266
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1114164590835571,
      "learning_rate": 8.560474406151921e-06,
      "loss": 0.5989,
      "step": 4267
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0410330295562744,
      "learning_rate": 8.559753998225965e-06,
      "loss": 0.5932,
      "step": 4268
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0758030414581299,
      "learning_rate": 8.559033440411155e-06,
      "loss": 0.5336,
      "step": 4269
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9148129820823669,
      "learning_rate": 8.558312732737825e-06,
      "loss": 0.512,
      "step": 4270
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9755415320396423,
      "learning_rate": 8.557591875236323e-06,
      "loss": 0.5372,
      "step": 4271
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9766758680343628,
      "learning_rate": 8.556870867937006e-06,
      "loss": 0.527,
      "step": 4272
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9935266375541687,
      "learning_rate": 8.55614971087023e-06,
      "loss": 0.5456,
      "step": 4273
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0467207431793213,
      "learning_rate": 8.555428404066359e-06,
      "loss": 0.5415,
      "step": 4274
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9002679586410522,
      "learning_rate": 8.554706947555766e-06,
      "loss": 0.5279,
      "step": 4275
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0087589025497437,
      "learning_rate": 8.553985341368832e-06,
      "loss": 0.6003,
      "step": 4276
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8842843770980835,
      "learning_rate": 8.553263585535937e-06,
      "loss": 0.5594,
      "step": 4277
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0302597284317017,
      "learning_rate": 8.552541680087472e-06,
      "loss": 0.5525,
      "step": 4278
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.047193169593811,
      "learning_rate": 8.551819625053837e-06,
      "loss": 0.5872,
      "step": 4279
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0108321905136108,
      "learning_rate": 8.551097420465432e-06,
      "loss": 0.5376,
      "step": 4280
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0369234085083008,
      "learning_rate": 8.55037506635267e-06,
      "loss": 0.5298,
      "step": 4281
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0145535469055176,
      "learning_rate": 8.549652562745963e-06,
      "loss": 0.5513,
      "step": 4282
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0360575914382935,
      "learning_rate": 8.548929909675736e-06,
      "loss": 0.5499,
      "step": 4283
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1149938106536865,
      "learning_rate": 8.548207107172417e-06,
      "loss": 0.5757,
      "step": 4284
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.109709620475769,
      "learning_rate": 8.547484155266439e-06,
      "loss": 0.5682,
      "step": 4285
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0783034563064575,
      "learning_rate": 8.546761053988244e-06,
      "loss": 0.614,
      "step": 4286
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0046306848526,
      "learning_rate": 8.546037803368279e-06,
      "loss": 0.5357,
      "step": 4287
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9678237438201904,
      "learning_rate": 8.545314403436998e-06,
      "loss": 0.506,
      "step": 4288
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1249730587005615,
      "learning_rate": 8.54459085422486e-06,
      "loss": 0.5468,
      "step": 4289
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0614315271377563,
      "learning_rate": 8.543867155762335e-06,
      "loss": 0.5888,
      "step": 4290
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1124675273895264,
      "learning_rate": 8.543143308079888e-06,
      "loss": 0.5672,
      "step": 4291
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.035871148109436,
      "learning_rate": 8.542419311208006e-06,
      "loss": 0.5639,
      "step": 4292
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9853594899177551,
      "learning_rate": 8.541695165177169e-06,
      "loss": 0.5951,
      "step": 4293
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0269336700439453,
      "learning_rate": 8.540970870017867e-06,
      "loss": 0.5504,
      "step": 4294
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9964527487754822,
      "learning_rate": 8.540246425760602e-06,
      "loss": 0.5321,
      "step": 4295
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0525084733963013,
      "learning_rate": 8.539521832435874e-06,
      "loss": 0.5324,
      "step": 4296
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0360336303710938,
      "learning_rate": 8.538797090074196e-06,
      "loss": 0.5566,
      "step": 4297
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0745279788970947,
      "learning_rate": 8.538072198706081e-06,
      "loss": 0.5588,
      "step": 4298
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0289655923843384,
      "learning_rate": 8.537347158362056e-06,
      "loss": 0.5235,
      "step": 4299
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0566655397415161,
      "learning_rate": 8.536621969072648e-06,
      "loss": 0.5395,
      "step": 4300
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.065811038017273,
      "learning_rate": 8.53589663086839e-06,
      "loss": 0.5819,
      "step": 4301
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0649478435516357,
      "learning_rate": 8.535171143779828e-06,
      "loss": 0.5556,
      "step": 4302
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9797233939170837,
      "learning_rate": 8.534445507837505e-06,
      "loss": 0.5467,
      "step": 4303
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0511661767959595,
      "learning_rate": 8.533719723071979e-06,
      "loss": 0.5709,
      "step": 4304
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0210515260696411,
      "learning_rate": 8.532993789513805e-06,
      "loss": 0.5575,
      "step": 4305
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1841431856155396,
      "learning_rate": 8.532267707193555e-06,
      "loss": 0.5748,
      "step": 4306
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0234102010726929,
      "learning_rate": 8.5315414761418e-06,
      "loss": 0.5749,
      "step": 4307
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0281085968017578,
      "learning_rate": 8.530815096389118e-06,
      "loss": 0.5112,
      "step": 4308
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0519088506698608,
      "learning_rate": 8.530088567966095e-06,
      "loss": 0.5671,
      "step": 4309
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9593432545661926,
      "learning_rate": 8.529361890903323e-06,
      "loss": 0.5305,
      "step": 4310
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2273467779159546,
      "learning_rate": 8.5286350652314e-06,
      "loss": 0.5833,
      "step": 4311
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0280483961105347,
      "learning_rate": 8.527908090980929e-06,
      "loss": 0.595,
      "step": 4312
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0453826189041138,
      "learning_rate": 8.527180968182522e-06,
      "loss": 0.5362,
      "step": 4313
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9932584762573242,
      "learning_rate": 8.526453696866794e-06,
      "loss": 0.5371,
      "step": 4314
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9493650197982788,
      "learning_rate": 8.525726277064368e-06,
      "loss": 0.5309,
      "step": 4315
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9597651958465576,
      "learning_rate": 8.524998708805874e-06,
      "loss": 0.5175,
      "step": 4316
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9586458206176758,
      "learning_rate": 8.524270992121948e-06,
      "loss": 0.5555,
      "step": 4317
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.064935564994812,
      "learning_rate": 8.523543127043228e-06,
      "loss": 0.5609,
      "step": 4318
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9935119152069092,
      "learning_rate": 8.522815113600366e-06,
      "loss": 0.5312,
      "step": 4319
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1048222780227661,
      "learning_rate": 8.522086951824014e-06,
      "loss": 0.611,
      "step": 4320
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9754494428634644,
      "learning_rate": 8.521358641744834e-06,
      "loss": 0.5286,
      "step": 4321
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.023982048034668,
      "learning_rate": 8.520630183393492e-06,
      "loss": 0.5568,
      "step": 4322
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0174022912979126,
      "learning_rate": 8.519901576800657e-06,
      "loss": 0.5716,
      "step": 4323
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0983679294586182,
      "learning_rate": 8.519172821997015e-06,
      "loss": 0.5184,
      "step": 4324
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0188732147216797,
      "learning_rate": 8.518443919013247e-06,
      "loss": 0.5623,
      "step": 4325
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9419271945953369,
      "learning_rate": 8.517714867880044e-06,
      "loss": 0.5156,
      "step": 4326
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0496870279312134,
      "learning_rate": 8.516985668628105e-06,
      "loss": 0.4981,
      "step": 4327
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0018783807754517,
      "learning_rate": 8.516256321288136e-06,
      "loss": 0.5079,
      "step": 4328
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0067249536514282,
      "learning_rate": 8.515526825890845e-06,
      "loss": 0.4978,
      "step": 4329
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0570882558822632,
      "learning_rate": 8.514797182466948e-06,
      "loss": 0.532,
      "step": 4330
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1334291696548462,
      "learning_rate": 8.51406739104717e-06,
      "loss": 0.5654,
      "step": 4331
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0686006546020508,
      "learning_rate": 8.513337451662238e-06,
      "loss": 0.5888,
      "step": 4332
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0836188793182373,
      "learning_rate": 8.512607364342887e-06,
      "loss": 0.5513,
      "step": 4333
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0141760110855103,
      "learning_rate": 8.51187712911986e-06,
      "loss": 0.5344,
      "step": 4334
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0270737409591675,
      "learning_rate": 8.511146746023905e-06,
      "loss": 0.5242,
      "step": 4335
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0236823558807373,
      "learning_rate": 8.510416215085775e-06,
      "loss": 0.5676,
      "step": 4336
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9484716653823853,
      "learning_rate": 8.509685536336229e-06,
      "loss": 0.5009,
      "step": 4337
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0540645122528076,
      "learning_rate": 8.508954709806034e-06,
      "loss": 0.5761,
      "step": 4338
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.025154948234558,
      "learning_rate": 8.508223735525963e-06,
      "loss": 0.5623,
      "step": 4339
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0674318075180054,
      "learning_rate": 8.507492613526795e-06,
      "loss": 0.5728,
      "step": 4340
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9796115756034851,
      "learning_rate": 8.506761343839316e-06,
      "loss": 0.5457,
      "step": 4341
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0570836067199707,
      "learning_rate": 8.506029926494315e-06,
      "loss": 0.5681,
      "step": 4342
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0061306953430176,
      "learning_rate": 8.50529836152259e-06,
      "loss": 0.6148,
      "step": 4343
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9319235682487488,
      "learning_rate": 8.504566648954947e-06,
      "loss": 0.5147,
      "step": 4344
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0190160274505615,
      "learning_rate": 8.503834788822191e-06,
      "loss": 0.5375,
      "step": 4345
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9299525022506714,
      "learning_rate": 8.503102781155141e-06,
      "loss": 0.5279,
      "step": 4346
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9546197652816772,
      "learning_rate": 8.502370625984622e-06,
      "loss": 0.529,
      "step": 4347
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.046825885772705,
      "learning_rate": 8.501638323341459e-06,
      "loss": 0.59,
      "step": 4348
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9880673289299011,
      "learning_rate": 8.500905873256486e-06,
      "loss": 0.5394,
      "step": 4349
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1426104307174683,
      "learning_rate": 8.500173275760546e-06,
      "loss": 0.5538,
      "step": 4350
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9619871377944946,
      "learning_rate": 8.499440530884486e-06,
      "loss": 0.5246,
      "step": 4351
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0348474979400635,
      "learning_rate": 8.498707638659159e-06,
      "loss": 0.5367,
      "step": 4352
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9267213344573975,
      "learning_rate": 8.497974599115424e-06,
      "loss": 0.5001,
      "step": 4353
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.062888503074646,
      "learning_rate": 8.497241412284147e-06,
      "loss": 0.5501,
      "step": 4354
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0686131715774536,
      "learning_rate": 8.496508078196202e-06,
      "loss": 0.574,
      "step": 4355
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9900593161582947,
      "learning_rate": 8.495774596882462e-06,
      "loss": 0.5111,
      "step": 4356
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.933822751045227,
      "learning_rate": 8.495040968373817e-06,
      "loss": 0.5655,
      "step": 4357
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.038696050643921,
      "learning_rate": 8.494307192701154e-06,
      "loss": 0.5161,
      "step": 4358
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9698143601417542,
      "learning_rate": 8.493573269895372e-06,
      "loss": 0.5127,
      "step": 4359
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.945016086101532,
      "learning_rate": 8.492839199987373e-06,
      "loss": 0.529,
      "step": 4360
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0088963508605957,
      "learning_rate": 8.492104983008065e-06,
      "loss": 0.5339,
      "step": 4361
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9921668767929077,
      "learning_rate": 8.491370618988367e-06,
      "loss": 0.5159,
      "step": 4362
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0715415477752686,
      "learning_rate": 8.490636107959194e-06,
      "loss": 0.5341,
      "step": 4363
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.040958046913147,
      "learning_rate": 8.489901449951478e-06,
      "loss": 0.5823,
      "step": 4364
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0582530498504639,
      "learning_rate": 8.489166644996154e-06,
      "loss": 0.5688,
      "step": 4365
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1656434535980225,
      "learning_rate": 8.48843169312416e-06,
      "loss": 0.5707,
      "step": 4366
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0881271362304688,
      "learning_rate": 8.487696594366444e-06,
      "loss": 0.5661,
      "step": 4367
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9278864860534668,
      "learning_rate": 8.486961348753954e-06,
      "loss": 0.4697,
      "step": 4368
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.033172607421875,
      "learning_rate": 8.486225956317655e-06,
      "loss": 0.5875,
      "step": 4369
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9845580458641052,
      "learning_rate": 8.48549041708851e-06,
      "loss": 0.526,
      "step": 4370
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9801870584487915,
      "learning_rate": 8.484754731097484e-06,
      "loss": 0.5087,
      "step": 4371
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0618585348129272,
      "learning_rate": 8.484018898375561e-06,
      "loss": 0.5513,
      "step": 4372
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9950966835021973,
      "learning_rate": 8.483282918953723e-06,
      "loss": 0.5884,
      "step": 4373
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.942801296710968,
      "learning_rate": 8.482546792862957e-06,
      "loss": 0.5631,
      "step": 4374
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0492613315582275,
      "learning_rate": 8.481810520134262e-06,
      "loss": 0.5348,
      "step": 4375
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9315921664237976,
      "learning_rate": 8.481074100798638e-06,
      "loss": 0.5685,
      "step": 4376
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0836108922958374,
      "learning_rate": 8.480337534887093e-06,
      "loss": 0.5692,
      "step": 4377
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0537856817245483,
      "learning_rate": 8.479600822430642e-06,
      "loss": 0.5957,
      "step": 4378
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0087220668792725,
      "learning_rate": 8.478863963460306e-06,
      "loss": 0.5093,
      "step": 4379
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0664055347442627,
      "learning_rate": 8.478126958007108e-06,
      "loss": 0.5462,
      "step": 4380
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9335545301437378,
      "learning_rate": 8.477389806102085e-06,
      "loss": 0.5823,
      "step": 4381
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0260354280471802,
      "learning_rate": 8.476652507776274e-06,
      "loss": 0.514,
      "step": 4382
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0136860609054565,
      "learning_rate": 8.475915063060721e-06,
      "loss": 0.5885,
      "step": 4383
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0180233716964722,
      "learning_rate": 8.475177471986476e-06,
      "loss": 0.5295,
      "step": 4384
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0764186382293701,
      "learning_rate": 8.474439734584597e-06,
      "loss": 0.5504,
      "step": 4385
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.021909236907959,
      "learning_rate": 8.473701850886147e-06,
      "loss": 0.5445,
      "step": 4386
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0124695301055908,
      "learning_rate": 8.472963820922195e-06,
      "loss": 0.5058,
      "step": 4387
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0495041608810425,
      "learning_rate": 8.47222564472382e-06,
      "loss": 0.5681,
      "step": 4388
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0147275924682617,
      "learning_rate": 8.471487322322101e-06,
      "loss": 0.5115,
      "step": 4389
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0069338083267212,
      "learning_rate": 8.47074885374813e-06,
      "loss": 0.58,
      "step": 4390
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0648608207702637,
      "learning_rate": 8.470010239032995e-06,
      "loss": 0.5825,
      "step": 4391
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0082082748413086,
      "learning_rate": 8.469271478207801e-06,
      "loss": 0.5018,
      "step": 4392
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9308993816375732,
      "learning_rate": 8.468532571303655e-06,
      "loss": 0.5125,
      "step": 4393
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0136820077896118,
      "learning_rate": 8.467793518351668e-06,
      "loss": 0.5703,
      "step": 4394
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1096161603927612,
      "learning_rate": 8.46705431938296e-06,
      "loss": 0.5736,
      "step": 4395
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9819938540458679,
      "learning_rate": 8.466314974428655e-06,
      "loss": 0.5676,
      "step": 4396
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0787750482559204,
      "learning_rate": 8.465575483519883e-06,
      "loss": 0.5306,
      "step": 4397
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0179063081741333,
      "learning_rate": 8.464835846687786e-06,
      "loss": 0.5864,
      "step": 4398
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9123398065567017,
      "learning_rate": 8.464096063963503e-06,
      "loss": 0.5144,
      "step": 4399
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0560660362243652,
      "learning_rate": 8.463356135378187e-06,
      "loss": 0.5308,
      "step": 4400
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.008315920829773,
      "learning_rate": 8.462616060962992e-06,
      "loss": 0.5408,
      "step": 4401
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.080171823501587,
      "learning_rate": 8.46187584074908e-06,
      "loss": 0.6,
      "step": 4402
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0192162990570068,
      "learning_rate": 8.461135474767618e-06,
      "loss": 0.5419,
      "step": 4403
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1443674564361572,
      "learning_rate": 8.460394963049784e-06,
      "loss": 0.5712,
      "step": 4404
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.009781837463379,
      "learning_rate": 8.459654305626754e-06,
      "loss": 0.5378,
      "step": 4405
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9513635635375977,
      "learning_rate": 8.458913502529718e-06,
      "loss": 0.5402,
      "step": 4406
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.040630578994751,
      "learning_rate": 8.458172553789866e-06,
      "loss": 0.5859,
      "step": 4407
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0000171661376953,
      "learning_rate": 8.457431459438398e-06,
      "loss": 0.5451,
      "step": 4408
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0893827676773071,
      "learning_rate": 8.456690219506519e-06,
      "loss": 0.5255,
      "step": 4409
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0486403703689575,
      "learning_rate": 8.45594883402544e-06,
      "loss": 0.5654,
      "step": 4410
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0393122434616089,
      "learning_rate": 8.455207303026378e-06,
      "loss": 0.5722,
      "step": 4411
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9944487810134888,
      "learning_rate": 8.454465626540555e-06,
      "loss": 0.5326,
      "step": 4412
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0593161582946777,
      "learning_rate": 8.453723804599203e-06,
      "loss": 0.5565,
      "step": 4413
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0495504140853882,
      "learning_rate": 8.452981837233555e-06,
      "loss": 0.5774,
      "step": 4414
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0957008600234985,
      "learning_rate": 8.452239724474856e-06,
      "loss": 0.5729,
      "step": 4415
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.020525574684143,
      "learning_rate": 8.451497466354349e-06,
      "loss": 0.5112,
      "step": 4416
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0432324409484863,
      "learning_rate": 8.450755062903293e-06,
      "loss": 0.5541,
      "step": 4417
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0233687162399292,
      "learning_rate": 8.450012514152943e-06,
      "loss": 0.5548,
      "step": 4418
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9818899631500244,
      "learning_rate": 8.44926982013457e-06,
      "loss": 0.5535,
      "step": 4419
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0102336406707764,
      "learning_rate": 8.448526980879444e-06,
      "loss": 0.5317,
      "step": 4420
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1956027746200562,
      "learning_rate": 8.447783996418843e-06,
      "loss": 0.6164,
      "step": 4421
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0055028200149536,
      "learning_rate": 8.447040866784051e-06,
      "loss": 0.5325,
      "step": 4422
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.037794828414917,
      "learning_rate": 8.446297592006361e-06,
      "loss": 0.4994,
      "step": 4423
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0306740999221802,
      "learning_rate": 8.445554172117066e-06,
      "loss": 0.5552,
      "step": 4424
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0212527513504028,
      "learning_rate": 8.444810607147472e-06,
      "loss": 0.5188,
      "step": 4425
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9898123741149902,
      "learning_rate": 8.444066897128888e-06,
      "loss": 0.516,
      "step": 4426
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4958157539367676,
      "learning_rate": 8.443323042092625e-06,
      "loss": 0.5338,
      "step": 4427
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0361689329147339,
      "learning_rate": 8.442579042070011e-06,
      "loss": 0.5656,
      "step": 4428
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9662915468215942,
      "learning_rate": 8.441834897092366e-06,
      "loss": 0.5384,
      "step": 4429
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1028213500976562,
      "learning_rate": 8.44109060719103e-06,
      "loss": 0.5347,
      "step": 4430
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0823792219161987,
      "learning_rate": 8.440346172397338e-06,
      "loss": 0.6062,
      "step": 4431
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0345669984817505,
      "learning_rate": 8.439601592742637e-06,
      "loss": 0.6027,
      "step": 4432
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0370168685913086,
      "learning_rate": 8.438856868258278e-06,
      "loss": 0.5847,
      "step": 4433
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9230498671531677,
      "learning_rate": 8.438111998975618e-06,
      "loss": 0.5397,
      "step": 4434
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0350542068481445,
      "learning_rate": 8.437366984926023e-06,
      "loss": 0.5518,
      "step": 4435
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9938403367996216,
      "learning_rate": 8.436621826140863e-06,
      "loss": 0.5528,
      "step": 4436
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.030972957611084,
      "learning_rate": 8.435876522651512e-06,
      "loss": 0.6143,
      "step": 4437
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1263898611068726,
      "learning_rate": 8.435131074489353e-06,
      "loss": 0.6138,
      "step": 4438
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.985565721988678,
      "learning_rate": 8.434385481685776e-06,
      "loss": 0.5098,
      "step": 4439
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0614559650421143,
      "learning_rate": 8.43363974427217e-06,
      "loss": 0.5787,
      "step": 4440
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0057392120361328,
      "learning_rate": 8.432893862279943e-06,
      "loss": 0.5294,
      "step": 4441
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.033789038658142,
      "learning_rate": 8.432147835740496e-06,
      "loss": 0.5534,
      "step": 4442
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0420987606048584,
      "learning_rate": 8.431401664685244e-06,
      "loss": 0.5472,
      "step": 4443
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0862915515899658,
      "learning_rate": 8.430655349145604e-06,
      "loss": 0.5556,
      "step": 4444
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0685113668441772,
      "learning_rate": 8.429908889153003e-06,
      "loss": 0.596,
      "step": 4445
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9933927655220032,
      "learning_rate": 8.429162284738868e-06,
      "loss": 0.5705,
      "step": 4446
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9709452390670776,
      "learning_rate": 8.42841553593464e-06,
      "loss": 0.5502,
      "step": 4447
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0400898456573486,
      "learning_rate": 8.42766864277176e-06,
      "loss": 0.5348,
      "step": 4448
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9806199669837952,
      "learning_rate": 8.426921605281677e-06,
      "loss": 0.5603,
      "step": 4449
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0576506853103638,
      "learning_rate": 8.426174423495848e-06,
      "loss": 0.5572,
      "step": 4450
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0964463949203491,
      "learning_rate": 8.425427097445733e-06,
      "loss": 0.5739,
      "step": 4451
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0517578125,
      "learning_rate": 8.424679627162798e-06,
      "loss": 0.5891,
      "step": 4452
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9388140439987183,
      "learning_rate": 8.423932012678516e-06,
      "loss": 0.5163,
      "step": 4453
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0317274332046509,
      "learning_rate": 8.42318425402437e-06,
      "loss": 0.6287,
      "step": 4454
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9393893480300903,
      "learning_rate": 8.422436351231843e-06,
      "loss": 0.5292,
      "step": 4455
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9703336358070374,
      "learning_rate": 8.421688304332428e-06,
      "loss": 0.5127,
      "step": 4456
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0111799240112305,
      "learning_rate": 8.42094011335762e-06,
      "loss": 0.51,
      "step": 4457
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0073695182800293,
      "learning_rate": 8.420191778338924e-06,
      "loss": 0.5102,
      "step": 4458
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9900495409965515,
      "learning_rate": 8.419443299307852e-06,
      "loss": 0.5851,
      "step": 4459
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0441724061965942,
      "learning_rate": 8.418694676295918e-06,
      "loss": 0.5308,
      "step": 4460
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9689183235168457,
      "learning_rate": 8.417945909334642e-06,
      "loss": 0.5456,
      "step": 4461
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0017889738082886,
      "learning_rate": 8.417196998455555e-06,
      "loss": 0.5471,
      "step": 4462
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0057268142700195,
      "learning_rate": 8.41644794369019e-06,
      "loss": 0.5082,
      "step": 4463
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9479397535324097,
      "learning_rate": 8.415698745070088e-06,
      "loss": 0.5388,
      "step": 4464
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0385222434997559,
      "learning_rate": 8.414949402626793e-06,
      "loss": 0.5791,
      "step": 4465
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9783814549446106,
      "learning_rate": 8.41419991639186e-06,
      "loss": 0.5192,
      "step": 4466
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1291273832321167,
      "learning_rate": 8.413450286396845e-06,
      "loss": 0.5347,
      "step": 4467
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8909294009208679,
      "learning_rate": 8.41270051267331e-06,
      "loss": 0.5149,
      "step": 4468
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0434598922729492,
      "learning_rate": 8.411950595252834e-06,
      "loss": 0.5483,
      "step": 4469
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0784733295440674,
      "learning_rate": 8.411200534166983e-06,
      "loss": 0.6113,
      "step": 4470
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.080352783203125,
      "learning_rate": 8.410450329447346e-06,
      "loss": 0.5783,
      "step": 4471
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0519651174545288,
      "learning_rate": 8.409699981125509e-06,
      "loss": 0.591,
      "step": 4472
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0252313613891602,
      "learning_rate": 8.408949489233068e-06,
      "loss": 0.5828,
      "step": 4473
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9834392666816711,
      "learning_rate": 8.408198853801623e-06,
      "loss": 0.5373,
      "step": 4474
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0300211906433105,
      "learning_rate": 8.40744807486278e-06,
      "loss": 0.5858,
      "step": 4475
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9656156301498413,
      "learning_rate": 8.406697152448152e-06,
      "loss": 0.5426,
      "step": 4476
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9705562591552734,
      "learning_rate": 8.405946086589359e-06,
      "loss": 0.5428,
      "step": 4477
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0767998695373535,
      "learning_rate": 8.405194877318023e-06,
      "loss": 0.6171,
      "step": 4478
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9853762984275818,
      "learning_rate": 8.404443524665777e-06,
      "loss": 0.5877,
      "step": 4479
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9431714415550232,
      "learning_rate": 8.40369202866426e-06,
      "loss": 0.5145,
      "step": 4480
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0816575288772583,
      "learning_rate": 8.40294038934511e-06,
      "loss": 0.5858,
      "step": 4481
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9906498193740845,
      "learning_rate": 8.402188606739977e-06,
      "loss": 0.5379,
      "step": 4482
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.089381217956543,
      "learning_rate": 8.401436680880518e-06,
      "loss": 0.5514,
      "step": 4483
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0199429988861084,
      "learning_rate": 8.400684611798395e-06,
      "loss": 0.5221,
      "step": 4484
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9694374799728394,
      "learning_rate": 8.39993239952527e-06,
      "loss": 0.5128,
      "step": 4485
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0346933603286743,
      "learning_rate": 8.399180044092821e-06,
      "loss": 0.5198,
      "step": 4486
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1518710851669312,
      "learning_rate": 8.398427545532726e-06,
      "loss": 0.5294,
      "step": 4487
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0096625089645386,
      "learning_rate": 8.397674903876667e-06,
      "loss": 0.5433,
      "step": 4488
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0959570407867432,
      "learning_rate": 8.396922119156339e-06,
      "loss": 0.5901,
      "step": 4489
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0759719610214233,
      "learning_rate": 8.396169191403438e-06,
      "loss": 0.5404,
      "step": 4490
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9815802574157715,
      "learning_rate": 8.395416120649667e-06,
      "loss": 0.5193,
      "step": 4491
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9859442710876465,
      "learning_rate": 8.394662906926734e-06,
      "loss": 0.5087,
      "step": 4492
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9523860812187195,
      "learning_rate": 8.393909550266354e-06,
      "loss": 0.5617,
      "step": 4493
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9746415615081787,
      "learning_rate": 8.393156050700252e-06,
      "loss": 0.5675,
      "step": 4494
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.037996768951416,
      "learning_rate": 8.39240240826015e-06,
      "loss": 0.5698,
      "step": 4495
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1229304075241089,
      "learning_rate": 8.391648622977787e-06,
      "loss": 0.6012,
      "step": 4496
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9897786378860474,
      "learning_rate": 8.390894694884896e-06,
      "loss": 0.5329,
      "step": 4497
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0292725563049316,
      "learning_rate": 8.390140624013228e-06,
      "loss": 0.5462,
      "step": 4498
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0851386785507202,
      "learning_rate": 8.38938641039453e-06,
      "loss": 0.589,
      "step": 4499
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9658882021903992,
      "learning_rate": 8.388632054060562e-06,
      "loss": 0.5191,
      "step": 4500
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.016357660293579,
      "learning_rate": 8.387877555043086e-06,
      "loss": 0.5857,
      "step": 4501
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9577678442001343,
      "learning_rate": 8.38712291337387e-06,
      "loss": 0.5332,
      "step": 4502
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0035228729248047,
      "learning_rate": 8.386368129084695e-06,
      "loss": 0.5338,
      "step": 4503
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.148341178894043,
      "learning_rate": 8.385613202207336e-06,
      "loss": 0.5602,
      "step": 4504
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.95160973072052,
      "learning_rate": 8.384858132773582e-06,
      "loss": 0.5396,
      "step": 4505
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.062878131866455,
      "learning_rate": 8.38410292081523e-06,
      "loss": 0.5118,
      "step": 4506
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9612262845039368,
      "learning_rate": 8.383347566364072e-06,
      "loss": 0.508,
      "step": 4507
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0526068210601807,
      "learning_rate": 8.38259206945192e-06,
      "loss": 0.5425,
      "step": 4508
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0237740278244019,
      "learning_rate": 8.381836430110585e-06,
      "loss": 0.5376,
      "step": 4509
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0377681255340576,
      "learning_rate": 8.38108064837188e-06,
      "loss": 0.5819,
      "step": 4510
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9746587872505188,
      "learning_rate": 8.380324724267631e-06,
      "loss": 0.5514,
      "step": 4511
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.966952383518219,
      "learning_rate": 8.379568657829669e-06,
      "loss": 0.5774,
      "step": 4512
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9846550226211548,
      "learning_rate": 8.378812449089826e-06,
      "loss": 0.5439,
      "step": 4513
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.055591106414795,
      "learning_rate": 8.378056098079946e-06,
      "loss": 0.594,
      "step": 4514
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0931600332260132,
      "learning_rate": 8.377299604831875e-06,
      "loss": 0.5985,
      "step": 4515
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9879661202430725,
      "learning_rate": 8.376542969377465e-06,
      "loss": 0.5831,
      "step": 4516
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0338890552520752,
      "learning_rate": 8.375786191748578e-06,
      "loss": 0.5988,
      "step": 4517
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1263805627822876,
      "learning_rate": 8.375029271977076e-06,
      "loss": 0.6065,
      "step": 4518
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9827618598937988,
      "learning_rate": 8.374272210094834e-06,
      "loss": 0.5548,
      "step": 4519
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9819245934486389,
      "learning_rate": 8.373515006133728e-06,
      "loss": 0.5061,
      "step": 4520
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9984309673309326,
      "learning_rate": 8.372757660125639e-06,
      "loss": 0.4948,
      "step": 4521
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0824671983718872,
      "learning_rate": 8.372000172102459e-06,
      "loss": 0.585,
      "step": 4522
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0909574031829834,
      "learning_rate": 8.37124254209608e-06,
      "loss": 0.5601,
      "step": 4523
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9879238605499268,
      "learning_rate": 8.370484770138407e-06,
      "loss": 0.5791,
      "step": 4524
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0516186952590942,
      "learning_rate": 8.369726856261346e-06,
      "loss": 0.5755,
      "step": 4525
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0363621711730957,
      "learning_rate": 8.36896880049681e-06,
      "loss": 0.5142,
      "step": 4526
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0198794603347778,
      "learning_rate": 8.368210602876716e-06,
      "loss": 0.5286,
      "step": 4527
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9703680872917175,
      "learning_rate": 8.36745226343299e-06,
      "loss": 0.5467,
      "step": 4528
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0106040239334106,
      "learning_rate": 8.366693782197566e-06,
      "loss": 0.5444,
      "step": 4529
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.034140944480896,
      "learning_rate": 8.365935159202378e-06,
      "loss": 0.5487,
      "step": 4530
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0963879823684692,
      "learning_rate": 8.365176394479368e-06,
      "loss": 0.573,
      "step": 4531
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.983534038066864,
      "learning_rate": 8.364417488060488e-06,
      "loss": 0.5616,
      "step": 4532
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0498030185699463,
      "learning_rate": 8.363658439977693e-06,
      "loss": 0.5222,
      "step": 4533
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9815921187400818,
      "learning_rate": 8.36289925026294e-06,
      "loss": 0.5376,
      "step": 4534
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9987731575965881,
      "learning_rate": 8.362139918948198e-06,
      "loss": 0.5405,
      "step": 4535
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0733778476715088,
      "learning_rate": 8.36138044606544e-06,
      "loss": 0.5804,
      "step": 4536
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0026512145996094,
      "learning_rate": 8.360620831646647e-06,
      "loss": 0.5573,
      "step": 4537
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.978951096534729,
      "learning_rate": 8.359861075723801e-06,
      "loss": 0.5474,
      "step": 4538
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0572445392608643,
      "learning_rate": 8.359101178328893e-06,
      "loss": 0.5784,
      "step": 4539
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9558642506599426,
      "learning_rate": 8.358341139493919e-06,
      "loss": 0.5252,
      "step": 4540
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0329904556274414,
      "learning_rate": 8.357580959250882e-06,
      "loss": 0.5189,
      "step": 4541
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0733036994934082,
      "learning_rate": 8.356820637631792e-06,
      "loss": 0.5723,
      "step": 4542
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.063766360282898,
      "learning_rate": 8.356060174668663e-06,
      "loss": 0.5409,
      "step": 4543
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9099135398864746,
      "learning_rate": 8.355299570393515e-06,
      "loss": 0.4935,
      "step": 4544
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9763380289077759,
      "learning_rate": 8.354538824838373e-06,
      "loss": 0.5566,
      "step": 4545
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9359689950942993,
      "learning_rate": 8.353777938035272e-06,
      "loss": 0.5214,
      "step": 4546
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9775251746177673,
      "learning_rate": 8.353016910016247e-06,
      "loss": 0.541,
      "step": 4547
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0470110177993774,
      "learning_rate": 8.352255740813347e-06,
      "loss": 0.5642,
      "step": 4548
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9434897303581238,
      "learning_rate": 8.351494430458617e-06,
      "loss": 0.5199,
      "step": 4549
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0338044166564941,
      "learning_rate": 8.350732978984116e-06,
      "loss": 0.5699,
      "step": 4550
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9947142601013184,
      "learning_rate": 8.349971386421906e-06,
      "loss": 0.556,
      "step": 4551
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9455569386482239,
      "learning_rate": 8.349209652804055e-06,
      "loss": 0.5067,
      "step": 4552
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9736185073852539,
      "learning_rate": 8.348447778162636e-06,
      "loss": 0.5689,
      "step": 4553
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0313166379928589,
      "learning_rate": 8.347685762529729e-06,
      "loss": 0.5833,
      "step": 4554
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.996852457523346,
      "learning_rate": 8.34692360593742e-06,
      "loss": 0.5084,
      "step": 4555
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0415087938308716,
      "learning_rate": 8.346161308417805e-06,
      "loss": 0.5731,
      "step": 4556
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9775435924530029,
      "learning_rate": 8.345398870002972e-06,
      "loss": 0.5587,
      "step": 4557
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0303115844726562,
      "learning_rate": 8.344636290725035e-06,
      "loss": 0.5762,
      "step": 4558
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9830759167671204,
      "learning_rate": 8.343873570616097e-06,
      "loss": 0.5606,
      "step": 4559
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0046902894973755,
      "learning_rate": 8.343110709708275e-06,
      "loss": 0.5648,
      "step": 4560
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0001426935195923,
      "learning_rate": 8.342347708033692e-06,
      "loss": 0.5636,
      "step": 4561
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.034018874168396,
      "learning_rate": 8.341584565624471e-06,
      "loss": 0.5831,
      "step": 4562
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9828287363052368,
      "learning_rate": 8.340821282512753e-06,
      "loss": 0.5553,
      "step": 4563
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0107008218765259,
      "learning_rate": 8.34005785873067e-06,
      "loss": 0.5308,
      "step": 4564
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.993648111820221,
      "learning_rate": 8.339294294310371e-06,
      "loss": 0.559,
      "step": 4565
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9884464740753174,
      "learning_rate": 8.338530589284005e-06,
      "loss": 0.5824,
      "step": 4566
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9992973804473877,
      "learning_rate": 8.33776674368373e-06,
      "loss": 0.4819,
      "step": 4567
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0691200494766235,
      "learning_rate": 8.337002757541708e-06,
      "loss": 0.6044,
      "step": 4568
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9904386401176453,
      "learning_rate": 8.33623863089011e-06,
      "loss": 0.529,
      "step": 4569
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9538317918777466,
      "learning_rate": 8.335474363761109e-06,
      "loss": 0.5066,
      "step": 4570
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0342077016830444,
      "learning_rate": 8.334709956186884e-06,
      "loss": 0.55,
      "step": 4571
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9945910573005676,
      "learning_rate": 8.333945408199624e-06,
      "loss": 0.5278,
      "step": 4572
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0571881532669067,
      "learning_rate": 8.333180719831521e-06,
      "loss": 0.5608,
      "step": 4573
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0826531648635864,
      "learning_rate": 8.332415891114774e-06,
      "loss": 0.5684,
      "step": 4574
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0074824094772339,
      "learning_rate": 8.331650922081586e-06,
      "loss": 0.5156,
      "step": 4575
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0355405807495117,
      "learning_rate": 8.330885812764168e-06,
      "loss": 0.5606,
      "step": 4576
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9580333232879639,
      "learning_rate": 8.330120563194736e-06,
      "loss": 0.5699,
      "step": 4577
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0509356260299683,
      "learning_rate": 8.32935517340551e-06,
      "loss": 0.5693,
      "step": 4578
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.944019079208374,
      "learning_rate": 8.328589643428722e-06,
      "loss": 0.534,
      "step": 4579
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0960191488265991,
      "learning_rate": 8.327823973296601e-06,
      "loss": 0.5914,
      "step": 4580
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0410923957824707,
      "learning_rate": 8.32705816304139e-06,
      "loss": 0.5643,
      "step": 4581
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0160187482833862,
      "learning_rate": 8.326292212695335e-06,
      "loss": 0.5954,
      "step": 4582
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0226081609725952,
      "learning_rate": 8.325526122290685e-06,
      "loss": 0.5585,
      "step": 4583
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9550665020942688,
      "learning_rate": 8.3247598918597e-06,
      "loss": 0.5219,
      "step": 4584
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.024646282196045,
      "learning_rate": 8.323993521434639e-06,
      "loss": 0.5408,
      "step": 4585
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9836820960044861,
      "learning_rate": 8.323227011047777e-06,
      "loss": 0.537,
      "step": 4586
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9069527387619019,
      "learning_rate": 8.322460360731386e-06,
      "loss": 0.5902,
      "step": 4587
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9728847742080688,
      "learning_rate": 8.321693570517745e-06,
      "loss": 0.5593,
      "step": 4588
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0425338745117188,
      "learning_rate": 8.320926640439145e-06,
      "loss": 0.5218,
      "step": 4589
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.983156681060791,
      "learning_rate": 8.320159570527876e-06,
      "loss": 0.5384,
      "step": 4590
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.932736873626709,
      "learning_rate": 8.319392360816239e-06,
      "loss": 0.536,
      "step": 4591
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9952834844589233,
      "learning_rate": 8.318625011336533e-06,
      "loss": 0.5843,
      "step": 4592
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.061063289642334,
      "learning_rate": 8.317857522121078e-06,
      "loss": 0.5549,
      "step": 4593
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0253628492355347,
      "learning_rate": 8.317089893202181e-06,
      "loss": 0.5438,
      "step": 4594
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9959260821342468,
      "learning_rate": 8.316322124612169e-06,
      "loss": 0.6049,
      "step": 4595
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0181363821029663,
      "learning_rate": 8.315554216383368e-06,
      "loss": 0.5881,
      "step": 4596
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.103354811668396,
      "learning_rate": 8.314786168548115e-06,
      "loss": 0.5834,
      "step": 4597
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0442967414855957,
      "learning_rate": 8.314017981138746e-06,
      "loss": 0.5289,
      "step": 4598
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.093725562095642,
      "learning_rate": 8.31324965418761e-06,
      "loss": 0.5644,
      "step": 4599
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0241702795028687,
      "learning_rate": 8.312481187727055e-06,
      "loss": 0.5611,
      "step": 4600
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9719318151473999,
      "learning_rate": 8.311712581789442e-06,
      "loss": 0.4764,
      "step": 4601
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0587811470031738,
      "learning_rate": 8.310943836407132e-06,
      "loss": 0.5606,
      "step": 4602
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0955878496170044,
      "learning_rate": 8.310174951612495e-06,
      "loss": 0.5183,
      "step": 4603
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0315818786621094,
      "learning_rate": 8.309405927437906e-06,
      "loss": 0.4809,
      "step": 4604
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.043058156967163,
      "learning_rate": 8.308636763915746e-06,
      "loss": 0.559,
      "step": 4605
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0042216777801514,
      "learning_rate": 8.307867461078402e-06,
      "loss": 0.5702,
      "step": 4606
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0011327266693115,
      "learning_rate": 8.307098018958266e-06,
      "loss": 0.5797,
      "step": 4607
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.989730954170227,
      "learning_rate": 8.306328437587738e-06,
      "loss": 0.5656,
      "step": 4608
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0619161128997803,
      "learning_rate": 8.305558716999221e-06,
      "loss": 0.5523,
      "step": 4609
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0472158193588257,
      "learning_rate": 8.304788857225126e-06,
      "loss": 0.5554,
      "step": 4610
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0451831817626953,
      "learning_rate": 8.304018858297867e-06,
      "loss": 0.537,
      "step": 4611
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0356922149658203,
      "learning_rate": 8.30324872024987e-06,
      "loss": 0.549,
      "step": 4612
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.051112413406372,
      "learning_rate": 8.30247844311356e-06,
      "loss": 0.4971,
      "step": 4613
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0303791761398315,
      "learning_rate": 8.301708026921371e-06,
      "loss": 0.5841,
      "step": 4614
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0167726278305054,
      "learning_rate": 8.300937471705742e-06,
      "loss": 0.5971,
      "step": 4615
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9281720519065857,
      "learning_rate": 8.300166777499119e-06,
      "loss": 0.5534,
      "step": 4616
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.017045497894287,
      "learning_rate": 8.299395944333955e-06,
      "loss": 0.6104,
      "step": 4617
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9842953085899353,
      "learning_rate": 8.298624972242704e-06,
      "loss": 0.5341,
      "step": 4618
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0903747081756592,
      "learning_rate": 8.297853861257831e-06,
      "loss": 0.5049,
      "step": 4619
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9999139904975891,
      "learning_rate": 8.297082611411805e-06,
      "loss": 0.5475,
      "step": 4620
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9975643157958984,
      "learning_rate": 8.296311222737099e-06,
      "loss": 0.511,
      "step": 4621
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0004945993423462,
      "learning_rate": 8.295539695266195e-06,
      "loss": 0.5207,
      "step": 4622
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.919409453868866,
      "learning_rate": 8.29476802903158e-06,
      "loss": 0.5177,
      "step": 4623
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9774051904678345,
      "learning_rate": 8.293996224065742e-06,
      "loss": 0.5436,
      "step": 4624
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9625217318534851,
      "learning_rate": 8.293224280401185e-06,
      "loss": 0.5572,
      "step": 4625
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.027803659439087,
      "learning_rate": 8.292452198070406e-06,
      "loss": 0.5563,
      "step": 4626
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9124547839164734,
      "learning_rate": 8.291679977105922e-06,
      "loss": 0.5344,
      "step": 4627
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9668528437614441,
      "learning_rate": 8.290907617540244e-06,
      "loss": 0.5308,
      "step": 4628
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0439895391464233,
      "learning_rate": 8.290135119405894e-06,
      "loss": 0.5132,
      "step": 4629
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.924747109413147,
      "learning_rate": 8.2893624827354e-06,
      "loss": 0.5168,
      "step": 4630
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9718056321144104,
      "learning_rate": 8.288589707561295e-06,
      "loss": 0.5367,
      "step": 4631
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.003222107887268,
      "learning_rate": 8.287816793916119e-06,
      "loss": 0.5405,
      "step": 4632
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0253312587738037,
      "learning_rate": 8.287043741832412e-06,
      "loss": 0.5878,
      "step": 4633
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0082941055297852,
      "learning_rate": 8.28627055134273e-06,
      "loss": 0.4953,
      "step": 4634
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9098595976829529,
      "learning_rate": 8.285497222479626e-06,
      "loss": 0.516,
      "step": 4635
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9030443429946899,
      "learning_rate": 8.284723755275666e-06,
      "loss": 0.5401,
      "step": 4636
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9865958094596863,
      "learning_rate": 8.283950149763413e-06,
      "loss": 0.5145,
      "step": 4637
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9161219000816345,
      "learning_rate": 8.283176405975444e-06,
      "loss": 0.5248,
      "step": 4638
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0371172428131104,
      "learning_rate": 8.282402523944338e-06,
      "loss": 0.5267,
      "step": 4639
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9949166774749756,
      "learning_rate": 8.28162850370268e-06,
      "loss": 0.5002,
      "step": 4640
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0591098070144653,
      "learning_rate": 8.28085434528306e-06,
      "loss": 0.5958,
      "step": 4641
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0102256536483765,
      "learning_rate": 8.28008004871808e-06,
      "loss": 0.5246,
      "step": 4642
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0576274394989014,
      "learning_rate": 8.279305614040337e-06,
      "loss": 0.541,
      "step": 4643
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0965458154678345,
      "learning_rate": 8.278531041282445e-06,
      "loss": 0.5362,
      "step": 4644
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.045287847518921,
      "learning_rate": 8.277756330477013e-06,
      "loss": 0.562,
      "step": 4645
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0357410907745361,
      "learning_rate": 8.276981481656668e-06,
      "loss": 0.5587,
      "step": 4646
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0280876159667969,
      "learning_rate": 8.276206494854029e-06,
      "loss": 0.5946,
      "step": 4647
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.971557080745697,
      "learning_rate": 8.275431370101734e-06,
      "loss": 0.5241,
      "step": 4648
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0403517484664917,
      "learning_rate": 8.274656107432418e-06,
      "loss": 0.5568,
      "step": 4649
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0311771631240845,
      "learning_rate": 8.273880706878724e-06,
      "loss": 0.5656,
      "step": 4650
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1487035751342773,
      "learning_rate": 8.273105168473304e-06,
      "loss": 0.536,
      "step": 4651
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0901763439178467,
      "learning_rate": 8.27232949224881e-06,
      "loss": 0.5805,
      "step": 4652
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0801087617874146,
      "learning_rate": 8.271553678237904e-06,
      "loss": 0.5626,
      "step": 4653
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0637428760528564,
      "learning_rate": 8.270777726473256e-06,
      "loss": 0.5283,
      "step": 4654
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9966744184494019,
      "learning_rate": 8.270001636987535e-06,
      "loss": 0.5295,
      "step": 4655
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.013501524925232,
      "learning_rate": 8.26922540981342e-06,
      "loss": 0.5601,
      "step": 4656
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0766243934631348,
      "learning_rate": 8.268449044983598e-06,
      "loss": 0.5249,
      "step": 4657
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9005308151245117,
      "learning_rate": 8.267672542530753e-06,
      "loss": 0.5389,
      "step": 4658
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9778000116348267,
      "learning_rate": 8.266895902487588e-06,
      "loss": 0.5196,
      "step": 4659
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9980995655059814,
      "learning_rate": 8.2661191248868e-06,
      "loss": 0.5653,
      "step": 4660
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0180249214172363,
      "learning_rate": 8.265342209761098e-06,
      "loss": 0.5532,
      "step": 4661
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9602007865905762,
      "learning_rate": 8.264565157143194e-06,
      "loss": 0.5783,
      "step": 4662
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.979564905166626,
      "learning_rate": 8.26378796706581e-06,
      "loss": 0.5458,
      "step": 4663
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0173044204711914,
      "learning_rate": 8.263010639561666e-06,
      "loss": 0.5548,
      "step": 4664
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9330337643623352,
      "learning_rate": 8.262233174663497e-06,
      "loss": 0.5383,
      "step": 4665
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0098026990890503,
      "learning_rate": 8.261455572404036e-06,
      "loss": 0.5428,
      "step": 4666
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0305527448654175,
      "learning_rate": 8.260677832816029e-06,
      "loss": 0.5413,
      "step": 4667
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0110151767730713,
      "learning_rate": 8.259899955932218e-06,
      "loss": 0.5382,
      "step": 4668
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0421013832092285,
      "learning_rate": 8.259121941785362e-06,
      "loss": 0.5617,
      "step": 4669
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9988119006156921,
      "learning_rate": 8.25834379040822e-06,
      "loss": 0.5215,
      "step": 4670
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9611369371414185,
      "learning_rate": 8.257565501833555e-06,
      "loss": 0.5705,
      "step": 4671
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9753782749176025,
      "learning_rate": 8.256787076094138e-06,
      "loss": 0.5314,
      "step": 4672
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.049460530281067,
      "learning_rate": 8.256008513222747e-06,
      "loss": 0.529,
      "step": 4673
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0090168714523315,
      "learning_rate": 8.255229813252167e-06,
      "loss": 0.5604,
      "step": 4674
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9995334148406982,
      "learning_rate": 8.25445097621518e-06,
      "loss": 0.536,
      "step": 4675
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0167433023452759,
      "learning_rate": 8.253672002144584e-06,
      "loss": 0.5293,
      "step": 4676
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9629647731781006,
      "learning_rate": 8.25289289107318e-06,
      "loss": 0.5628,
      "step": 4677
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9389039278030396,
      "learning_rate": 8.252113643033774e-06,
      "loss": 0.5432,
      "step": 4678
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0852667093276978,
      "learning_rate": 8.251334258059173e-06,
      "loss": 0.633,
      "step": 4679
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9621460437774658,
      "learning_rate": 8.250554736182199e-06,
      "loss": 0.5695,
      "step": 4680
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9658225774765015,
      "learning_rate": 8.249775077435671e-06,
      "loss": 0.5503,
      "step": 4681
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.971916139125824,
      "learning_rate": 8.24899528185242e-06,
      "loss": 0.4941,
      "step": 4682
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0347654819488525,
      "learning_rate": 8.24821534946528e-06,
      "loss": 0.5648,
      "step": 4683
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9471699595451355,
      "learning_rate": 8.247435280307093e-06,
      "loss": 0.557,
      "step": 4684
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9978876709938049,
      "learning_rate": 8.246655074410703e-06,
      "loss": 0.5893,
      "step": 4685
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.061025857925415,
      "learning_rate": 8.24587473180896e-06,
      "loss": 0.5648,
      "step": 4686
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0649787187576294,
      "learning_rate": 8.245094252534727e-06,
      "loss": 0.538,
      "step": 4687
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0082396268844604,
      "learning_rate": 8.244313636620862e-06,
      "loss": 0.5599,
      "step": 4688
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9994105696678162,
      "learning_rate": 8.243532884100236e-06,
      "loss": 0.546,
      "step": 4689
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.007741928100586,
      "learning_rate": 8.242751995005721e-06,
      "loss": 0.5514,
      "step": 4690
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9305049180984497,
      "learning_rate": 8.241970969370205e-06,
      "loss": 0.5207,
      "step": 4691
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0813062191009521,
      "learning_rate": 8.241189807226566e-06,
      "loss": 0.5586,
      "step": 4692
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.026228427886963,
      "learning_rate": 8.240408508607703e-06,
      "loss": 0.5504,
      "step": 4693
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.02143394947052,
      "learning_rate": 8.239627073546507e-06,
      "loss": 0.5425,
      "step": 4694
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9124248623847961,
      "learning_rate": 8.238845502075886e-06,
      "loss": 0.5542,
      "step": 4695
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9577934145927429,
      "learning_rate": 8.238063794228748e-06,
      "loss": 0.5093,
      "step": 4696
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9729812145233154,
      "learning_rate": 8.237281950038008e-06,
      "loss": 0.6089,
      "step": 4697
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0045956373214722,
      "learning_rate": 8.236499969536585e-06,
      "loss": 0.5789,
      "step": 4698
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9919505715370178,
      "learning_rate": 8.23571785275741e-06,
      "loss": 0.5477,
      "step": 4699
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0783262252807617,
      "learning_rate": 8.234935599733412e-06,
      "loss": 0.5961,
      "step": 4700
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9970394372940063,
      "learning_rate": 8.234153210497528e-06,
      "loss": 0.4814,
      "step": 4701
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.00620436668396,
      "learning_rate": 8.233370685082704e-06,
      "loss": 0.5642,
      "step": 4702
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9523754715919495,
      "learning_rate": 8.232588023521888e-06,
      "loss": 0.53,
      "step": 4703
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0080584287643433,
      "learning_rate": 8.231805225848035e-06,
      "loss": 0.5723,
      "step": 4704
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9508122205734253,
      "learning_rate": 8.23102229209411e-06,
      "loss": 0.512,
      "step": 4705
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0608426332473755,
      "learning_rate": 8.230239222293073e-06,
      "loss": 0.4941,
      "step": 4706
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0919255018234253,
      "learning_rate": 8.229456016477899e-06,
      "loss": 0.5382,
      "step": 4707
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0042272806167603,
      "learning_rate": 8.228672674681568e-06,
      "loss": 0.5456,
      "step": 4708
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.12455153465271,
      "learning_rate": 8.227889196937062e-06,
      "loss": 0.6251,
      "step": 4709
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0752999782562256,
      "learning_rate": 8.227105583277372e-06,
      "loss": 0.497,
      "step": 4710
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9840076565742493,
      "learning_rate": 8.22632183373549e-06,
      "loss": 0.5404,
      "step": 4711
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.017511248588562,
      "learning_rate": 8.225537948344423e-06,
      "loss": 0.5328,
      "step": 4712
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9743256568908691,
      "learning_rate": 8.224753927137171e-06,
      "loss": 0.5761,
      "step": 4713
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.004582166671753,
      "learning_rate": 8.22396977014675e-06,
      "loss": 0.5328,
      "step": 4714
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.04586923122406,
      "learning_rate": 8.223185477406175e-06,
      "loss": 0.569,
      "step": 4715
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.086186170578003,
      "learning_rate": 8.222401048948476e-06,
      "loss": 0.5482,
      "step": 4716
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9625822305679321,
      "learning_rate": 8.221616484806676e-06,
      "loss": 0.5437,
      "step": 4717
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0753564834594727,
      "learning_rate": 8.220831785013814e-06,
      "loss": 0.5511,
      "step": 4718
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9520816802978516,
      "learning_rate": 8.22004694960293e-06,
      "loss": 0.5087,
      "step": 4719
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.970871090888977,
      "learning_rate": 8.21926197860707e-06,
      "loss": 0.5151,
      "step": 4720
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0056312084197998,
      "learning_rate": 8.218476872059288e-06,
      "loss": 0.5275,
      "step": 4721
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.066782832145691,
      "learning_rate": 8.217691629992641e-06,
      "loss": 0.5548,
      "step": 4722
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9306086897850037,
      "learning_rate": 8.216906252440193e-06,
      "loss": 0.5172,
      "step": 4723
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0483423471450806,
      "learning_rate": 8.216120739435013e-06,
      "loss": 0.5743,
      "step": 4724
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0240111351013184,
      "learning_rate": 8.215335091010177e-06,
      "loss": 0.516,
      "step": 4725
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0304317474365234,
      "learning_rate": 8.214549307198765e-06,
      "loss": 0.537,
      "step": 4726
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0802671909332275,
      "learning_rate": 8.213763388033867e-06,
      "loss": 0.5904,
      "step": 4727
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0293980836868286,
      "learning_rate": 8.212977333548569e-06,
      "loss": 0.5909,
      "step": 4728
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0778398513793945,
      "learning_rate": 8.212191143775973e-06,
      "loss": 0.5599,
      "step": 4729
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0456862449645996,
      "learning_rate": 8.211404818749184e-06,
      "loss": 0.6011,
      "step": 4730
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9759236574172974,
      "learning_rate": 8.21061835850131e-06,
      "loss": 0.5174,
      "step": 4731
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0486438274383545,
      "learning_rate": 8.209831763065465e-06,
      "loss": 0.5671,
      "step": 4732
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0495485067367554,
      "learning_rate": 8.209045032474773e-06,
      "loss": 0.5122,
      "step": 4733
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9698861241340637,
      "learning_rate": 8.208258166762355e-06,
      "loss": 0.5258,
      "step": 4734
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0011286735534668,
      "learning_rate": 8.207471165961347e-06,
      "loss": 0.5479,
      "step": 4735
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9875169992446899,
      "learning_rate": 8.206684030104886e-06,
      "loss": 0.565,
      "step": 4736
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9940193891525269,
      "learning_rate": 8.205896759226115e-06,
      "loss": 0.5387,
      "step": 4737
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9382634162902832,
      "learning_rate": 8.205109353358186e-06,
      "loss": 0.571,
      "step": 4738
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9879447221755981,
      "learning_rate": 8.20432181253425e-06,
      "loss": 0.5197,
      "step": 4739
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9900509119033813,
      "learning_rate": 8.203534136787473e-06,
      "loss": 0.5059,
      "step": 4740
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0219135284423828,
      "learning_rate": 8.202746326151015e-06,
      "loss": 0.5262,
      "step": 4741
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9983688592910767,
      "learning_rate": 8.20195838065805e-06,
      "loss": 0.6165,
      "step": 4742
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0055594444274902,
      "learning_rate": 8.201170300341757e-06,
      "loss": 0.552,
      "step": 4743
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9651437997817993,
      "learning_rate": 8.20038208523532e-06,
      "loss": 0.5851,
      "step": 4744
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1710509061813354,
      "learning_rate": 8.199593735371924e-06,
      "loss": 0.5762,
      "step": 4745
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0982491970062256,
      "learning_rate": 8.198805250784769e-06,
      "loss": 0.5491,
      "step": 4746
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0060925483703613,
      "learning_rate": 8.198016631507053e-06,
      "loss": 0.5129,
      "step": 4747
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0412200689315796,
      "learning_rate": 8.19722787757198e-06,
      "loss": 0.5161,
      "step": 4748
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.054504632949829,
      "learning_rate": 8.196438989012765e-06,
      "loss": 0.5459,
      "step": 4749
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0147706270217896,
      "learning_rate": 8.195649965862622e-06,
      "loss": 0.5291,
      "step": 4750
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9944101572036743,
      "learning_rate": 8.194860808154778e-06,
      "loss": 0.5146,
      "step": 4751
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.054218053817749,
      "learning_rate": 8.194071515922456e-06,
      "loss": 0.5231,
      "step": 4752
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0329952239990234,
      "learning_rate": 8.193282089198897e-06,
      "loss": 0.5629,
      "step": 4753
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0480854511260986,
      "learning_rate": 8.192492528017337e-06,
      "loss": 0.5488,
      "step": 4754
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9405835866928101,
      "learning_rate": 8.191702832411023e-06,
      "loss": 0.5193,
      "step": 4755
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0800672769546509,
      "learning_rate": 8.190913002413204e-06,
      "loss": 0.5868,
      "step": 4756
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.053136944770813,
      "learning_rate": 8.19012303805714e-06,
      "loss": 0.5606,
      "step": 4757
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0696057081222534,
      "learning_rate": 8.189332939376092e-06,
      "loss": 0.5747,
      "step": 4758
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9500235319137573,
      "learning_rate": 8.188542706403331e-06,
      "loss": 0.499,
      "step": 4759
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.010105848312378,
      "learning_rate": 8.187752339172126e-06,
      "loss": 0.5386,
      "step": 4760
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9794458150863647,
      "learning_rate": 8.18696183771576e-06,
      "loss": 0.5711,
      "step": 4761
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9968729019165039,
      "learning_rate": 8.18617120206752e-06,
      "loss": 0.5591,
      "step": 4762
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9325933456420898,
      "learning_rate": 8.185380432260693e-06,
      "loss": 0.5214,
      "step": 4763
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1231193542480469,
      "learning_rate": 8.184589528328576e-06,
      "loss": 0.5749,
      "step": 4764
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0722376108169556,
      "learning_rate": 8.183798490304473e-06,
      "loss": 0.5411,
      "step": 4765
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.111397385597229,
      "learning_rate": 8.183007318221691e-06,
      "loss": 0.6101,
      "step": 4766
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.991192638874054,
      "learning_rate": 8.182216012113543e-06,
      "loss": 0.5273,
      "step": 4767
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9147621989250183,
      "learning_rate": 8.181424572013352e-06,
      "loss": 0.503,
      "step": 4768
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.976429283618927,
      "learning_rate": 8.180632997954437e-06,
      "loss": 0.4936,
      "step": 4769
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9914647936820984,
      "learning_rate": 8.179841289970132e-06,
      "loss": 0.5334,
      "step": 4770
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0159574747085571,
      "learning_rate": 8.179049448093771e-06,
      "loss": 0.5456,
      "step": 4771
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.998140811920166,
      "learning_rate": 8.178257472358697e-06,
      "loss": 0.5649,
      "step": 4772
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0405910015106201,
      "learning_rate": 8.177465362798259e-06,
      "loss": 0.5688,
      "step": 4773
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0000064373016357,
      "learning_rate": 8.176673119445807e-06,
      "loss": 0.5521,
      "step": 4774
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9818173050880432,
      "learning_rate": 8.1758807423347e-06,
      "loss": 0.5873,
      "step": 4775
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9706140160560608,
      "learning_rate": 8.175088231498304e-06,
      "loss": 0.5542,
      "step": 4776
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0345028638839722,
      "learning_rate": 8.174295586969987e-06,
      "loss": 0.4983,
      "step": 4777
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0244017839431763,
      "learning_rate": 8.173502808783127e-06,
      "loss": 0.576,
      "step": 4778
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.979910671710968,
      "learning_rate": 8.172709896971103e-06,
      "loss": 0.5135,
      "step": 4779
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1312154531478882,
      "learning_rate": 8.1719168515673e-06,
      "loss": 0.6129,
      "step": 4780
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.084006428718567,
      "learning_rate": 8.171123672605116e-06,
      "loss": 0.5455,
      "step": 4781
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0457392930984497,
      "learning_rate": 8.170330360117944e-06,
      "loss": 0.5901,
      "step": 4782
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0750620365142822,
      "learning_rate": 8.169536914139189e-06,
      "loss": 0.509,
      "step": 4783
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.007377028465271,
      "learning_rate": 8.168743334702262e-06,
      "loss": 0.5299,
      "step": 4784
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.197182297706604,
      "learning_rate": 8.167949621840576e-06,
      "loss": 0.5613,
      "step": 4785
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0056759119033813,
      "learning_rate": 8.16715577558755e-06,
      "loss": 0.5237,
      "step": 4786
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9658280611038208,
      "learning_rate": 8.166361795976614e-06,
      "loss": 0.5577,
      "step": 4787
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0165833234786987,
      "learning_rate": 8.165567683041197e-06,
      "loss": 0.5531,
      "step": 4788
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9536402821540833,
      "learning_rate": 8.164773436814736e-06,
      "loss": 0.518,
      "step": 4789
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0761758089065552,
      "learning_rate": 8.163979057330677e-06,
      "loss": 0.537,
      "step": 4790
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0675421953201294,
      "learning_rate": 8.163184544622467e-06,
      "loss": 0.4321,
      "step": 4791
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0709632635116577,
      "learning_rate": 8.162389898723558e-06,
      "loss": 0.5416,
      "step": 4792
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.984924852848053,
      "learning_rate": 8.161595119667413e-06,
      "loss": 0.5235,
      "step": 4793
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.097145915031433,
      "learning_rate": 8.160800207487495e-06,
      "loss": 0.5539,
      "step": 4794
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9144260883331299,
      "learning_rate": 8.160005162217275e-06,
      "loss": 0.5375,
      "step": 4795
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.924608051776886,
      "learning_rate": 8.159209983890232e-06,
      "loss": 0.5342,
      "step": 4796
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0322277545928955,
      "learning_rate": 8.158414672539845e-06,
      "loss": 0.5645,
      "step": 4797
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0819236040115356,
      "learning_rate": 8.157619228199605e-06,
      "loss": 0.535,
      "step": 4798
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2202394008636475,
      "learning_rate": 8.156823650903003e-06,
      "loss": 0.6181,
      "step": 4799
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0472339391708374,
      "learning_rate": 8.156027940683539e-06,
      "loss": 0.5069,
      "step": 4800
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0055584907531738,
      "learning_rate": 8.15523209757472e-06,
      "loss": 0.6032,
      "step": 4801
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0556094646453857,
      "learning_rate": 8.15443612161005e-06,
      "loss": 0.5541,
      "step": 4802
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9796180725097656,
      "learning_rate": 8.15364001282305e-06,
      "loss": 0.5515,
      "step": 4803
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0150129795074463,
      "learning_rate": 8.15284377124724e-06,
      "loss": 0.5023,
      "step": 4804
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9804368019104004,
      "learning_rate": 8.152047396916145e-06,
      "loss": 0.4709,
      "step": 4805
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1215652227401733,
      "learning_rate": 8.1512508898633e-06,
      "loss": 0.5868,
      "step": 4806
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9547824263572693,
      "learning_rate": 8.150454250122245e-06,
      "loss": 0.5221,
      "step": 4807
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0794047117233276,
      "learning_rate": 8.149657477726518e-06,
      "loss": 0.518,
      "step": 4808
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0394928455352783,
      "learning_rate": 8.148860572709674e-06,
      "loss": 0.5546,
      "step": 4809
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.95167076587677,
      "learning_rate": 8.148063535105261e-06,
      "loss": 0.5344,
      "step": 4810
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.120106816291809,
      "learning_rate": 8.147266364946848e-06,
      "loss": 0.5504,
      "step": 4811
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0547051429748535,
      "learning_rate": 8.146469062267995e-06,
      "loss": 0.5594,
      "step": 4812
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.05607271194458,
      "learning_rate": 8.145671627102277e-06,
      "loss": 0.5548,
      "step": 4813
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9690526723861694,
      "learning_rate": 8.14487405948327e-06,
      "loss": 0.5183,
      "step": 4814
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9852856993675232,
      "learning_rate": 8.144076359444555e-06,
      "loss": 0.4959,
      "step": 4815
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0635384321212769,
      "learning_rate": 8.143278527019722e-06,
      "loss": 0.5398,
      "step": 4816
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1019240617752075,
      "learning_rate": 8.142480562242365e-06,
      "loss": 0.5441,
      "step": 4817
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0027168989181519,
      "learning_rate": 8.141682465146084e-06,
      "loss": 0.5409,
      "step": 4818
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9465576410293579,
      "learning_rate": 8.140884235764484e-06,
      "loss": 0.544,
      "step": 4819
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.001171588897705,
      "learning_rate": 8.140085874131174e-06,
      "loss": 0.5424,
      "step": 4820
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9729893803596497,
      "learning_rate": 8.139287380279773e-06,
      "loss": 0.5462,
      "step": 4821
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1184935569763184,
      "learning_rate": 8.138488754243899e-06,
      "loss": 0.5888,
      "step": 4822
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1190617084503174,
      "learning_rate": 8.137689996057183e-06,
      "loss": 0.5694,
      "step": 4823
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.957481861114502,
      "learning_rate": 8.136891105753258e-06,
      "loss": 0.4887,
      "step": 4824
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9850767254829407,
      "learning_rate": 8.136092083365758e-06,
      "loss": 0.5534,
      "step": 4825
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.052511215209961,
      "learning_rate": 8.13529292892833e-06,
      "loss": 0.5776,
      "step": 4826
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9635382294654846,
      "learning_rate": 8.134493642474625e-06,
      "loss": 0.5626,
      "step": 4827
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9893434047698975,
      "learning_rate": 8.133694224038297e-06,
      "loss": 0.5278,
      "step": 4828
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0003610849380493,
      "learning_rate": 8.132894673653007e-06,
      "loss": 0.5565,
      "step": 4829
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0402261018753052,
      "learning_rate": 8.13209499135242e-06,
      "loss": 0.5364,
      "step": 4830
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0688753128051758,
      "learning_rate": 8.131295177170208e-06,
      "loss": 0.5734,
      "step": 4831
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0642523765563965,
      "learning_rate": 8.13049523114005e-06,
      "loss": 0.5675,
      "step": 4832
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9267446398735046,
      "learning_rate": 8.129695153295627e-06,
      "loss": 0.5026,
      "step": 4833
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9632136821746826,
      "learning_rate": 8.128894943670631e-06,
      "loss": 0.5302,
      "step": 4834
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9400724172592163,
      "learning_rate": 8.12809460229875e-06,
      "loss": 0.507,
      "step": 4835
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9593424797058105,
      "learning_rate": 8.127294129213691e-06,
      "loss": 0.5504,
      "step": 4836
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0008407831192017,
      "learning_rate": 8.126493524449153e-06,
      "loss": 0.5677,
      "step": 4837
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0743299722671509,
      "learning_rate": 8.12569278803885e-06,
      "loss": 0.5388,
      "step": 4838
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0238865613937378,
      "learning_rate": 8.124891920016495e-06,
      "loss": 0.5762,
      "step": 4839
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0223515033721924,
      "learning_rate": 8.124090920415814e-06,
      "loss": 0.574,
      "step": 4840
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.026674509048462,
      "learning_rate": 8.123289789270532e-06,
      "loss": 0.5846,
      "step": 4841
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0493360757827759,
      "learning_rate": 8.12248852661438e-06,
      "loss": 0.5678,
      "step": 4842
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9677516222000122,
      "learning_rate": 8.121687132481101e-06,
      "loss": 0.5142,
      "step": 4843
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9218686819076538,
      "learning_rate": 8.120885606904436e-06,
      "loss": 0.5231,
      "step": 4844
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9612337350845337,
      "learning_rate": 8.120083949918137e-06,
      "loss": 0.5023,
      "step": 4845
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.932696521282196,
      "learning_rate": 8.119282161555952e-06,
      "loss": 0.5063,
      "step": 4846
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0821776390075684,
      "learning_rate": 8.11848024185165e-06,
      "loss": 0.5861,
      "step": 4847
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.029322862625122,
      "learning_rate": 8.117678190838991e-06,
      "loss": 0.544,
      "step": 4848
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0611780881881714,
      "learning_rate": 8.116876008551751e-06,
      "loss": 0.5172,
      "step": 4849
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9280547499656677,
      "learning_rate": 8.116073695023704e-06,
      "loss": 0.5443,
      "step": 4850
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0576995611190796,
      "learning_rate": 8.115271250288635e-06,
      "loss": 0.5182,
      "step": 4851
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1136101484298706,
      "learning_rate": 8.114468674380328e-06,
      "loss": 0.5999,
      "step": 4852
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0218087434768677,
      "learning_rate": 8.113665967332582e-06,
      "loss": 0.5447,
      "step": 4853
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9869479537010193,
      "learning_rate": 8.112863129179194e-06,
      "loss": 0.5787,
      "step": 4854
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.043057918548584,
      "learning_rate": 8.112060159953966e-06,
      "loss": 0.5773,
      "step": 4855
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0893371105194092,
      "learning_rate": 8.111257059690714e-06,
      "loss": 0.5851,
      "step": 4856
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0911871194839478,
      "learning_rate": 8.110453828423248e-06,
      "loss": 0.5698,
      "step": 4857
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0049121379852295,
      "learning_rate": 8.109650466185394e-06,
      "loss": 0.5398,
      "step": 4858
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0009570121765137,
      "learning_rate": 8.108846973010975e-06,
      "loss": 0.502,
      "step": 4859
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.035396695137024,
      "learning_rate": 8.108043348933825e-06,
      "loss": 0.5625,
      "step": 4860
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0491327047348022,
      "learning_rate": 8.107239593987781e-06,
      "loss": 0.5217,
      "step": 4861
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0376914739608765,
      "learning_rate": 8.10643570820669e-06,
      "loss": 0.5596,
      "step": 4862
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.05499267578125,
      "learning_rate": 8.105631691624394e-06,
      "loss": 0.5712,
      "step": 4863
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0017470121383667,
      "learning_rate": 8.104827544274754e-06,
      "loss": 0.6343,
      "step": 4864
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.988599956035614,
      "learning_rate": 8.104023266191625e-06,
      "loss": 0.5115,
      "step": 4865
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9934467673301697,
      "learning_rate": 8.103218857408875e-06,
      "loss": 0.5522,
      "step": 4866
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1765915155410767,
      "learning_rate": 8.102414317960373e-06,
      "loss": 0.5593,
      "step": 4867
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0065627098083496,
      "learning_rate": 8.10160964788e-06,
      "loss": 0.5288,
      "step": 4868
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.084834337234497,
      "learning_rate": 8.100804847201632e-06,
      "loss": 0.5505,
      "step": 4869
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1313133239746094,
      "learning_rate": 8.09999991595916e-06,
      "loss": 0.5272,
      "step": 4870
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1298694610595703,
      "learning_rate": 8.099194854186475e-06,
      "loss": 0.5865,
      "step": 4871
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9154739379882812,
      "learning_rate": 8.098389661917475e-06,
      "loss": 0.4787,
      "step": 4872
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0516656637191772,
      "learning_rate": 8.097584339186066e-06,
      "loss": 0.5343,
      "step": 4873
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2043085098266602,
      "learning_rate": 8.096778886026155e-06,
      "loss": 0.5639,
      "step": 4874
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0819754600524902,
      "learning_rate": 8.09597330247166e-06,
      "loss": 0.5108,
      "step": 4875
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.084563970565796,
      "learning_rate": 8.095167588556498e-06,
      "loss": 0.5942,
      "step": 4876
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0509734153747559,
      "learning_rate": 8.094361744314597e-06,
      "loss": 0.5664,
      "step": 4877
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0124056339263916,
      "learning_rate": 8.093555769779887e-06,
      "loss": 0.497,
      "step": 4878
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.034844160079956,
      "learning_rate": 8.092749664986304e-06,
      "loss": 0.5735,
      "step": 4879
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.966972291469574,
      "learning_rate": 8.091943429967792e-06,
      "loss": 0.5141,
      "step": 4880
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1028048992156982,
      "learning_rate": 8.0911370647583e-06,
      "loss": 0.621,
      "step": 4881
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9943513870239258,
      "learning_rate": 8.090330569391778e-06,
      "loss": 0.5266,
      "step": 4882
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.095138669013977,
      "learning_rate": 8.089523943902187e-06,
      "loss": 0.5717,
      "step": 4883
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0169485807418823,
      "learning_rate": 8.088717188323489e-06,
      "loss": 0.5974,
      "step": 4884
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1654067039489746,
      "learning_rate": 8.087910302689656e-06,
      "loss": 0.5394,
      "step": 4885
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9405874013900757,
      "learning_rate": 8.087103287034664e-06,
      "loss": 0.552,
      "step": 4886
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9728289246559143,
      "learning_rate": 8.086296141392489e-06,
      "loss": 0.5,
      "step": 4887
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9715489745140076,
      "learning_rate": 8.08548886579712e-06,
      "loss": 0.5554,
      "step": 4888
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0222160816192627,
      "learning_rate": 8.08468146028255e-06,
      "loss": 0.5288,
      "step": 4889
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9580404758453369,
      "learning_rate": 8.083873924882775e-06,
      "loss": 0.5233,
      "step": 4890
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1099965572357178,
      "learning_rate": 8.083066259631796e-06,
      "loss": 0.5475,
      "step": 4891
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9971556067466736,
      "learning_rate": 8.082258464563621e-06,
      "loss": 0.5567,
      "step": 4892
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9796202182769775,
      "learning_rate": 8.081450539712266e-06,
      "loss": 0.5169,
      "step": 4893
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.042241096496582,
      "learning_rate": 8.080642485111747e-06,
      "loss": 0.5239,
      "step": 4894
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9608373641967773,
      "learning_rate": 8.07983430079609e-06,
      "loss": 0.5181,
      "step": 4895
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9804826974868774,
      "learning_rate": 8.079025986799326e-06,
      "loss": 0.5064,
      "step": 4896
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0063300132751465,
      "learning_rate": 8.078217543155488e-06,
      "loss": 0.4989,
      "step": 4897
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.039695382118225,
      "learning_rate": 8.077408969898619e-06,
      "loss": 0.609,
      "step": 4898
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9815044403076172,
      "learning_rate": 8.076600267062761e-06,
      "loss": 0.5646,
      "step": 4899
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0451741218566895,
      "learning_rate": 8.07579143468197e-06,
      "loss": 0.5568,
      "step": 4900
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9567393660545349,
      "learning_rate": 8.074982472790302e-06,
      "loss": 0.4964,
      "step": 4901
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0217117071151733,
      "learning_rate": 8.074173381421819e-06,
      "loss": 0.5487,
      "step": 4902
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0442255735397339,
      "learning_rate": 8.073364160610589e-06,
      "loss": 0.5254,
      "step": 4903
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9374282956123352,
      "learning_rate": 8.072554810390685e-06,
      "loss": 0.5208,
      "step": 4904
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1306034326553345,
      "learning_rate": 8.071745330796187e-06,
      "loss": 0.6253,
      "step": 4905
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9648675918579102,
      "learning_rate": 8.070935721861178e-06,
      "loss": 0.5291,
      "step": 4906
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9852427840232849,
      "learning_rate": 8.07012598361975e-06,
      "loss": 0.5502,
      "step": 4907
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9443341493606567,
      "learning_rate": 8.069316116105996e-06,
      "loss": 0.6045,
      "step": 4908
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9153281450271606,
      "learning_rate": 8.068506119354019e-06,
      "loss": 0.505,
      "step": 4909
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0826537609100342,
      "learning_rate": 8.067695993397923e-06,
      "loss": 0.5711,
      "step": 4910
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9605743885040283,
      "learning_rate": 8.066885738271821e-06,
      "loss": 0.4974,
      "step": 4911
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.985244870185852,
      "learning_rate": 8.06607535400983e-06,
      "loss": 0.5372,
      "step": 4912
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0411839485168457,
      "learning_rate": 8.06526484064607e-06,
      "loss": 0.5248,
      "step": 4913
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0325897932052612,
      "learning_rate": 8.064454198214673e-06,
      "loss": 0.5599,
      "step": 4914
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0130306482315063,
      "learning_rate": 8.063643426749769e-06,
      "loss": 0.5652,
      "step": 4915
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0241799354553223,
      "learning_rate": 8.062832526285498e-06,
      "loss": 0.5673,
      "step": 4916
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.090224027633667,
      "learning_rate": 8.062021496856004e-06,
      "loss": 0.5921,
      "step": 4917
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0551925897598267,
      "learning_rate": 8.061210338495437e-06,
      "loss": 0.5828,
      "step": 4918
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9556765556335449,
      "learning_rate": 8.060399051237952e-06,
      "loss": 0.5779,
      "step": 4919
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.074124813079834,
      "learning_rate": 8.059587635117709e-06,
      "loss": 0.5351,
      "step": 4920
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9773049354553223,
      "learning_rate": 8.058776090168874e-06,
      "loss": 0.5298,
      "step": 4921
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9283863306045532,
      "learning_rate": 8.057964416425618e-06,
      "loss": 0.5124,
      "step": 4922
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0056229829788208,
      "learning_rate": 8.05715261392212e-06,
      "loss": 0.4826,
      "step": 4923
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9548004269599915,
      "learning_rate": 8.05634068269256e-06,
      "loss": 0.5044,
      "step": 4924
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9597302675247192,
      "learning_rate": 8.055528622771124e-06,
      "loss": 0.5362,
      "step": 4925
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9519512057304382,
      "learning_rate": 8.05471643419201e-06,
      "loss": 0.5208,
      "step": 4926
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9922435283660889,
      "learning_rate": 8.053904116989413e-06,
      "loss": 0.5491,
      "step": 4927
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.922353208065033,
      "learning_rate": 8.053091671197537e-06,
      "loss": 0.4871,
      "step": 4928
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0409049987792969,
      "learning_rate": 8.052279096850591e-06,
      "loss": 0.5317,
      "step": 4929
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9409239292144775,
      "learning_rate": 8.051466393982792e-06,
      "loss": 0.5379,
      "step": 4930
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0207635164260864,
      "learning_rate": 8.050653562628356e-06,
      "loss": 0.5513,
      "step": 4931
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0265536308288574,
      "learning_rate": 8.049840602821512e-06,
      "loss": 0.5972,
      "step": 4932
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1316540241241455,
      "learning_rate": 8.04902751459649e-06,
      "loss": 0.5525,
      "step": 4933
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9593460559844971,
      "learning_rate": 8.048214297987526e-06,
      "loss": 0.4956,
      "step": 4934
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9455051422119141,
      "learning_rate": 8.047400953028863e-06,
      "loss": 0.5173,
      "step": 4935
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9488646388053894,
      "learning_rate": 8.046587479754746e-06,
      "loss": 0.5095,
      "step": 4936
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0066953897476196,
      "learning_rate": 8.04577387819943e-06,
      "loss": 0.5887,
      "step": 4937
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0051743984222412,
      "learning_rate": 8.044960148397168e-06,
      "loss": 0.5565,
      "step": 4938
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0139719247817993,
      "learning_rate": 8.04414629038223e-06,
      "loss": 0.5635,
      "step": 4939
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9318371415138245,
      "learning_rate": 8.04333230418888e-06,
      "loss": 0.5089,
      "step": 4940
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.042931318283081,
      "learning_rate": 8.042518189851394e-06,
      "loss": 0.5411,
      "step": 4941
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9732179641723633,
      "learning_rate": 8.04170394740405e-06,
      "loss": 0.562,
      "step": 4942
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9993977546691895,
      "learning_rate": 8.040889576881136e-06,
      "loss": 0.5448,
      "step": 4943
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.030236005783081,
      "learning_rate": 8.04007507831694e-06,
      "loss": 0.5764,
      "step": 4944
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0133357048034668,
      "learning_rate": 8.039260451745758e-06,
      "loss": 0.5193,
      "step": 4945
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9909392595291138,
      "learning_rate": 8.03844569720189e-06,
      "loss": 0.5408,
      "step": 4946
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0355006456375122,
      "learning_rate": 8.037630814719644e-06,
      "loss": 0.5258,
      "step": 4947
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9290564060211182,
      "learning_rate": 8.036815804333334e-06,
      "loss": 0.5275,
      "step": 4948
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.990656316280365,
      "learning_rate": 8.036000666077273e-06,
      "loss": 0.549,
      "step": 4949
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9807307720184326,
      "learning_rate": 8.035185399985784e-06,
      "loss": 0.522,
      "step": 4950
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0541224479675293,
      "learning_rate": 8.034370006093198e-06,
      "loss": 0.6065,
      "step": 4951
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0626296997070312,
      "learning_rate": 8.033554484433848e-06,
      "loss": 0.5326,
      "step": 4952
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0675690174102783,
      "learning_rate": 8.032738835042068e-06,
      "loss": 0.5979,
      "step": 4953
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0939754247665405,
      "learning_rate": 8.031923057952208e-06,
      "loss": 0.5368,
      "step": 4954
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9942882061004639,
      "learning_rate": 8.031107153198617e-06,
      "loss": 0.5531,
      "step": 4955
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.094529151916504,
      "learning_rate": 8.030291120815647e-06,
      "loss": 0.5452,
      "step": 4956
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1173778772354126,
      "learning_rate": 8.029474960837657e-06,
      "loss": 0.5411,
      "step": 4957
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.025722622871399,
      "learning_rate": 8.028658673299019e-06,
      "loss": 0.5644,
      "step": 4958
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9596103429794312,
      "learning_rate": 8.027842258234097e-06,
      "loss": 0.514,
      "step": 4959
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.069469690322876,
      "learning_rate": 8.027025715677273e-06,
      "loss": 0.6062,
      "step": 4960
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0660532712936401,
      "learning_rate": 8.026209045662925e-06,
      "loss": 0.5804,
      "step": 4961
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0204991102218628,
      "learning_rate": 8.025392248225444e-06,
      "loss": 0.5192,
      "step": 4962
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0006287097930908,
      "learning_rate": 8.024575323399217e-06,
      "loss": 0.5342,
      "step": 4963
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9939424991607666,
      "learning_rate": 8.023758271218646e-06,
      "loss": 0.5538,
      "step": 4964
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.973581075668335,
      "learning_rate": 8.022941091718133e-06,
      "loss": 0.5441,
      "step": 4965
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0039817094802856,
      "learning_rate": 8.022123784932085e-06,
      "loss": 0.5714,
      "step": 4966
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0360323190689087,
      "learning_rate": 8.02130635089492e-06,
      "loss": 0.5494,
      "step": 4967
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0309659242630005,
      "learning_rate": 8.020488789641054e-06,
      "loss": 0.5939,
      "step": 4968
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9577841758728027,
      "learning_rate": 8.019671101204914e-06,
      "loss": 0.5239,
      "step": 4969
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9899508357048035,
      "learning_rate": 8.018853285620926e-06,
      "loss": 0.5017,
      "step": 4970
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0647516250610352,
      "learning_rate": 8.018035342923529e-06,
      "loss": 0.5435,
      "step": 4971
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9604344964027405,
      "learning_rate": 8.017217273147165e-06,
      "loss": 0.4843,
      "step": 4972
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9627401232719421,
      "learning_rate": 8.016399076326275e-06,
      "loss": 0.611,
      "step": 4973
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0629135370254517,
      "learning_rate": 8.015580752495314e-06,
      "loss": 0.5744,
      "step": 4974
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0638678073883057,
      "learning_rate": 8.014762301688737e-06,
      "loss": 0.5838,
      "step": 4975
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9654185175895691,
      "learning_rate": 8.013943723941009e-06,
      "loss": 0.444,
      "step": 4976
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0113495588302612,
      "learning_rate": 8.013125019286594e-06,
      "loss": 0.5789,
      "step": 4977
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0057951211929321,
      "learning_rate": 8.012306187759966e-06,
      "loss": 0.5557,
      "step": 4978
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0672248601913452,
      "learning_rate": 8.011487229395605e-06,
      "loss": 0.579,
      "step": 4979
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9103686213493347,
      "learning_rate": 8.010668144227991e-06,
      "loss": 0.5334,
      "step": 4980
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0784097909927368,
      "learning_rate": 8.009848932291617e-06,
      "loss": 0.505,
      "step": 4981
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0375616550445557,
      "learning_rate": 8.009029593620974e-06,
      "loss": 0.5159,
      "step": 4982
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.013041615486145,
      "learning_rate": 8.008210128250563e-06,
      "loss": 0.56,
      "step": 4983
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0100730657577515,
      "learning_rate": 8.007390536214888e-06,
      "loss": 0.551,
      "step": 4984
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9632449746131897,
      "learning_rate": 8.006570817548457e-06,
      "loss": 0.517,
      "step": 4985
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0761468410491943,
      "learning_rate": 8.005750972285793e-06,
      "loss": 0.5718,
      "step": 4986
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9765868782997131,
      "learning_rate": 8.004931000461408e-06,
      "loss": 0.5141,
      "step": 4987
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0333558320999146,
      "learning_rate": 8.004110902109832e-06,
      "loss": 0.5547,
      "step": 4988
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.029192328453064,
      "learning_rate": 8.003290677265599e-06,
      "loss": 0.5921,
      "step": 4989
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0383979082107544,
      "learning_rate": 8.002470325963241e-06,
      "loss": 0.5385,
      "step": 4990
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9121312499046326,
      "learning_rate": 8.001649848237303e-06,
      "loss": 0.4893,
      "step": 4991
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9828303456306458,
      "learning_rate": 8.000829244122333e-06,
      "loss": 0.5365,
      "step": 4992
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0785502195358276,
      "learning_rate": 8.00000851365288e-06,
      "loss": 0.589,
      "step": 4993
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.953129768371582,
      "learning_rate": 7.999187656863507e-06,
      "loss": 0.5109,
      "step": 4994
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9617950320243835,
      "learning_rate": 7.998366673788775e-06,
      "loss": 0.5067,
      "step": 4995
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0531142950057983,
      "learning_rate": 7.997545564463251e-06,
      "loss": 0.5743,
      "step": 4996
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0768109560012817,
      "learning_rate": 7.996724328921514e-06,
      "loss": 0.4972,
      "step": 4997
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1927998065948486,
      "learning_rate": 7.99590296719814e-06,
      "loss": 0.5797,
      "step": 4998
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0408989191055298,
      "learning_rate": 7.995081479327712e-06,
      "loss": 0.5515,
      "step": 4999
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0196020603179932,
      "learning_rate": 7.994259865344822e-06,
      "loss": 0.5542,
      "step": 5000
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0699565410614014,
      "learning_rate": 7.993438125284068e-06,
      "loss": 0.5788,
      "step": 5001
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0551797151565552,
      "learning_rate": 7.992616259180045e-06,
      "loss": 0.5452,
      "step": 5002
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0628966093063354,
      "learning_rate": 7.991794267067363e-06,
      "loss": 0.5319,
      "step": 5003
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0177741050720215,
      "learning_rate": 7.99097214898063e-06,
      "loss": 0.5379,
      "step": 5004
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9868799448013306,
      "learning_rate": 7.99014990495447e-06,
      "loss": 0.5373,
      "step": 5005
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0062427520751953,
      "learning_rate": 7.989327535023495e-06,
      "loss": 0.5484,
      "step": 5006
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0482970476150513,
      "learning_rate": 7.988505039222339e-06,
      "loss": 0.5368,
      "step": 5007
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0803232192993164,
      "learning_rate": 7.987682417585629e-06,
      "loss": 0.5596,
      "step": 5008
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0112433433532715,
      "learning_rate": 7.98685967014801e-06,
      "loss": 0.5734,
      "step": 5009
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0411220788955688,
      "learning_rate": 7.986036796944116e-06,
      "loss": 0.5594,
      "step": 5010
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0272407531738281,
      "learning_rate": 7.985213798008605e-06,
      "loss": 0.5552,
      "step": 5011
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0483700037002563,
      "learning_rate": 7.984390673376123e-06,
      "loss": 0.5545,
      "step": 5012
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9998469948768616,
      "learning_rate": 7.983567423081331e-06,
      "loss": 0.5405,
      "step": 5013
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9811612963676453,
      "learning_rate": 7.982744047158897e-06,
      "loss": 0.5598,
      "step": 5014
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.082160472869873,
      "learning_rate": 7.981920545643485e-06,
      "loss": 0.5253,
      "step": 5015
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.977547287940979,
      "learning_rate": 7.981096918569773e-06,
      "loss": 0.5051,
      "step": 5016
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0095993280410767,
      "learning_rate": 7.980273165972438e-06,
      "loss": 0.5184,
      "step": 5017
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9310224652290344,
      "learning_rate": 7.979449287886171e-06,
      "loss": 0.4903,
      "step": 5018
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0223238468170166,
      "learning_rate": 7.978625284345657e-06,
      "loss": 0.5627,
      "step": 5019
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9764466881752014,
      "learning_rate": 7.977801155385595e-06,
      "loss": 0.536,
      "step": 5020
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0332834720611572,
      "learning_rate": 7.976976901040686e-06,
      "loss": 0.577,
      "step": 5021
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0004802942276,
      "learning_rate": 7.976152521345635e-06,
      "loss": 0.6022,
      "step": 5022
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1783488988876343,
      "learning_rate": 7.975328016335154e-06,
      "loss": 0.5923,
      "step": 5023
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9814012050628662,
      "learning_rate": 7.974503386043961e-06,
      "loss": 0.5552,
      "step": 5024
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.096435546875,
      "learning_rate": 7.973678630506778e-06,
      "loss": 0.6054,
      "step": 5025
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9659005403518677,
      "learning_rate": 7.972853749758334e-06,
      "loss": 0.5512,
      "step": 5026
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0144753456115723,
      "learning_rate": 7.972028743833357e-06,
      "loss": 0.5553,
      "step": 5027
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0260987281799316,
      "learning_rate": 7.971203612766591e-06,
      "loss": 0.526,
      "step": 5028
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9555620551109314,
      "learning_rate": 7.970378356592779e-06,
      "loss": 0.582,
      "step": 5029
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9632571935653687,
      "learning_rate": 7.969552975346664e-06,
      "loss": 0.5346,
      "step": 5030
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0778241157531738,
      "learning_rate": 7.968727469063005e-06,
      "loss": 0.5058,
      "step": 5031
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.007095456123352,
      "learning_rate": 7.967901837776559e-06,
      "loss": 0.5496,
      "step": 5032
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.007356882095337,
      "learning_rate": 7.967076081522091e-06,
      "loss": 0.5745,
      "step": 5033
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2195940017700195,
      "learning_rate": 7.966250200334373e-06,
      "loss": 0.6314,
      "step": 5034
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0311719179153442,
      "learning_rate": 7.965424194248176e-06,
      "loss": 0.5251,
      "step": 5035
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9696710109710693,
      "learning_rate": 7.964598063298282e-06,
      "loss": 0.5846,
      "step": 5036
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9161157011985779,
      "learning_rate": 7.963771807519477e-06,
      "loss": 0.4878,
      "step": 5037
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9984802603721619,
      "learning_rate": 7.962945426946552e-06,
      "loss": 0.5166,
      "step": 5038
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9821481704711914,
      "learning_rate": 7.962118921614302e-06,
      "loss": 0.5727,
      "step": 5039
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0021934509277344,
      "learning_rate": 7.961292291557529e-06,
      "loss": 0.5063,
      "step": 5040
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0840123891830444,
      "learning_rate": 7.960465536811039e-06,
      "loss": 0.5433,
      "step": 5041
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9790361523628235,
      "learning_rate": 7.959638657409643e-06,
      "loss": 0.5349,
      "step": 5042
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9270578026771545,
      "learning_rate": 7.95881165338816e-06,
      "loss": 0.5016,
      "step": 5043
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.036125898361206,
      "learning_rate": 7.957984524781413e-06,
      "loss": 0.5588,
      "step": 5044
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0060368776321411,
      "learning_rate": 7.957157271624225e-06,
      "loss": 0.5297,
      "step": 5045
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0325381755828857,
      "learning_rate": 7.956329893951432e-06,
      "loss": 0.5324,
      "step": 5046
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1573082208633423,
      "learning_rate": 7.95550239179787e-06,
      "loss": 0.5842,
      "step": 5047
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8682379126548767,
      "learning_rate": 7.954674765198386e-06,
      "loss": 0.4966,
      "step": 5048
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.066186785697937,
      "learning_rate": 7.953847014187826e-06,
      "loss": 0.5796,
      "step": 5049
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0737568140029907,
      "learning_rate": 7.953019138801045e-06,
      "loss": 0.5641,
      "step": 5050
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0351073741912842,
      "learning_rate": 7.952191139072898e-06,
      "loss": 0.574,
      "step": 5051
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9687965512275696,
      "learning_rate": 7.951363015038254e-06,
      "loss": 0.5135,
      "step": 5052
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0956878662109375,
      "learning_rate": 7.950534766731982e-06,
      "loss": 0.5986,
      "step": 5053
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.957947850227356,
      "learning_rate": 7.949706394188951e-06,
      "loss": 0.5402,
      "step": 5054
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0185531377792358,
      "learning_rate": 7.948877897444047e-06,
      "loss": 0.5415,
      "step": 5055
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0777101516723633,
      "learning_rate": 7.948049276532156e-06,
      "loss": 0.5738,
      "step": 5056
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9509685039520264,
      "learning_rate": 7.94722053148816e-06,
      "loss": 0.5157,
      "step": 5057
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0248844623565674,
      "learning_rate": 7.946391662346964e-06,
      "loss": 0.5821,
      "step": 5058
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0210704803466797,
      "learning_rate": 7.945562669143463e-06,
      "loss": 0.5644,
      "step": 5059
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9601178765296936,
      "learning_rate": 7.944733551912566e-06,
      "loss": 0.5246,
      "step": 5060
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9830024242401123,
      "learning_rate": 7.943904310689184e-06,
      "loss": 0.5272,
      "step": 5061
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0547600984573364,
      "learning_rate": 7.94307494550823e-06,
      "loss": 0.5822,
      "step": 5062
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.100956678390503,
      "learning_rate": 7.94224545640463e-06,
      "loss": 0.5451,
      "step": 5063
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9732983708381653,
      "learning_rate": 7.941415843413309e-06,
      "loss": 0.5272,
      "step": 5064
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9543006420135498,
      "learning_rate": 7.940586106569198e-06,
      "loss": 0.5295,
      "step": 5065
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0225476026535034,
      "learning_rate": 7.939756245907237e-06,
      "loss": 0.549,
      "step": 5066
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0729137659072876,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.5776,
      "step": 5067
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0118247270584106,
      "learning_rate": 7.938096153269535e-06,
      "loss": 0.5463,
      "step": 5068
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.978184163570404,
      "learning_rate": 7.937265921363695e-06,
      "loss": 0.4906,
      "step": 5069
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0645402669906616,
      "learning_rate": 7.936435565779806e-06,
      "loss": 0.5251,
      "step": 5070
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0021170377731323,
      "learning_rate": 7.93560508655283e-06,
      "loss": 0.5221,
      "step": 5071
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.007643222808838,
      "learning_rate": 7.934774483717736e-06,
      "loss": 0.5161,
      "step": 5072
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0118756294250488,
      "learning_rate": 7.933943757309498e-06,
      "loss": 0.5745,
      "step": 5073
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0875252485275269,
      "learning_rate": 7.933112907363096e-06,
      "loss": 0.519,
      "step": 5074
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0094273090362549,
      "learning_rate": 7.93228193391351e-06,
      "loss": 0.5552,
      "step": 5075
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0161619186401367,
      "learning_rate": 7.931450836995736e-06,
      "loss": 0.5124,
      "step": 5076
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9365780353546143,
      "learning_rate": 7.930619616644761e-06,
      "loss": 0.5455,
      "step": 5077
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0313904285430908,
      "learning_rate": 7.929788272895591e-06,
      "loss": 0.5447,
      "step": 5078
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0682475566864014,
      "learning_rate": 7.928956805783228e-06,
      "loss": 0.5592,
      "step": 5079
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0047205686569214,
      "learning_rate": 7.928125215342685e-06,
      "loss": 0.5656,
      "step": 5080
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.002293586730957,
      "learning_rate": 7.927293501608975e-06,
      "loss": 0.5491,
      "step": 5081
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9850450158119202,
      "learning_rate": 7.926461664617117e-06,
      "loss": 0.5376,
      "step": 5082
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9095563888549805,
      "learning_rate": 7.92562970440214e-06,
      "loss": 0.5299,
      "step": 5083
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.984972357749939,
      "learning_rate": 7.924797620999074e-06,
      "loss": 0.5106,
      "step": 5084
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1025408506393433,
      "learning_rate": 7.923965414442953e-06,
      "loss": 0.5869,
      "step": 5085
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0418413877487183,
      "learning_rate": 7.923133084768822e-06,
      "loss": 0.5485,
      "step": 5086
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.098783016204834,
      "learning_rate": 7.922300632011726e-06,
      "loss": 0.5581,
      "step": 5087
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9901795387268066,
      "learning_rate": 7.921468056206715e-06,
      "loss": 0.5622,
      "step": 5088
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.031957983970642,
      "learning_rate": 7.920635357388848e-06,
      "loss": 0.4864,
      "step": 5089
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0441521406173706,
      "learning_rate": 7.919802535593185e-06,
      "loss": 0.5574,
      "step": 5090
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9432700872421265,
      "learning_rate": 7.918969590854797e-06,
      "loss": 0.5624,
      "step": 5091
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9714290499687195,
      "learning_rate": 7.91813652320875e-06,
      "loss": 0.5236,
      "step": 5092
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9984144568443298,
      "learning_rate": 7.91730333269013e-06,
      "loss": 0.5561,
      "step": 5093
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0645394325256348,
      "learning_rate": 7.916470019334012e-06,
      "loss": 0.5818,
      "step": 5094
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0106472969055176,
      "learning_rate": 7.915636583175489e-06,
      "loss": 0.5075,
      "step": 5095
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1064532995224,
      "learning_rate": 7.91480302424965e-06,
      "loss": 0.5447,
      "step": 5096
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0424033403396606,
      "learning_rate": 7.913969342591597e-06,
      "loss": 0.5357,
      "step": 5097
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0222606658935547,
      "learning_rate": 7.913135538236432e-06,
      "loss": 0.6008,
      "step": 5098
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1493009328842163,
      "learning_rate": 7.912301611219264e-06,
      "loss": 0.5439,
      "step": 5099
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9946584105491638,
      "learning_rate": 7.911467561575204e-06,
      "loss": 0.5631,
      "step": 5100
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0782363414764404,
      "learning_rate": 7.910633389339376e-06,
      "loss": 0.5791,
      "step": 5101
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9767991304397583,
      "learning_rate": 7.909799094546899e-06,
      "loss": 0.4918,
      "step": 5102
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9834350943565369,
      "learning_rate": 7.908964677232906e-06,
      "loss": 0.5096,
      "step": 5103
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.052124261856079,
      "learning_rate": 7.90813013743253e-06,
      "loss": 0.5607,
      "step": 5104
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1817008256912231,
      "learning_rate": 7.90729547518091e-06,
      "loss": 0.5512,
      "step": 5105
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0303164720535278,
      "learning_rate": 7.906460690513192e-06,
      "loss": 0.532,
      "step": 5106
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.050991177558899,
      "learning_rate": 7.905625783464525e-06,
      "loss": 0.552,
      "step": 5107
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9917927384376526,
      "learning_rate": 7.904790754070063e-06,
      "loss": 0.5332,
      "step": 5108
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9641286134719849,
      "learning_rate": 7.90395560236497e-06,
      "loss": 0.4993,
      "step": 5109
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9856176376342773,
      "learning_rate": 7.903120328384406e-06,
      "loss": 0.5131,
      "step": 5110
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9945058822631836,
      "learning_rate": 7.902284932163545e-06,
      "loss": 0.5193,
      "step": 5111
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0896492004394531,
      "learning_rate": 7.901449413737562e-06,
      "loss": 0.5403,
      "step": 5112
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9672039151191711,
      "learning_rate": 7.90061377314164e-06,
      "loss": 0.5629,
      "step": 5113
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9481355547904968,
      "learning_rate": 7.899778010410958e-06,
      "loss": 0.5466,
      "step": 5114
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9383245706558228,
      "learning_rate": 7.898942125580715e-06,
      "loss": 0.518,
      "step": 5115
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9123684167861938,
      "learning_rate": 7.898106118686102e-06,
      "loss": 0.5126,
      "step": 5116
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0579379796981812,
      "learning_rate": 7.897269989762322e-06,
      "loss": 0.5693,
      "step": 5117
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0428487062454224,
      "learning_rate": 7.896433738844583e-06,
      "loss": 0.5411,
      "step": 5118
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0016083717346191,
      "learning_rate": 7.895597365968093e-06,
      "loss": 0.5569,
      "step": 5119
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9868251085281372,
      "learning_rate": 7.894760871168074e-06,
      "loss": 0.5551,
      "step": 5120
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.116563320159912,
      "learning_rate": 7.893924254479744e-06,
      "loss": 0.5593,
      "step": 5121
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0450432300567627,
      "learning_rate": 7.893087515938329e-06,
      "loss": 0.5684,
      "step": 5122
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0664788484573364,
      "learning_rate": 7.892250655579063e-06,
      "loss": 0.551,
      "step": 5123
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0176259279251099,
      "learning_rate": 7.891413673437185e-06,
      "loss": 0.5314,
      "step": 5124
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0659630298614502,
      "learning_rate": 7.890576569547937e-06,
      "loss": 0.5712,
      "step": 5125
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.021335482597351,
      "learning_rate": 7.889739343946561e-06,
      "loss": 0.5733,
      "step": 5126
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.977240800857544,
      "learning_rate": 7.888901996668317e-06,
      "loss": 0.525,
      "step": 5127
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9783990979194641,
      "learning_rate": 7.888064527748458e-06,
      "loss": 0.5608,
      "step": 5128
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9815640449523926,
      "learning_rate": 7.887226937222252e-06,
      "loss": 0.5808,
      "step": 5129
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9848017692565918,
      "learning_rate": 7.88638922512496e-06,
      "loss": 0.5739,
      "step": 5130
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0390576124191284,
      "learning_rate": 7.88555139149186e-06,
      "loss": 0.5748,
      "step": 5131
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9983201026916504,
      "learning_rate": 7.884713436358228e-06,
      "loss": 0.5368,
      "step": 5132
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0555384159088135,
      "learning_rate": 7.883875359759349e-06,
      "loss": 0.5677,
      "step": 5133
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0259679555892944,
      "learning_rate": 7.883037161730511e-06,
      "loss": 0.5416,
      "step": 5134
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9975308179855347,
      "learning_rate": 7.882198842307008e-06,
      "loss": 0.5452,
      "step": 5135
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0763533115386963,
      "learning_rate": 7.881360401524138e-06,
      "loss": 0.5529,
      "step": 5136
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.104356288909912,
      "learning_rate": 7.880521839417206e-06,
      "loss": 0.5224,
      "step": 5137
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9988602995872498,
      "learning_rate": 7.879683156021518e-06,
      "loss": 0.5517,
      "step": 5138
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1201856136322021,
      "learning_rate": 7.87884435137239e-06,
      "loss": 0.5465,
      "step": 5139
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9202961325645447,
      "learning_rate": 7.878005425505143e-06,
      "loss": 0.5252,
      "step": 5140
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9737326502799988,
      "learning_rate": 7.877166378455098e-06,
      "loss": 0.5632,
      "step": 5141
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0532732009887695,
      "learning_rate": 7.876327210257586e-06,
      "loss": 0.5459,
      "step": 5142
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9994428157806396,
      "learning_rate": 7.875487920947941e-06,
      "loss": 0.5726,
      "step": 5143
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0257031917572021,
      "learning_rate": 7.874648510561503e-06,
      "loss": 0.5696,
      "step": 5144
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0415347814559937,
      "learning_rate": 7.873808979133616e-06,
      "loss": 0.5748,
      "step": 5145
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9889886975288391,
      "learning_rate": 7.872969326699631e-06,
      "loss": 0.494,
      "step": 5146
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0992611646652222,
      "learning_rate": 7.8721295532949e-06,
      "loss": 0.5662,
      "step": 5147
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.073646068572998,
      "learning_rate": 7.871289658954789e-06,
      "loss": 0.5554,
      "step": 5148
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.016093373298645,
      "learning_rate": 7.870449643714654e-06,
      "loss": 0.5469,
      "step": 5149
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9302458763122559,
      "learning_rate": 7.869609507609874e-06,
      "loss": 0.498,
      "step": 5150
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9844838380813599,
      "learning_rate": 7.868769250675818e-06,
      "loss": 0.5399,
      "step": 5151
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8882426619529724,
      "learning_rate": 7.867928872947869e-06,
      "loss": 0.4986,
      "step": 5152
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0245869159698486,
      "learning_rate": 7.867088374461413e-06,
      "loss": 0.5663,
      "step": 5153
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9834217429161072,
      "learning_rate": 7.866247755251838e-06,
      "loss": 0.5518,
      "step": 5154
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9708665609359741,
      "learning_rate": 7.865407015354542e-06,
      "loss": 0.514,
      "step": 5155
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0061088800430298,
      "learning_rate": 7.864566154804925e-06,
      "loss": 0.5584,
      "step": 5156
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0202516317367554,
      "learning_rate": 7.86372517363839e-06,
      "loss": 0.504,
      "step": 5157
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0482490062713623,
      "learning_rate": 7.862884071890353e-06,
      "loss": 0.5764,
      "step": 5158
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9607723355293274,
      "learning_rate": 7.862042849596225e-06,
      "loss": 0.5018,
      "step": 5159
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9499772787094116,
      "learning_rate": 7.86120150679143e-06,
      "loss": 0.563,
      "step": 5160
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9344480633735657,
      "learning_rate": 7.860360043511392e-06,
      "loss": 0.5223,
      "step": 5161
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9974209070205688,
      "learning_rate": 7.859518459791543e-06,
      "loss": 0.5606,
      "step": 5162
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0160338878631592,
      "learning_rate": 7.85867675566732e-06,
      "loss": 0.5219,
      "step": 5163
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9889779090881348,
      "learning_rate": 7.857834931174164e-06,
      "loss": 0.5416,
      "step": 5164
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9936977028846741,
      "learning_rate": 7.85699298634752e-06,
      "loss": 0.5473,
      "step": 5165
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0194740295410156,
      "learning_rate": 7.856150921222838e-06,
      "loss": 0.5352,
      "step": 5166
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9760123491287231,
      "learning_rate": 7.85530873583558e-06,
      "loss": 0.5736,
      "step": 5167
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9568659067153931,
      "learning_rate": 7.854466430221203e-06,
      "loss": 0.5255,
      "step": 5168
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9879942536354065,
      "learning_rate": 7.853624004415172e-06,
      "loss": 0.5914,
      "step": 5169
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9788195490837097,
      "learning_rate": 7.852781458452964e-06,
      "loss": 0.5409,
      "step": 5170
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0303471088409424,
      "learning_rate": 7.851938792370053e-06,
      "loss": 0.5852,
      "step": 5171
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8927707076072693,
      "learning_rate": 7.85109600620192e-06,
      "loss": 0.4724,
      "step": 5172
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1031830310821533,
      "learning_rate": 7.85025309998405e-06,
      "loss": 0.5288,
      "step": 5173
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9610477089881897,
      "learning_rate": 7.849410073751942e-06,
      "loss": 0.4959,
      "step": 5174
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0511902570724487,
      "learning_rate": 7.848566927541084e-06,
      "loss": 0.533,
      "step": 5175
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9335895776748657,
      "learning_rate": 7.847723661386985e-06,
      "loss": 0.5203,
      "step": 5176
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0692973136901855,
      "learning_rate": 7.846880275325149e-06,
      "loss": 0.5507,
      "step": 5177
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9702898859977722,
      "learning_rate": 7.846036769391086e-06,
      "loss": 0.5529,
      "step": 5178
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0902308225631714,
      "learning_rate": 7.845193143620316e-06,
      "loss": 0.5497,
      "step": 5179
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0889067649841309,
      "learning_rate": 7.84434939804836e-06,
      "loss": 0.5724,
      "step": 5180
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1026750802993774,
      "learning_rate": 7.843505532710748e-06,
      "loss": 0.5676,
      "step": 5181
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.019676923751831,
      "learning_rate": 7.84266154764301e-06,
      "loss": 0.5744,
      "step": 5182
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9964596629142761,
      "learning_rate": 7.84181744288068e-06,
      "loss": 0.5693,
      "step": 5183
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9298571348190308,
      "learning_rate": 7.840973218459305e-06,
      "loss": 0.4892,
      "step": 5184
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0911149978637695,
      "learning_rate": 7.84012887441443e-06,
      "loss": 0.5437,
      "step": 5185
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1494719982147217,
      "learning_rate": 7.839284410781609e-06,
      "loss": 0.5176,
      "step": 5186
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.035965085029602,
      "learning_rate": 7.838439827596398e-06,
      "loss": 0.5471,
      "step": 5187
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.102126121520996,
      "learning_rate": 7.83759512489436e-06,
      "loss": 0.5752,
      "step": 5188
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9802284240722656,
      "learning_rate": 7.836750302711065e-06,
      "loss": 0.5237,
      "step": 5189
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9825965762138367,
      "learning_rate": 7.83590536108208e-06,
      "loss": 0.522,
      "step": 5190
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.017299771308899,
      "learning_rate": 7.835060300042986e-06,
      "loss": 0.5584,
      "step": 5191
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.996579110622406,
      "learning_rate": 7.834215119629366e-06,
      "loss": 0.5229,
      "step": 5192
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.006910800933838,
      "learning_rate": 7.833369819876809e-06,
      "loss": 0.5416,
      "step": 5193
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0680913925170898,
      "learning_rate": 7.832524400820902e-06,
      "loss": 0.5333,
      "step": 5194
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0967390537261963,
      "learning_rate": 7.831678862497248e-06,
      "loss": 0.5562,
      "step": 5195
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.077828288078308,
      "learning_rate": 7.830833204941446e-06,
      "loss": 0.5854,
      "step": 5196
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9674387574195862,
      "learning_rate": 7.829987428189108e-06,
      "loss": 0.5844,
      "step": 5197
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9557288289070129,
      "learning_rate": 7.829141532275843e-06,
      "loss": 0.5209,
      "step": 5198
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9728726744651794,
      "learning_rate": 7.82829551723727e-06,
      "loss": 0.5654,
      "step": 5199
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9683870077133179,
      "learning_rate": 7.827449383109012e-06,
      "loss": 0.5174,
      "step": 5200
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0592831373214722,
      "learning_rate": 7.826603129926696e-06,
      "loss": 0.5505,
      "step": 5201
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9644933342933655,
      "learning_rate": 7.825756757725956e-06,
      "loss": 0.5594,
      "step": 5202
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.012763261795044,
      "learning_rate": 7.824910266542426e-06,
      "loss": 0.5953,
      "step": 5203
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9749206304550171,
      "learning_rate": 7.824063656411756e-06,
      "loss": 0.4719,
      "step": 5204
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.92559415102005,
      "learning_rate": 7.823216927369588e-06,
      "loss": 0.5293,
      "step": 5205
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0100061893463135,
      "learning_rate": 7.822370079451576e-06,
      "loss": 0.5859,
      "step": 5206
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9328200221061707,
      "learning_rate": 7.821523112693377e-06,
      "loss": 0.509,
      "step": 5207
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9542256593704224,
      "learning_rate": 7.820676027130657e-06,
      "loss": 0.4838,
      "step": 5208
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9602751135826111,
      "learning_rate": 7.81982882279908e-06,
      "loss": 0.5223,
      "step": 5209
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0051568746566772,
      "learning_rate": 7.818981499734323e-06,
      "loss": 0.5252,
      "step": 5210
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9804307222366333,
      "learning_rate": 7.818134057972062e-06,
      "loss": 0.516,
      "step": 5211
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0683549642562866,
      "learning_rate": 7.817286497547977e-06,
      "loss": 0.5317,
      "step": 5212
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.01795494556427,
      "learning_rate": 7.81643881849776e-06,
      "loss": 0.562,
      "step": 5213
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0241423845291138,
      "learning_rate": 7.815591020857101e-06,
      "loss": 0.5633,
      "step": 5214
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.044502854347229,
      "learning_rate": 7.8147431046617e-06,
      "loss": 0.5379,
      "step": 5215
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.967847466468811,
      "learning_rate": 7.813895069947257e-06,
      "loss": 0.5734,
      "step": 5216
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9767869710922241,
      "learning_rate": 7.813046916749483e-06,
      "loss": 0.5801,
      "step": 5217
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1450245380401611,
      "learning_rate": 7.812198645104088e-06,
      "loss": 0.5427,
      "step": 5218
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0699288845062256,
      "learning_rate": 7.811350255046792e-06,
      "loss": 0.5262,
      "step": 5219
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0704792737960815,
      "learning_rate": 7.810501746613316e-06,
      "loss": 0.514,
      "step": 5220
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0542374849319458,
      "learning_rate": 7.809653119839389e-06,
      "loss": 0.5324,
      "step": 5221
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0917741060256958,
      "learning_rate": 7.808804374760742e-06,
      "loss": 0.5393,
      "step": 5222
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0410189628601074,
      "learning_rate": 7.807955511413114e-06,
      "loss": 0.5527,
      "step": 5223
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.987148106098175,
      "learning_rate": 7.80710652983225e-06,
      "loss": 0.4854,
      "step": 5224
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.000052809715271,
      "learning_rate": 7.806257430053893e-06,
      "loss": 0.5382,
      "step": 5225
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1064789295196533,
      "learning_rate": 7.8054082121138e-06,
      "loss": 0.5361,
      "step": 5226
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9538825750350952,
      "learning_rate": 7.804558876047724e-06,
      "loss": 0.536,
      "step": 5227
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.048471450805664,
      "learning_rate": 7.80370942189143e-06,
      "loss": 0.5302,
      "step": 5228
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.125544548034668,
      "learning_rate": 7.802859849680686e-06,
      "loss": 0.5418,
      "step": 5229
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9804906845092773,
      "learning_rate": 7.802010159451267e-06,
      "loss": 0.5573,
      "step": 5230
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0753605365753174,
      "learning_rate": 7.801160351238945e-06,
      "loss": 0.5406,
      "step": 5231
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.953654944896698,
      "learning_rate": 7.800310425079505e-06,
      "loss": 0.4982,
      "step": 5232
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9982483983039856,
      "learning_rate": 7.799460381008736e-06,
      "loss": 0.5012,
      "step": 5233
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.003063678741455,
      "learning_rate": 7.798610219062428e-06,
      "loss": 0.5596,
      "step": 5234
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0453439950942993,
      "learning_rate": 7.79775993927638e-06,
      "loss": 0.5208,
      "step": 5235
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.057268500328064,
      "learning_rate": 7.796909541686392e-06,
      "loss": 0.4797,
      "step": 5236
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9837533831596375,
      "learning_rate": 7.796059026328274e-06,
      "loss": 0.5483,
      "step": 5237
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.974611222743988,
      "learning_rate": 7.795208393237839e-06,
      "loss": 0.511,
      "step": 5238
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0048149824142456,
      "learning_rate": 7.794357642450899e-06,
      "loss": 0.5638,
      "step": 5239
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0645167827606201,
      "learning_rate": 7.793506774003282e-06,
      "loss": 0.5545,
      "step": 5240
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9813336730003357,
      "learning_rate": 7.792655787930811e-06,
      "loss": 0.5464,
      "step": 5241
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0172266960144043,
      "learning_rate": 7.791804684269322e-06,
      "loss": 0.5456,
      "step": 5242
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.032725214958191,
      "learning_rate": 7.790953463054647e-06,
      "loss": 0.5566,
      "step": 5243
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9283890128135681,
      "learning_rate": 7.790102124322633e-06,
      "loss": 0.5315,
      "step": 5244
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0746551752090454,
      "learning_rate": 7.789250668109124e-06,
      "loss": 0.5924,
      "step": 5245
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0125327110290527,
      "learning_rate": 7.788399094449971e-06,
      "loss": 0.5477,
      "step": 5246
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9725445508956909,
      "learning_rate": 7.787547403381033e-06,
      "loss": 0.512,
      "step": 5247
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0623197555541992,
      "learning_rate": 7.786695594938172e-06,
      "loss": 0.5346,
      "step": 5248
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1318409442901611,
      "learning_rate": 7.785843669157253e-06,
      "loss": 0.588,
      "step": 5249
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0636366605758667,
      "learning_rate": 7.784991626074148e-06,
      "loss": 0.6046,
      "step": 5250
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0380812883377075,
      "learning_rate": 7.784139465724734e-06,
      "loss": 0.5747,
      "step": 5251
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9197916388511658,
      "learning_rate": 7.783287188144893e-06,
      "loss": 0.5228,
      "step": 5252
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9440740942955017,
      "learning_rate": 7.78243479337051e-06,
      "loss": 0.5606,
      "step": 5253
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9165782332420349,
      "learning_rate": 7.781582281437479e-06,
      "loss": 0.5276,
      "step": 5254
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1359775066375732,
      "learning_rate": 7.780729652381694e-06,
      "loss": 0.5725,
      "step": 5255
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9660843014717102,
      "learning_rate": 7.779876906239055e-06,
      "loss": 0.5514,
      "step": 5256
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0322363376617432,
      "learning_rate": 7.779024043045471e-06,
      "loss": 0.5686,
      "step": 5257
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0766687393188477,
      "learning_rate": 7.778171062836853e-06,
      "loss": 0.5995,
      "step": 5258
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1299717426300049,
      "learning_rate": 7.777317965649114e-06,
      "loss": 0.5667,
      "step": 5259
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9850207567214966,
      "learning_rate": 7.776464751518177e-06,
      "loss": 0.5553,
      "step": 5260
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0228867530822754,
      "learning_rate": 7.775611420479971e-06,
      "loss": 0.4989,
      "step": 5261
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.973818838596344,
      "learning_rate": 7.774757972570423e-06,
      "loss": 0.5209,
      "step": 5262
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9740568995475769,
      "learning_rate": 7.773904407825467e-06,
      "loss": 0.5751,
      "step": 5263
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.050536870956421,
      "learning_rate": 7.773050726281048e-06,
      "loss": 0.5365,
      "step": 5264
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1395479440689087,
      "learning_rate": 7.772196927973109e-06,
      "loss": 0.5459,
      "step": 5265
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0125073194503784,
      "learning_rate": 7.771343012937602e-06,
      "loss": 0.5437,
      "step": 5266
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.010521650314331,
      "learning_rate": 7.77048898121048e-06,
      "loss": 0.5383,
      "step": 5267
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0275487899780273,
      "learning_rate": 7.769634832827706e-06,
      "loss": 0.5211,
      "step": 5268
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.042787790298462,
      "learning_rate": 7.768780567825243e-06,
      "loss": 0.53,
      "step": 5269
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0025073289871216,
      "learning_rate": 7.767926186239064e-06,
      "loss": 0.5722,
      "step": 5270
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.065557599067688,
      "learning_rate": 7.76707168810514e-06,
      "loss": 0.567,
      "step": 5271
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9811010956764221,
      "learning_rate": 7.766217073459454e-06,
      "loss": 0.5216,
      "step": 5272
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9990840554237366,
      "learning_rate": 7.765362342337991e-06,
      "loss": 0.5442,
      "step": 5273
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9401424527168274,
      "learning_rate": 7.76450749477674e-06,
      "loss": 0.5128,
      "step": 5274
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9551236629486084,
      "learning_rate": 7.763652530811692e-06,
      "loss": 0.5355,
      "step": 5275
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9799797534942627,
      "learning_rate": 7.762797450478853e-06,
      "loss": 0.5377,
      "step": 5276
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9672342538833618,
      "learning_rate": 7.761942253814225e-06,
      "loss": 0.5489,
      "step": 5277
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9345821738243103,
      "learning_rate": 7.761086940853814e-06,
      "loss": 0.4499,
      "step": 5278
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0644389390945435,
      "learning_rate": 7.76023151163364e-06,
      "loss": 0.5205,
      "step": 5279
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0195364952087402,
      "learning_rate": 7.759375966189718e-06,
      "loss": 0.5371,
      "step": 5280
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0258504152297974,
      "learning_rate": 7.758520304558072e-06,
      "loss": 0.5222,
      "step": 5281
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0528091192245483,
      "learning_rate": 7.757664526774733e-06,
      "loss": 0.5424,
      "step": 5282
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0659440755844116,
      "learning_rate": 7.756808632875737e-06,
      "loss": 0.5677,
      "step": 5283
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.067395806312561,
      "learning_rate": 7.755952622897117e-06,
      "loss": 0.5145,
      "step": 5284
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0888980627059937,
      "learning_rate": 7.755096496874918e-06,
      "loss": 0.537,
      "step": 5285
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9735004305839539,
      "learning_rate": 7.75424025484519e-06,
      "loss": 0.5465,
      "step": 5286
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0772993564605713,
      "learning_rate": 7.753383896843988e-06,
      "loss": 0.5832,
      "step": 5287
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9806780219078064,
      "learning_rate": 7.752527422907368e-06,
      "loss": 0.5101,
      "step": 5288
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9502233862876892,
      "learning_rate": 7.751670833071393e-06,
      "loss": 0.5332,
      "step": 5289
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0269991159439087,
      "learning_rate": 7.750814127372131e-06,
      "loss": 0.4955,
      "step": 5290
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9721434712409973,
      "learning_rate": 7.749957305845656e-06,
      "loss": 0.5073,
      "step": 5291
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0321543216705322,
      "learning_rate": 7.749100368528047e-06,
      "loss": 0.5809,
      "step": 5292
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.013351321220398,
      "learning_rate": 7.748243315455382e-06,
      "loss": 0.552,
      "step": 5293
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9572513103485107,
      "learning_rate": 7.747386146663753e-06,
      "loss": 0.5345,
      "step": 5294
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9991185665130615,
      "learning_rate": 7.746528862189251e-06,
      "loss": 0.5737,
      "step": 5295
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0627458095550537,
      "learning_rate": 7.745671462067974e-06,
      "loss": 0.5519,
      "step": 5296
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0270791053771973,
      "learning_rate": 7.74481394633602e-06,
      "loss": 0.5641,
      "step": 5297
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.138097882270813,
      "learning_rate": 7.743956315029502e-06,
      "loss": 0.5608,
      "step": 5298
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0272469520568848,
      "learning_rate": 7.743098568184529e-06,
      "loss": 0.5283,
      "step": 5299
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0091753005981445,
      "learning_rate": 7.742240705837217e-06,
      "loss": 0.5103,
      "step": 5300
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0320264101028442,
      "learning_rate": 7.741382728023687e-06,
      "loss": 0.5482,
      "step": 5301
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9640737175941467,
      "learning_rate": 7.74052463478007e-06,
      "loss": 0.5287,
      "step": 5302
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.966127872467041,
      "learning_rate": 7.739666426142493e-06,
      "loss": 0.5348,
      "step": 5303
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0670241117477417,
      "learning_rate": 7.738808102147093e-06,
      "loss": 0.5128,
      "step": 5304
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.044314980506897,
      "learning_rate": 7.737949662830012e-06,
      "loss": 0.5914,
      "step": 5305
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0198527574539185,
      "learning_rate": 7.737091108227395e-06,
      "loss": 0.5561,
      "step": 5306
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.060970664024353,
      "learning_rate": 7.736232438375391e-06,
      "loss": 0.5686,
      "step": 5307
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.951311469078064,
      "learning_rate": 7.735373653310161e-06,
      "loss": 0.5144,
      "step": 5308
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0283550024032593,
      "learning_rate": 7.73451475306786e-06,
      "loss": 0.5794,
      "step": 5309
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9955611824989319,
      "learning_rate": 7.733655737684657e-06,
      "loss": 0.5263,
      "step": 5310
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0055062770843506,
      "learning_rate": 7.732796607196719e-06,
      "loss": 0.5571,
      "step": 5311
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0637763738632202,
      "learning_rate": 7.731937361640223e-06,
      "loss": 0.5657,
      "step": 5312
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9737985730171204,
      "learning_rate": 7.73107800105135e-06,
      "loss": 0.543,
      "step": 5313
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0452426671981812,
      "learning_rate": 7.730218525466283e-06,
      "loss": 0.5642,
      "step": 5314
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9778900742530823,
      "learning_rate": 7.729358934921209e-06,
      "loss": 0.5404,
      "step": 5315
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0338850021362305,
      "learning_rate": 7.728499229452326e-06,
      "loss": 0.514,
      "step": 5316
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.054598331451416,
      "learning_rate": 7.727639409095833e-06,
      "loss": 0.5899,
      "step": 5317
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.118851661682129,
      "learning_rate": 7.726779473887933e-06,
      "loss": 0.5827,
      "step": 5318
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0014888048171997,
      "learning_rate": 7.725919423864837e-06,
      "loss": 0.5633,
      "step": 5319
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0067678689956665,
      "learning_rate": 7.725059259062753e-06,
      "loss": 0.5687,
      "step": 5320
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0196219682693481,
      "learning_rate": 7.724198979517905e-06,
      "loss": 0.5275,
      "step": 5321
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0631487369537354,
      "learning_rate": 7.723338585266515e-06,
      "loss": 0.5602,
      "step": 5322
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1154228448867798,
      "learning_rate": 7.722478076344812e-06,
      "loss": 0.5358,
      "step": 5323
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.059723138809204,
      "learning_rate": 7.721617452789028e-06,
      "loss": 0.5893,
      "step": 5324
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0555039644241333,
      "learning_rate": 7.7207567146354e-06,
      "loss": 0.5798,
      "step": 5325
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.053282618522644,
      "learning_rate": 7.71989586192017e-06,
      "loss": 0.5244,
      "step": 5326
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.051269769668579,
      "learning_rate": 7.719034894679589e-06,
      "loss": 0.5534,
      "step": 5327
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.103200078010559,
      "learning_rate": 7.718173812949908e-06,
      "loss": 0.53,
      "step": 5328
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0847004652023315,
      "learning_rate": 7.717312616767382e-06,
      "loss": 0.5591,
      "step": 5329
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9675300717353821,
      "learning_rate": 7.716451306168276e-06,
      "loss": 0.5453,
      "step": 5330
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9376102685928345,
      "learning_rate": 7.715589881188852e-06,
      "loss": 0.5355,
      "step": 5331
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0276098251342773,
      "learning_rate": 7.71472834186539e-06,
      "loss": 0.5211,
      "step": 5332
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0249872207641602,
      "learning_rate": 7.713866688234157e-06,
      "loss": 0.5386,
      "step": 5333
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.946427583694458,
      "learning_rate": 7.713004920331441e-06,
      "loss": 0.5545,
      "step": 5334
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9513344168663025,
      "learning_rate": 7.712143038193525e-06,
      "loss": 0.508,
      "step": 5335
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.002827763557434,
      "learning_rate": 7.7112810418567e-06,
      "loss": 0.528,
      "step": 5336
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0068122148513794,
      "learning_rate": 7.710418931357263e-06,
      "loss": 0.553,
      "step": 5337
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.025156855583191,
      "learning_rate": 7.709556706731514e-06,
      "loss": 0.5589,
      "step": 5338
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.054255723953247,
      "learning_rate": 7.708694368015758e-06,
      "loss": 0.5058,
      "step": 5339
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8967742919921875,
      "learning_rate": 7.707831915246304e-06,
      "loss": 0.5334,
      "step": 5340
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0459626913070679,
      "learning_rate": 7.706969348459469e-06,
      "loss": 0.5869,
      "step": 5341
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0625965595245361,
      "learning_rate": 7.70610666769157e-06,
      "loss": 0.6205,
      "step": 5342
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9735672473907471,
      "learning_rate": 7.705243872978935e-06,
      "loss": 0.5298,
      "step": 5343
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9772914052009583,
      "learning_rate": 7.704380964357889e-06,
      "loss": 0.5148,
      "step": 5344
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0570646524429321,
      "learning_rate": 7.70351794186477e-06,
      "loss": 0.5635,
      "step": 5345
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.028130054473877,
      "learning_rate": 7.702654805535915e-06,
      "loss": 0.5314,
      "step": 5346
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.987157940864563,
      "learning_rate": 7.701791555407669e-06,
      "loss": 0.5216,
      "step": 5347
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0099393129348755,
      "learning_rate": 7.700928191516378e-06,
      "loss": 0.5943,
      "step": 5348
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9394066333770752,
      "learning_rate": 7.700064713898398e-06,
      "loss": 0.5601,
      "step": 5349
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0110291242599487,
      "learning_rate": 7.699201122590086e-06,
      "loss": 0.5385,
      "step": 5350
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0360554456710815,
      "learning_rate": 7.6983374176278e-06,
      "loss": 0.5548,
      "step": 5351
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9967292547225952,
      "learning_rate": 7.697473599047918e-06,
      "loss": 0.5413,
      "step": 5352
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9504690766334534,
      "learning_rate": 7.696609666886805e-06,
      "loss": 0.5463,
      "step": 5353
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9553558826446533,
      "learning_rate": 7.695745621180839e-06,
      "loss": 0.4722,
      "step": 5354
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9541303515434265,
      "learning_rate": 7.694881461966402e-06,
      "loss": 0.5693,
      "step": 5355
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.013753056526184,
      "learning_rate": 7.694017189279882e-06,
      "loss": 0.5445,
      "step": 5356
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9317124485969543,
      "learning_rate": 7.69315280315767e-06,
      "loss": 0.4885,
      "step": 5357
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0442945957183838,
      "learning_rate": 7.692288303636163e-06,
      "loss": 0.5415,
      "step": 5358
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9768821597099304,
      "learning_rate": 7.69142369075176e-06,
      "loss": 0.5353,
      "step": 5359
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9622286558151245,
      "learning_rate": 7.690558964540872e-06,
      "loss": 0.5459,
      "step": 5360
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0875238180160522,
      "learning_rate": 7.6896941250399e-06,
      "loss": 0.533,
      "step": 5361
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0347795486450195,
      "learning_rate": 7.688829172285267e-06,
      "loss": 0.5402,
      "step": 5362
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9879282712936401,
      "learning_rate": 7.687964106313392e-06,
      "loss": 0.5114,
      "step": 5363
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0509141683578491,
      "learning_rate": 7.687098927160701e-06,
      "loss": 0.5641,
      "step": 5364
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0029528141021729,
      "learning_rate": 7.68623363486362e-06,
      "loss": 0.569,
      "step": 5365
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0377953052520752,
      "learning_rate": 7.685368229458584e-06,
      "loss": 0.5288,
      "step": 5366
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0632033348083496,
      "learning_rate": 7.684502710982035e-06,
      "loss": 0.5371,
      "step": 5367
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9152517914772034,
      "learning_rate": 7.683637079470418e-06,
      "loss": 0.5469,
      "step": 5368
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9898026585578918,
      "learning_rate": 7.682771334960178e-06,
      "loss": 0.5199,
      "step": 5369
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9831115007400513,
      "learning_rate": 7.681905477487769e-06,
      "loss": 0.5911,
      "step": 5370
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8863547444343567,
      "learning_rate": 7.68103950708965e-06,
      "loss": 0.4814,
      "step": 5371
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9888505935668945,
      "learning_rate": 7.680173423802282e-06,
      "loss": 0.5537,
      "step": 5372
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0175485610961914,
      "learning_rate": 7.679307227662136e-06,
      "loss": 0.5367,
      "step": 5373
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0653303861618042,
      "learning_rate": 7.678440918705686e-06,
      "loss": 0.5694,
      "step": 5374
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.096785545349121,
      "learning_rate": 7.677574496969404e-06,
      "loss": 0.5849,
      "step": 5375
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0689427852630615,
      "learning_rate": 7.676707962489775e-06,
      "loss": 0.5733,
      "step": 5376
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0037180185317993,
      "learning_rate": 7.675841315303284e-06,
      "loss": 0.6059,
      "step": 5377
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9984425902366638,
      "learning_rate": 7.674974555446425e-06,
      "loss": 0.5509,
      "step": 5378
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9228800535202026,
      "learning_rate": 7.674107682955693e-06,
      "loss": 0.5234,
      "step": 5379
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0302813053131104,
      "learning_rate": 7.67324069786759e-06,
      "loss": 0.5437,
      "step": 5380
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0414074659347534,
      "learning_rate": 7.67237360021862e-06,
      "loss": 0.5869,
      "step": 5381
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0754034519195557,
      "learning_rate": 7.671506390045293e-06,
      "loss": 0.5862,
      "step": 5382
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.090884804725647,
      "learning_rate": 7.670639067384126e-06,
      "loss": 0.5666,
      "step": 5383
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.976880669593811,
      "learning_rate": 7.66977163227164e-06,
      "loss": 0.5274,
      "step": 5384
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9574406147003174,
      "learning_rate": 7.668904084744357e-06,
      "loss": 0.5016,
      "step": 5385
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9235224723815918,
      "learning_rate": 7.668036424838808e-06,
      "loss": 0.5219,
      "step": 5386
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9787632822990417,
      "learning_rate": 7.667168652591524e-06,
      "loss": 0.5527,
      "step": 5387
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0407617092132568,
      "learning_rate": 7.66630076803905e-06,
      "loss": 0.5838,
      "step": 5388
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9891582727432251,
      "learning_rate": 7.665432771217922e-06,
      "loss": 0.5295,
      "step": 5389
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0179615020751953,
      "learning_rate": 7.664564662164696e-06,
      "loss": 0.5429,
      "step": 5390
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1185835599899292,
      "learning_rate": 7.66369644091592e-06,
      "loss": 0.5839,
      "step": 5391
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9839662313461304,
      "learning_rate": 7.662828107508153e-06,
      "loss": 0.537,
      "step": 5392
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.948878824710846,
      "learning_rate": 7.661959661977958e-06,
      "loss": 0.5315,
      "step": 5393
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1196187734603882,
      "learning_rate": 7.661091104361902e-06,
      "loss": 0.566,
      "step": 5394
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9894713163375854,
      "learning_rate": 7.660222434696556e-06,
      "loss": 0.5725,
      "step": 5395
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.982667088508606,
      "learning_rate": 7.6593536530185e-06,
      "loss": 0.5424,
      "step": 5396
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.060299038887024,
      "learning_rate": 7.658484759364308e-06,
      "loss": 0.5497,
      "step": 5397
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.089518666267395,
      "learning_rate": 7.657615753770575e-06,
      "loss": 0.573,
      "step": 5398
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8669412136077881,
      "learning_rate": 7.656746636273889e-06,
      "loss": 0.4693,
      "step": 5399
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.986979603767395,
      "learning_rate": 7.655877406910841e-06,
      "loss": 0.5226,
      "step": 5400
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1012288331985474,
      "learning_rate": 7.655008065718036e-06,
      "loss": 0.5856,
      "step": 5401
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.949129045009613,
      "learning_rate": 7.654138612732078e-06,
      "loss": 0.5477,
      "step": 5402
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9540390372276306,
      "learning_rate": 7.653269047989575e-06,
      "loss": 0.529,
      "step": 5403
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9496141076087952,
      "learning_rate": 7.652399371527142e-06,
      "loss": 0.5673,
      "step": 5404
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.135327696800232,
      "learning_rate": 7.651529583381398e-06,
      "loss": 0.5832,
      "step": 5405
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0860304832458496,
      "learning_rate": 7.65065968358897e-06,
      "loss": 0.6126,
      "step": 5406
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0657199621200562,
      "learning_rate": 7.649789672186483e-06,
      "loss": 0.5898,
      "step": 5407
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0875470638275146,
      "learning_rate": 7.648919549210567e-06,
      "loss": 0.5724,
      "step": 5408
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9945139288902283,
      "learning_rate": 7.648049314697869e-06,
      "loss": 0.495,
      "step": 5409
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.04204523563385,
      "learning_rate": 7.647178968685024e-06,
      "loss": 0.5271,
      "step": 5410
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.11293625831604,
      "learning_rate": 7.646308511208682e-06,
      "loss": 0.6148,
      "step": 5411
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.973831057548523,
      "learning_rate": 7.645437942305491e-06,
      "loss": 0.5045,
      "step": 5412
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0557979345321655,
      "learning_rate": 7.644567262012115e-06,
      "loss": 0.5711,
      "step": 5413
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0014163255691528,
      "learning_rate": 7.643696470365209e-06,
      "loss": 0.562,
      "step": 5414
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0061209201812744,
      "learning_rate": 7.642825567401444e-06,
      "loss": 0.5078,
      "step": 5415
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.074394702911377,
      "learning_rate": 7.641954553157487e-06,
      "loss": 0.5228,
      "step": 5416
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9882181286811829,
      "learning_rate": 7.641083427670014e-06,
      "loss": 0.5472,
      "step": 5417
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0100023746490479,
      "learning_rate": 7.640212190975707e-06,
      "loss": 0.5354,
      "step": 5418
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0182669162750244,
      "learning_rate": 7.639340843111247e-06,
      "loss": 0.5168,
      "step": 5419
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8914739489555359,
      "learning_rate": 7.638469384113328e-06,
      "loss": 0.4661,
      "step": 5420
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.998095691204071,
      "learning_rate": 7.637597814018638e-06,
      "loss": 0.5238,
      "step": 5421
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0295097827911377,
      "learning_rate": 7.636726132863883e-06,
      "loss": 0.5437,
      "step": 5422
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.030522346496582,
      "learning_rate": 7.635854340685762e-06,
      "loss": 0.541,
      "step": 5423
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0026613473892212,
      "learning_rate": 7.634982437520984e-06,
      "loss": 0.5135,
      "step": 5424
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9932113289833069,
      "learning_rate": 7.634110423406262e-06,
      "loss": 0.5158,
      "step": 5425
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0206046104431152,
      "learning_rate": 7.633238298378315e-06,
      "loss": 0.5665,
      "step": 5426
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0064122676849365,
      "learning_rate": 7.632366062473862e-06,
      "loss": 0.5221,
      "step": 5427
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0268290042877197,
      "learning_rate": 7.631493715729632e-06,
      "loss": 0.5089,
      "step": 5428
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0285868644714355,
      "learning_rate": 7.630621258182354e-06,
      "loss": 0.5365,
      "step": 5429
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9036509990692139,
      "learning_rate": 7.62974868986877e-06,
      "loss": 0.5043,
      "step": 5430
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.057509183883667,
      "learning_rate": 7.628876010825614e-06,
      "loss": 0.5889,
      "step": 5431
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.924913227558136,
      "learning_rate": 7.628003221089635e-06,
      "loss": 0.4883,
      "step": 5432
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1624161005020142,
      "learning_rate": 7.6271303206975825e-06,
      "loss": 0.5788,
      "step": 5433
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0000855922698975,
      "learning_rate": 7.626257309686211e-06,
      "loss": 0.531,
      "step": 5434
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.98502117395401,
      "learning_rate": 7.6253841880922805e-06,
      "loss": 0.533,
      "step": 5435
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1036839485168457,
      "learning_rate": 7.624510955952555e-06,
      "loss": 0.515,
      "step": 5436
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0275412797927856,
      "learning_rate": 7.623637613303805e-06,
      "loss": 0.5339,
      "step": 5437
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9513323307037354,
      "learning_rate": 7.6227641601827996e-06,
      "loss": 0.4962,
      "step": 5438
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9934118986129761,
      "learning_rate": 7.62189059662632e-06,
      "loss": 0.5453,
      "step": 5439
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0119845867156982,
      "learning_rate": 7.621016922671147e-06,
      "loss": 0.5397,
      "step": 5440
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9635437726974487,
      "learning_rate": 7.620143138354072e-06,
      "loss": 0.5361,
      "step": 5441
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.966569185256958,
      "learning_rate": 7.6192692437118825e-06,
      "loss": 0.5614,
      "step": 5442
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.117817759513855,
      "learning_rate": 7.618395238781377e-06,
      "loss": 0.6088,
      "step": 5443
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0499943494796753,
      "learning_rate": 7.617521123599356e-06,
      "loss": 0.5452,
      "step": 5444
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0381006002426147,
      "learning_rate": 7.616646898202629e-06,
      "loss": 0.565,
      "step": 5445
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.005886435508728,
      "learning_rate": 7.6157725626280014e-06,
      "loss": 0.4925,
      "step": 5446
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9612359404563904,
      "learning_rate": 7.61489811691229e-06,
      "loss": 0.5114,
      "step": 5447
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0364564657211304,
      "learning_rate": 7.614023561092319e-06,
      "loss": 0.5505,
      "step": 5448
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0441381931304932,
      "learning_rate": 7.613148895204906e-06,
      "loss": 0.5538,
      "step": 5449
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.001044750213623,
      "learning_rate": 7.612274119286884e-06,
      "loss": 0.5626,
      "step": 5450
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1037360429763794,
      "learning_rate": 7.611399233375087e-06,
      "loss": 0.562,
      "step": 5451
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.037367820739746,
      "learning_rate": 7.610524237506354e-06,
      "loss": 0.5773,
      "step": 5452
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0317175388336182,
      "learning_rate": 7.6096491317175246e-06,
      "loss": 0.5643,
      "step": 5453
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9355440139770508,
      "learning_rate": 7.608773916045449e-06,
      "loss": 0.5216,
      "step": 5454
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9212518334388733,
      "learning_rate": 7.607898590526979e-06,
      "loss": 0.529,
      "step": 5455
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9198362231254578,
      "learning_rate": 7.607023155198973e-06,
      "loss": 0.494,
      "step": 5456
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0639890432357788,
      "learning_rate": 7.606147610098289e-06,
      "loss": 0.5544,
      "step": 5457
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9888911247253418,
      "learning_rate": 7.605271955261796e-06,
      "loss": 0.5496,
      "step": 5458
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0444262027740479,
      "learning_rate": 7.604396190726364e-06,
      "loss": 0.5366,
      "step": 5459
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2127799987792969,
      "learning_rate": 7.603520316528869e-06,
      "loss": 0.575,
      "step": 5460
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0242631435394287,
      "learning_rate": 7.60264433270619e-06,
      "loss": 0.545,
      "step": 5461
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.918563961982727,
      "learning_rate": 7.601768239295213e-06,
      "loss": 0.4738,
      "step": 5462
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0891157388687134,
      "learning_rate": 7.600892036332825e-06,
      "loss": 0.5843,
      "step": 5463
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0414520502090454,
      "learning_rate": 7.600015723855922e-06,
      "loss": 0.5781,
      "step": 5464
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1219522953033447,
      "learning_rate": 7.599139301901401e-06,
      "loss": 0.5522,
      "step": 5465
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0756316184997559,
      "learning_rate": 7.5982627705061666e-06,
      "loss": 0.5975,
      "step": 5466
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9472103118896484,
      "learning_rate": 7.597386129707126e-06,
      "loss": 0.5374,
      "step": 5467
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0254567861557007,
      "learning_rate": 7.596509379541191e-06,
      "loss": 0.5118,
      "step": 5468
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9630821943283081,
      "learning_rate": 7.595632520045277e-06,
      "loss": 0.5424,
      "step": 5469
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0659990310668945,
      "learning_rate": 7.594755551256308e-06,
      "loss": 0.5271,
      "step": 5470
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.037732720375061,
      "learning_rate": 7.593878473211209e-06,
      "loss": 0.5695,
      "step": 5471
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9575313329696655,
      "learning_rate": 7.593001285946913e-06,
      "loss": 0.4812,
      "step": 5472
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9590383172035217,
      "learning_rate": 7.592123989500351e-06,
      "loss": 0.5403,
      "step": 5473
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.141347885131836,
      "learning_rate": 7.591246583908465e-06,
      "loss": 0.5727,
      "step": 5474
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0539191961288452,
      "learning_rate": 7.590369069208201e-06,
      "loss": 0.5665,
      "step": 5475
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.944985032081604,
      "learning_rate": 7.589491445436505e-06,
      "loss": 0.551,
      "step": 5476
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9608230590820312,
      "learning_rate": 7.588613712630334e-06,
      "loss": 0.5212,
      "step": 5477
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.004446029663086,
      "learning_rate": 7.587735870826643e-06,
      "loss": 0.5286,
      "step": 5478
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9937500953674316,
      "learning_rate": 7.586857920062399e-06,
      "loss": 0.5697,
      "step": 5479
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0046453475952148,
      "learning_rate": 7.585979860374566e-06,
      "loss": 0.5417,
      "step": 5480
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.925399899482727,
      "learning_rate": 7.5851016918001165e-06,
      "loss": 0.5237,
      "step": 5481
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9018027782440186,
      "learning_rate": 7.584223414376028e-06,
      "loss": 0.5335,
      "step": 5482
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0428508520126343,
      "learning_rate": 7.583345028139282e-06,
      "loss": 0.5563,
      "step": 5483
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0010719299316406,
      "learning_rate": 7.582466533126863e-06,
      "loss": 0.5427,
      "step": 5484
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0468254089355469,
      "learning_rate": 7.581587929375761e-06,
      "loss": 0.5808,
      "step": 5485
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.953605055809021,
      "learning_rate": 7.580709216922973e-06,
      "loss": 0.54,
      "step": 5486
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0275905132293701,
      "learning_rate": 7.579830395805499e-06,
      "loss": 0.5562,
      "step": 5487
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1463297605514526,
      "learning_rate": 7.578951466060341e-06,
      "loss": 0.5444,
      "step": 5488
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9410524368286133,
      "learning_rate": 7.578072427724506e-06,
      "loss": 0.5091,
      "step": 5489
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0085787773132324,
      "learning_rate": 7.577193280835011e-06,
      "loss": 0.5477,
      "step": 5490
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9225336313247681,
      "learning_rate": 7.5763140254288716e-06,
      "loss": 0.5188,
      "step": 5491
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0280065536499023,
      "learning_rate": 7.575434661543113e-06,
      "loss": 0.5693,
      "step": 5492
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0376582145690918,
      "learning_rate": 7.574555189214756e-06,
      "loss": 0.5671,
      "step": 5493
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0159308910369873,
      "learning_rate": 7.573675608480841e-06,
      "loss": 0.5355,
      "step": 5494
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0317509174346924,
      "learning_rate": 7.5727959193783974e-06,
      "loss": 0.5673,
      "step": 5495
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.031361699104309,
      "learning_rate": 7.571916121944467e-06,
      "loss": 0.4955,
      "step": 5496
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9690631031990051,
      "learning_rate": 7.571036216216097e-06,
      "loss": 0.5287,
      "step": 5497
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0832093954086304,
      "learning_rate": 7.570156202230335e-06,
      "loss": 0.5871,
      "step": 5498
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0991196632385254,
      "learning_rate": 7.569276080024237e-06,
      "loss": 0.5868,
      "step": 5499
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.053709626197815,
      "learning_rate": 7.5683958496348596e-06,
      "loss": 0.5653,
      "step": 5500
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9709888100624084,
      "learning_rate": 7.567515511099268e-06,
      "loss": 0.5257,
      "step": 5501
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9813860654830933,
      "learning_rate": 7.56663506445453e-06,
      "loss": 0.5658,
      "step": 5502
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0247528553009033,
      "learning_rate": 7.5657545097377205e-06,
      "loss": 0.5182,
      "step": 5503
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8857021331787109,
      "learning_rate": 7.564873846985912e-06,
      "loss": 0.5192,
      "step": 5504
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0899372100830078,
      "learning_rate": 7.563993076236189e-06,
      "loss": 0.5456,
      "step": 5505
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.18221914768219,
      "learning_rate": 7.563112197525637e-06,
      "loss": 0.5913,
      "step": 5506
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0839096307754517,
      "learning_rate": 7.562231210891347e-06,
      "loss": 0.5285,
      "step": 5507
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0166727304458618,
      "learning_rate": 7.561350116370413e-06,
      "loss": 0.5327,
      "step": 5508
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0474815368652344,
      "learning_rate": 7.560468913999937e-06,
      "loss": 0.5666,
      "step": 5509
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0153292417526245,
      "learning_rate": 7.559587603817022e-06,
      "loss": 0.5532,
      "step": 5510
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0052329301834106,
      "learning_rate": 7.558706185858777e-06,
      "loss": 0.5549,
      "step": 5511
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9917075634002686,
      "learning_rate": 7.557824660162316e-06,
      "loss": 0.5391,
      "step": 5512
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1074151992797852,
      "learning_rate": 7.556943026764756e-06,
      "loss": 0.5031,
      "step": 5513
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0473936796188354,
      "learning_rate": 7.55606128570322e-06,
      "loss": 0.5169,
      "step": 5514
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0295766592025757,
      "learning_rate": 7.5551794370148366e-06,
      "loss": 0.5779,
      "step": 5515
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9668201804161072,
      "learning_rate": 7.554297480736734e-06,
      "loss": 0.5236,
      "step": 5516
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9963002800941467,
      "learning_rate": 7.553415416906051e-06,
      "loss": 0.5361,
      "step": 5517
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1146910190582275,
      "learning_rate": 7.552533245559927e-06,
      "loss": 0.5512,
      "step": 5518
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0025383234024048,
      "learning_rate": 7.551650966735509e-06,
      "loss": 0.5026,
      "step": 5519
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.053562879562378,
      "learning_rate": 7.550768580469945e-06,
      "loss": 0.5612,
      "step": 5520
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.030248761177063,
      "learning_rate": 7.549886086800389e-06,
      "loss": 0.5254,
      "step": 5521
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8703546524047852,
      "learning_rate": 7.549003485763999e-06,
      "loss": 0.4456,
      "step": 5522
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0638717412948608,
      "learning_rate": 7.548120777397941e-06,
      "loss": 0.5633,
      "step": 5523
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.140637755393982,
      "learning_rate": 7.547237961739382e-06,
      "loss": 0.5914,
      "step": 5524
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0732502937316895,
      "learning_rate": 7.546355038825492e-06,
      "loss": 0.5189,
      "step": 5525
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9874221086502075,
      "learning_rate": 7.545472008693451e-06,
      "loss": 0.5542,
      "step": 5526
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.110288143157959,
      "learning_rate": 7.544588871380439e-06,
      "loss": 0.5781,
      "step": 5527
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9584753513336182,
      "learning_rate": 7.54370562692364e-06,
      "loss": 0.5693,
      "step": 5528
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9943515658378601,
      "learning_rate": 7.542822275360246e-06,
      "loss": 0.5642,
      "step": 5529
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0656532049179077,
      "learning_rate": 7.541938816727453e-06,
      "loss": 0.5973,
      "step": 5530
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0154662132263184,
      "learning_rate": 7.5410552510624594e-06,
      "loss": 0.5365,
      "step": 5531
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0014674663543701,
      "learning_rate": 7.540171578402466e-06,
      "loss": 0.5718,
      "step": 5532
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0146523714065552,
      "learning_rate": 7.539287798784688e-06,
      "loss": 0.5304,
      "step": 5533
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.035952091217041,
      "learning_rate": 7.538403912246333e-06,
      "loss": 0.5214,
      "step": 5534
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9672898054122925,
      "learning_rate": 7.537519918824619e-06,
      "loss": 0.5037,
      "step": 5535
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9362606406211853,
      "learning_rate": 7.5366358185567676e-06,
      "loss": 0.4639,
      "step": 5536
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9829898476600647,
      "learning_rate": 7.5357516114800075e-06,
      "loss": 0.5478,
      "step": 5537
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9308422207832336,
      "learning_rate": 7.534867297631569e-06,
      "loss": 0.5041,
      "step": 5538
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0422790050506592,
      "learning_rate": 7.533982877048685e-06,
      "loss": 0.5872,
      "step": 5539
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0372891426086426,
      "learning_rate": 7.5330983497685975e-06,
      "loss": 0.5368,
      "step": 5540
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0850752592086792,
      "learning_rate": 7.532213715828551e-06,
      "loss": 0.5311,
      "step": 5541
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9258276224136353,
      "learning_rate": 7.531328975265795e-06,
      "loss": 0.5529,
      "step": 5542
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9780939817428589,
      "learning_rate": 7.53044412811758e-06,
      "loss": 0.503,
      "step": 5543
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0371925830841064,
      "learning_rate": 7.529559174421167e-06,
      "loss": 0.5487,
      "step": 5544
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9781390428543091,
      "learning_rate": 7.528674114213816e-06,
      "loss": 0.5579,
      "step": 5545
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.094603419303894,
      "learning_rate": 7.527788947532795e-06,
      "loss": 0.5784,
      "step": 5546
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.128503441810608,
      "learning_rate": 7.526903674415373e-06,
      "loss": 0.5358,
      "step": 5547
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0087530612945557,
      "learning_rate": 7.526018294898832e-06,
      "loss": 0.5644,
      "step": 5548
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.00899076461792,
      "learning_rate": 7.525132809020443e-06,
      "loss": 0.5114,
      "step": 5549
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9247457981109619,
      "learning_rate": 7.524247216817499e-06,
      "loss": 0.5141,
      "step": 5550
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0578961372375488,
      "learning_rate": 7.5233615183272836e-06,
      "loss": 0.5232,
      "step": 5551
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1226731538772583,
      "learning_rate": 7.522475713587095e-06,
      "loss": 0.5662,
      "step": 5552
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0937976837158203,
      "learning_rate": 7.521589802634228e-06,
      "loss": 0.5649,
      "step": 5553
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.984603226184845,
      "learning_rate": 7.520703785505987e-06,
      "loss": 0.5423,
      "step": 5554
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9459813237190247,
      "learning_rate": 7.519817662239678e-06,
      "loss": 0.5214,
      "step": 5555
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.017005443572998,
      "learning_rate": 7.518931432872614e-06,
      "loss": 0.5657,
      "step": 5556
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0255146026611328,
      "learning_rate": 7.518045097442111e-06,
      "loss": 0.5667,
      "step": 5557
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0680776834487915,
      "learning_rate": 7.517158655985483e-06,
      "loss": 0.5562,
      "step": 5558
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.114147424697876,
      "learning_rate": 7.516272108540066e-06,
      "loss": 0.5831,
      "step": 5559
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0147895812988281,
      "learning_rate": 7.515385455143183e-06,
      "loss": 0.5874,
      "step": 5560
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1398597955703735,
      "learning_rate": 7.514498695832169e-06,
      "loss": 0.5688,
      "step": 5561
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9944059252738953,
      "learning_rate": 7.51361183064436e-06,
      "loss": 0.5354,
      "step": 5562
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0544904470443726,
      "learning_rate": 7.512724859617103e-06,
      "loss": 0.5276,
      "step": 5563
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1566762924194336,
      "learning_rate": 7.511837782787743e-06,
      "loss": 0.5288,
      "step": 5564
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0607764720916748,
      "learning_rate": 7.510950600193632e-06,
      "loss": 0.5206,
      "step": 5565
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9686064124107361,
      "learning_rate": 7.510063311872125e-06,
      "loss": 0.5209,
      "step": 5566
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0985875129699707,
      "learning_rate": 7.509175917860586e-06,
      "loss": 0.5759,
      "step": 5567
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0419986248016357,
      "learning_rate": 7.508288418196377e-06,
      "loss": 0.557,
      "step": 5568
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0015439987182617,
      "learning_rate": 7.507400812916868e-06,
      "loss": 0.5079,
      "step": 5569
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.097322940826416,
      "learning_rate": 7.5065131020594316e-06,
      "loss": 0.5759,
      "step": 5570
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0685824155807495,
      "learning_rate": 7.5056252856614505e-06,
      "loss": 0.526,
      "step": 5571
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0418754816055298,
      "learning_rate": 7.504737363760306e-06,
      "loss": 0.6284,
      "step": 5572
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9443859457969666,
      "learning_rate": 7.503849336393382e-06,
      "loss": 0.5359,
      "step": 5573
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0443962812423706,
      "learning_rate": 7.502961203598074e-06,
      "loss": 0.5534,
      "step": 5574
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9699767231941223,
      "learning_rate": 7.502072965411776e-06,
      "loss": 0.5226,
      "step": 5575
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1486808061599731,
      "learning_rate": 7.501184621871891e-06,
      "loss": 0.5182,
      "step": 5576
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9507920145988464,
      "learning_rate": 7.5002961730158204e-06,
      "loss": 0.5199,
      "step": 5577
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.052960991859436,
      "learning_rate": 7.499407618880979e-06,
      "loss": 0.5421,
      "step": 5578
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9882809519767761,
      "learning_rate": 7.498518959504775e-06,
      "loss": 0.5601,
      "step": 5579
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.05231511592865,
      "learning_rate": 7.49763019492463e-06,
      "loss": 0.5815,
      "step": 5580
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0310633182525635,
      "learning_rate": 7.4967413251779655e-06,
      "loss": 0.5782,
      "step": 5581
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0598621368408203,
      "learning_rate": 7.495852350302209e-06,
      "loss": 0.5534,
      "step": 5582
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.952500581741333,
      "learning_rate": 7.494963270334794e-06,
      "loss": 0.5795,
      "step": 5583
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0291993618011475,
      "learning_rate": 7.494074085313155e-06,
      "loss": 0.4821,
      "step": 5584
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.014594316482544,
      "learning_rate": 7.493184795274731e-06,
      "loss": 0.5457,
      "step": 5585
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0594139099121094,
      "learning_rate": 7.49229540025697e-06,
      "loss": 0.6289,
      "step": 5586
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.008696436882019,
      "learning_rate": 7.4914059002973185e-06,
      "loss": 0.5707,
      "step": 5587
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9493990540504456,
      "learning_rate": 7.490516295433232e-06,
      "loss": 0.4909,
      "step": 5588
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0605522394180298,
      "learning_rate": 7.489626585702169e-06,
      "loss": 0.5402,
      "step": 5589
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.025477647781372,
      "learning_rate": 7.4887367711415905e-06,
      "loss": 0.5494,
      "step": 5590
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9939894676208496,
      "learning_rate": 7.487846851788965e-06,
      "loss": 0.5431,
      "step": 5591
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.996059238910675,
      "learning_rate": 7.486956827681761e-06,
      "loss": 0.5496,
      "step": 5592
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0278451442718506,
      "learning_rate": 7.4860666988574585e-06,
      "loss": 0.5322,
      "step": 5593
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0295774936676025,
      "learning_rate": 7.485176465353534e-06,
      "loss": 0.5237,
      "step": 5594
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8795758485794067,
      "learning_rate": 7.484286127207476e-06,
      "loss": 0.4798,
      "step": 5595
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.974780797958374,
      "learning_rate": 7.48339568445677e-06,
      "loss": 0.5342,
      "step": 5596
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9459009170532227,
      "learning_rate": 7.482505137138911e-06,
      "loss": 0.5407,
      "step": 5597
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0662119388580322,
      "learning_rate": 7.4816144852913975e-06,
      "loss": 0.5727,
      "step": 5598
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0604021549224854,
      "learning_rate": 7.480723728951731e-06,
      "loss": 0.5644,
      "step": 5599
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9835381507873535,
      "learning_rate": 7.479832868157416e-06,
      "loss": 0.5444,
      "step": 5600
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9299740195274353,
      "learning_rate": 7.4789419029459675e-06,
      "loss": 0.482,
      "step": 5601
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9987683892250061,
      "learning_rate": 7.478050833354897e-06,
      "loss": 0.5687,
      "step": 5602
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0130796432495117,
      "learning_rate": 7.47715965942173e-06,
      "loss": 0.5344,
      "step": 5603
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9855684041976929,
      "learning_rate": 7.476268381183984e-06,
      "loss": 0.4738,
      "step": 5604
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0145164728164673,
      "learning_rate": 7.475376998679193e-06,
      "loss": 0.5809,
      "step": 5605
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0176178216934204,
      "learning_rate": 7.474485511944887e-06,
      "loss": 0.554,
      "step": 5606
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9874436259269714,
      "learning_rate": 7.4735939210186036e-06,
      "loss": 0.5148,
      "step": 5607
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0104726552963257,
      "learning_rate": 7.472702225937884e-06,
      "loss": 0.5412,
      "step": 5608
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.001220464706421,
      "learning_rate": 7.471810426740278e-06,
      "loss": 0.538,
      "step": 5609
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9531809091567993,
      "learning_rate": 7.470918523463333e-06,
      "loss": 0.521,
      "step": 5610
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0025501251220703,
      "learning_rate": 7.470026516144604e-06,
      "loss": 0.575,
      "step": 5611
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9598075151443481,
      "learning_rate": 7.469134404821652e-06,
      "loss": 0.5063,
      "step": 5612
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0706355571746826,
      "learning_rate": 7.468242189532039e-06,
      "loss": 0.5339,
      "step": 5613
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.079030156135559,
      "learning_rate": 7.467349870313334e-06,
      "loss": 0.4836,
      "step": 5614
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0258371829986572,
      "learning_rate": 7.466457447203109e-06,
      "loss": 0.5309,
      "step": 5615
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9211562871932983,
      "learning_rate": 7.465564920238941e-06,
      "loss": 0.5266,
      "step": 5616
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.933488667011261,
      "learning_rate": 7.464672289458411e-06,
      "loss": 0.5295,
      "step": 5617
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.12042236328125,
      "learning_rate": 7.463779554899107e-06,
      "loss": 0.5473,
      "step": 5618
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0543317794799805,
      "learning_rate": 7.462886716598614e-06,
      "loss": 0.5447,
      "step": 5619
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9691358804702759,
      "learning_rate": 7.46199377459453e-06,
      "loss": 0.5441,
      "step": 5620
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1141588687896729,
      "learning_rate": 7.461100728924455e-06,
      "loss": 0.501,
      "step": 5621
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9861559867858887,
      "learning_rate": 7.460207579625988e-06,
      "loss": 0.5273,
      "step": 5622
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.97755366563797,
      "learning_rate": 7.459314326736738e-06,
      "loss": 0.5061,
      "step": 5623
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.093091368675232,
      "learning_rate": 7.458420970294317e-06,
      "loss": 0.5569,
      "step": 5624
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9785102605819702,
      "learning_rate": 7.457527510336342e-06,
      "loss": 0.4819,
      "step": 5625
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0222108364105225,
      "learning_rate": 7.456633946900432e-06,
      "loss": 0.5498,
      "step": 5626
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9800930023193359,
      "learning_rate": 7.455740280024212e-06,
      "loss": 0.5178,
      "step": 5627
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.966901421546936,
      "learning_rate": 7.454846509745311e-06,
      "loss": 0.5308,
      "step": 5628
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0559923648834229,
      "learning_rate": 7.453952636101366e-06,
      "loss": 0.5995,
      "step": 5629
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0091379880905151,
      "learning_rate": 7.45305865913001e-06,
      "loss": 0.5898,
      "step": 5630
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1073163747787476,
      "learning_rate": 7.452164578868889e-06,
      "loss": 0.5816,
      "step": 5631
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9831271171569824,
      "learning_rate": 7.451270395355647e-06,
      "loss": 0.5804,
      "step": 5632
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9418684244155884,
      "learning_rate": 7.450376108627937e-06,
      "loss": 0.48,
      "step": 5633
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.993733286857605,
      "learning_rate": 7.449481718723411e-06,
      "loss": 0.5548,
      "step": 5634
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9845336675643921,
      "learning_rate": 7.448587225679733e-06,
      "loss": 0.5687,
      "step": 5635
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9980434775352478,
      "learning_rate": 7.447692629534565e-06,
      "loss": 0.5142,
      "step": 5636
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0565739870071411,
      "learning_rate": 7.446797930325574e-06,
      "loss": 0.5431,
      "step": 5637
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9953488707542419,
      "learning_rate": 7.445903128090435e-06,
      "loss": 0.5196,
      "step": 5638
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.039448857307434,
      "learning_rate": 7.445008222866823e-06,
      "loss": 0.5051,
      "step": 5639
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.963239848613739,
      "learning_rate": 7.444113214692422e-06,
      "loss": 0.5137,
      "step": 5640
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0066325664520264,
      "learning_rate": 7.443218103604915e-06,
      "loss": 0.5287,
      "step": 5641
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0743483304977417,
      "learning_rate": 7.442322889641992e-06,
      "loss": 0.5878,
      "step": 5642
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9948568940162659,
      "learning_rate": 7.441427572841349e-06,
      "loss": 0.6129,
      "step": 5643
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9837034344673157,
      "learning_rate": 7.440532153240685e-06,
      "loss": 0.5177,
      "step": 5644
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9905748963356018,
      "learning_rate": 7.4396366308777015e-06,
      "loss": 0.5653,
      "step": 5645
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0937864780426025,
      "learning_rate": 7.4387410057901056e-06,
      "loss": 0.5016,
      "step": 5646
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9721953272819519,
      "learning_rate": 7.4378452780156094e-06,
      "loss": 0.4845,
      "step": 5647
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0243889093399048,
      "learning_rate": 7.436949447591931e-06,
      "loss": 0.5188,
      "step": 5648
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9671458601951599,
      "learning_rate": 7.4360535145567865e-06,
      "loss": 0.5298,
      "step": 5649
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9595158696174622,
      "learning_rate": 7.435157478947905e-06,
      "loss": 0.4837,
      "step": 5650
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9490304589271545,
      "learning_rate": 7.434261340803013e-06,
      "loss": 0.5266,
      "step": 5651
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0103788375854492,
      "learning_rate": 7.433365100159844e-06,
      "loss": 0.5178,
      "step": 5652
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.060921311378479,
      "learning_rate": 7.432468757056136e-06,
      "loss": 0.5291,
      "step": 5653
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9731550812721252,
      "learning_rate": 7.431572311529629e-06,
      "loss": 0.5375,
      "step": 5654
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0851389169692993,
      "learning_rate": 7.4306757636180725e-06,
      "loss": 0.4945,
      "step": 5655
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0156821012496948,
      "learning_rate": 7.429779113359214e-06,
      "loss": 0.5008,
      "step": 5656
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9673095941543579,
      "learning_rate": 7.428882360790811e-06,
      "loss": 0.5113,
      "step": 5657
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.134093165397644,
      "learning_rate": 7.427985505950619e-06,
      "loss": 0.5406,
      "step": 5658
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0507636070251465,
      "learning_rate": 7.427088548876406e-06,
      "loss": 0.5929,
      "step": 5659
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0085718631744385,
      "learning_rate": 7.426191489605936e-06,
      "loss": 0.5333,
      "step": 5660
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.981497049331665,
      "learning_rate": 7.425294328176984e-06,
      "loss": 0.5076,
      "step": 5661
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9927743673324585,
      "learning_rate": 7.4243970646273236e-06,
      "loss": 0.5324,
      "step": 5662
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0711027383804321,
      "learning_rate": 7.423499698994737e-06,
      "loss": 0.5218,
      "step": 5663
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.946141242980957,
      "learning_rate": 7.422602231317009e-06,
      "loss": 0.5257,
      "step": 5664
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0562503337860107,
      "learning_rate": 7.421704661631929e-06,
      "loss": 0.5366,
      "step": 5665
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9718359708786011,
      "learning_rate": 7.42080698997729e-06,
      "loss": 0.5756,
      "step": 5666
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9516698122024536,
      "learning_rate": 7.419909216390889e-06,
      "loss": 0.536,
      "step": 5667
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.954897940158844,
      "learning_rate": 7.4190113409105304e-06,
      "loss": 0.5016,
      "step": 5668
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0027780532836914,
      "learning_rate": 7.418113363574018e-06,
      "loss": 0.5256,
      "step": 5669
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.044217586517334,
      "learning_rate": 7.417215284419165e-06,
      "loss": 0.5591,
      "step": 5670
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0571179389953613,
      "learning_rate": 7.416317103483784e-06,
      "loss": 0.5826,
      "step": 5671
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1507805585861206,
      "learning_rate": 7.415418820805698e-06,
      "loss": 0.5302,
      "step": 5672
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.052793025970459,
      "learning_rate": 7.414520436422725e-06,
      "loss": 0.6083,
      "step": 5673
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9399493932723999,
      "learning_rate": 7.413621950372698e-06,
      "loss": 0.5089,
      "step": 5674
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9684324860572815,
      "learning_rate": 7.4127233626934456e-06,
      "loss": 0.5471,
      "step": 5675
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.084831714630127,
      "learning_rate": 7.411824673422808e-06,
      "loss": 0.5573,
      "step": 5676
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0572097301483154,
      "learning_rate": 7.410925882598621e-06,
      "loss": 0.5921,
      "step": 5677
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9332970380783081,
      "learning_rate": 7.410026990258734e-06,
      "loss": 0.5345,
      "step": 5678
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0227848291397095,
      "learning_rate": 7.409127996440993e-06,
      "loss": 0.5114,
      "step": 5679
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0537148714065552,
      "learning_rate": 7.408228901183254e-06,
      "loss": 0.5592,
      "step": 5680
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.004742980003357,
      "learning_rate": 7.407329704523372e-06,
      "loss": 0.536,
      "step": 5681
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0189913511276245,
      "learning_rate": 7.406430406499212e-06,
      "loss": 0.5426,
      "step": 5682
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0214310884475708,
      "learning_rate": 7.405531007148638e-06,
      "loss": 0.5429,
      "step": 5683
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9545047283172607,
      "learning_rate": 7.404631506509523e-06,
      "loss": 0.5079,
      "step": 5684
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.060755729675293,
      "learning_rate": 7.403731904619739e-06,
      "loss": 0.586,
      "step": 5685
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.942334771156311,
      "learning_rate": 7.402832201517166e-06,
      "loss": 0.5332,
      "step": 5686
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0206327438354492,
      "learning_rate": 7.40193239723969e-06,
      "loss": 0.5492,
      "step": 5687
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.026884913444519,
      "learning_rate": 7.401032491825194e-06,
      "loss": 0.5491,
      "step": 5688
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0408568382263184,
      "learning_rate": 7.400132485311573e-06,
      "loss": 0.548,
      "step": 5689
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1180294752120972,
      "learning_rate": 7.399232377736722e-06,
      "loss": 0.5797,
      "step": 5690
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.064048409461975,
      "learning_rate": 7.398332169138544e-06,
      "loss": 0.5514,
      "step": 5691
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0275312662124634,
      "learning_rate": 7.397431859554941e-06,
      "loss": 0.5385,
      "step": 5692
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9816993474960327,
      "learning_rate": 7.396531449023821e-06,
      "loss": 0.5183,
      "step": 5693
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0414061546325684,
      "learning_rate": 7.395630937583099e-06,
      "loss": 0.5726,
      "step": 5694
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9948675036430359,
      "learning_rate": 7.394730325270693e-06,
      "loss": 0.5384,
      "step": 5695
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0395992994308472,
      "learning_rate": 7.393829612124524e-06,
      "loss": 0.5407,
      "step": 5696
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0142568349838257,
      "learning_rate": 7.392928798182516e-06,
      "loss": 0.5272,
      "step": 5697
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9534479975700378,
      "learning_rate": 7.392027883482602e-06,
      "loss": 0.5845,
      "step": 5698
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9710732102394104,
      "learning_rate": 7.391126868062714e-06,
      "loss": 0.5356,
      "step": 5699
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9691248536109924,
      "learning_rate": 7.390225751960792e-06,
      "loss": 0.5359,
      "step": 5700
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9411775469779968,
      "learning_rate": 7.389324535214779e-06,
      "loss": 0.5567,
      "step": 5701
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0573960542678833,
      "learning_rate": 7.388423217862621e-06,
      "loss": 0.5633,
      "step": 5702
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0169092416763306,
      "learning_rate": 7.387521799942271e-06,
      "loss": 0.5405,
      "step": 5703
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9815866351127625,
      "learning_rate": 7.386620281491683e-06,
      "loss": 0.5868,
      "step": 5704
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.961209237575531,
      "learning_rate": 7.385718662548817e-06,
      "loss": 0.5813,
      "step": 5705
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.036128282546997,
      "learning_rate": 7.384816943151638e-06,
      "loss": 0.5422,
      "step": 5706
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9688419103622437,
      "learning_rate": 7.383915123338113e-06,
      "loss": 0.5279,
      "step": 5707
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8959729671478271,
      "learning_rate": 7.3830132031462165e-06,
      "loss": 0.5514,
      "step": 5708
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.007442593574524,
      "learning_rate": 7.382111182613923e-06,
      "loss": 0.4924,
      "step": 5709
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9503515958786011,
      "learning_rate": 7.381209061779214e-06,
      "loss": 0.5705,
      "step": 5710
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9991022944450378,
      "learning_rate": 7.380306840680076e-06,
      "loss": 0.5538,
      "step": 5711
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0210731029510498,
      "learning_rate": 7.379404519354496e-06,
      "loss": 0.5804,
      "step": 5712
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9408265948295593,
      "learning_rate": 7.378502097840471e-06,
      "loss": 0.5335,
      "step": 5713
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.086181879043579,
      "learning_rate": 7.377599576175995e-06,
      "loss": 0.594,
      "step": 5714
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.025308609008789,
      "learning_rate": 7.376696954399073e-06,
      "loss": 0.5439,
      "step": 5715
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9805166721343994,
      "learning_rate": 7.37579423254771e-06,
      "loss": 0.5418,
      "step": 5716
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.048662781715393,
      "learning_rate": 7.374891410659917e-06,
      "loss": 0.5599,
      "step": 5717
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.037967324256897,
      "learning_rate": 7.373988488773708e-06,
      "loss": 0.5504,
      "step": 5718
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1014705896377563,
      "learning_rate": 7.3730854669271015e-06,
      "loss": 0.5071,
      "step": 5719
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.019339680671692,
      "learning_rate": 7.372182345158122e-06,
      "loss": 0.5773,
      "step": 5720
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0042972564697266,
      "learning_rate": 7.3712791235047976e-06,
      "loss": 0.5491,
      "step": 5721
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1469221115112305,
      "learning_rate": 7.370375802005157e-06,
      "loss": 0.5723,
      "step": 5722
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1442276239395142,
      "learning_rate": 7.369472380697236e-06,
      "loss": 0.4971,
      "step": 5723
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9209834933280945,
      "learning_rate": 7.368568859619078e-06,
      "loss": 0.5027,
      "step": 5724
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0353718996047974,
      "learning_rate": 7.3676652388087234e-06,
      "loss": 0.5424,
      "step": 5725
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0237194299697876,
      "learning_rate": 7.366761518304223e-06,
      "loss": 0.535,
      "step": 5726
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0912165641784668,
      "learning_rate": 7.365857698143628e-06,
      "loss": 0.5355,
      "step": 5727
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.13982355594635,
      "learning_rate": 7.364953778364996e-06,
      "loss": 0.5697,
      "step": 5728
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.951877772808075,
      "learning_rate": 7.364049759006387e-06,
      "loss": 0.5509,
      "step": 5729
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0230592489242554,
      "learning_rate": 7.363145640105867e-06,
      "loss": 0.5146,
      "step": 5730
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9922885894775391,
      "learning_rate": 7.362241421701505e-06,
      "loss": 0.533,
      "step": 5731
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.080177664756775,
      "learning_rate": 7.3613371038313744e-06,
      "loss": 0.5416,
      "step": 5732
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9975622892379761,
      "learning_rate": 7.360432686533552e-06,
      "loss": 0.5887,
      "step": 5733
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0143722295761108,
      "learning_rate": 7.359528169846121e-06,
      "loss": 0.557,
      "step": 5734
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9244665503501892,
      "learning_rate": 7.358623553807167e-06,
      "loss": 0.477,
      "step": 5735
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9375993013381958,
      "learning_rate": 7.35771883845478e-06,
      "loss": 0.5025,
      "step": 5736
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9574780464172363,
      "learning_rate": 7.356814023827055e-06,
      "loss": 0.4916,
      "step": 5737
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0635297298431396,
      "learning_rate": 7.35590910996209e-06,
      "loss": 0.5399,
      "step": 5738
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9631228446960449,
      "learning_rate": 7.355004096897987e-06,
      "loss": 0.5071,
      "step": 5739
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9968248009681702,
      "learning_rate": 7.354098984672856e-06,
      "loss": 0.4824,
      "step": 5740
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0746512413024902,
      "learning_rate": 7.353193773324805e-06,
      "loss": 0.5767,
      "step": 5741
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0765202045440674,
      "learning_rate": 7.35228846289195e-06,
      "loss": 0.508,
      "step": 5742
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9962648749351501,
      "learning_rate": 7.351383053412411e-06,
      "loss": 0.5793,
      "step": 5743
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9487490057945251,
      "learning_rate": 7.350477544924313e-06,
      "loss": 0.5729,
      "step": 5744
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.005916953086853,
      "learning_rate": 7.349571937465782e-06,
      "loss": 0.5754,
      "step": 5745
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9979771375656128,
      "learning_rate": 7.348666231074948e-06,
      "loss": 0.5288,
      "step": 5746
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0600558519363403,
      "learning_rate": 7.3477604257899515e-06,
      "loss": 0.545,
      "step": 5747
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0493375062942505,
      "learning_rate": 7.346854521648929e-06,
      "loss": 0.577,
      "step": 5748
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9606602787971497,
      "learning_rate": 7.345948518690029e-06,
      "loss": 0.4939,
      "step": 5749
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0327515602111816,
      "learning_rate": 7.345042416951395e-06,
      "loss": 0.5413,
      "step": 5750
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9119663834571838,
      "learning_rate": 7.344136216471185e-06,
      "loss": 0.5097,
      "step": 5751
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.002037525177002,
      "learning_rate": 7.343229917287552e-06,
      "loss": 0.5294,
      "step": 5752
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1066184043884277,
      "learning_rate": 7.34232351943866e-06,
      "loss": 0.5014,
      "step": 5753
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0165209770202637,
      "learning_rate": 7.341417022962671e-06,
      "loss": 0.4836,
      "step": 5754
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0941275358200073,
      "learning_rate": 7.340510427897759e-06,
      "loss": 0.4963,
      "step": 5755
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1376608610153198,
      "learning_rate": 7.339603734282093e-06,
      "loss": 0.5642,
      "step": 5756
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0889432430267334,
      "learning_rate": 7.338696942153855e-06,
      "loss": 0.5388,
      "step": 5757
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9839189648628235,
      "learning_rate": 7.337790051551221e-06,
      "loss": 0.5082,
      "step": 5758
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0140478610992432,
      "learning_rate": 7.3368830625123835e-06,
      "loss": 0.5426,
      "step": 5759
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1105388402938843,
      "learning_rate": 7.335975975075529e-06,
      "loss": 0.6198,
      "step": 5760
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.046632170677185,
      "learning_rate": 7.3350687892788505e-06,
      "loss": 0.4986,
      "step": 5761
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.060750961303711,
      "learning_rate": 7.33416150516055e-06,
      "loss": 0.518,
      "step": 5762
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9920255541801453,
      "learning_rate": 7.333254122758828e-06,
      "loss": 0.5693,
      "step": 5763
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9959121942520142,
      "learning_rate": 7.332346642111893e-06,
      "loss": 0.5546,
      "step": 5764
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.037704586982727,
      "learning_rate": 7.331439063257953e-06,
      "loss": 0.5586,
      "step": 5765
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9203774333000183,
      "learning_rate": 7.330531386235225e-06,
      "loss": 0.5295,
      "step": 5766
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0745431184768677,
      "learning_rate": 7.329623611081927e-06,
      "loss": 0.5458,
      "step": 5767
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0159822702407837,
      "learning_rate": 7.3287157378362846e-06,
      "loss": 0.5106,
      "step": 5768
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0377622842788696,
      "learning_rate": 7.327807766536521e-06,
      "loss": 0.5749,
      "step": 5769
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.03034245967865,
      "learning_rate": 7.3268996972208725e-06,
      "loss": 0.5834,
      "step": 5770
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0114792585372925,
      "learning_rate": 7.325991529927572e-06,
      "loss": 0.5514,
      "step": 5771
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0172260999679565,
      "learning_rate": 7.325083264694859e-06,
      "loss": 0.4866,
      "step": 5772
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0338290929794312,
      "learning_rate": 7.324174901560978e-06,
      "loss": 0.5613,
      "step": 5773
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9678194522857666,
      "learning_rate": 7.323266440564177e-06,
      "loss": 0.5218,
      "step": 5774
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0081979036331177,
      "learning_rate": 7.32235788174271e-06,
      "loss": 0.4992,
      "step": 5775
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9918741583824158,
      "learning_rate": 7.32144922513483e-06,
      "loss": 0.5158,
      "step": 5776
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0231093168258667,
      "learning_rate": 7.320540470778799e-06,
      "loss": 0.5438,
      "step": 5777
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0513994693756104,
      "learning_rate": 7.319631618712881e-06,
      "loss": 0.5442,
      "step": 5778
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9481300115585327,
      "learning_rate": 7.318722668975347e-06,
      "loss": 0.4835,
      "step": 5779
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9858001470565796,
      "learning_rate": 7.317813621604466e-06,
      "loss": 0.536,
      "step": 5780
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9564962983131409,
      "learning_rate": 7.316904476638515e-06,
      "loss": 0.5025,
      "step": 5781
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9979397654533386,
      "learning_rate": 7.315995234115778e-06,
      "loss": 0.5011,
      "step": 5782
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.912788987159729,
      "learning_rate": 7.315085894074539e-06,
      "loss": 0.5239,
      "step": 5783
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0001533031463623,
      "learning_rate": 7.314176456553086e-06,
      "loss": 0.5199,
      "step": 5784
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9417988061904907,
      "learning_rate": 7.3132669215897125e-06,
      "loss": 0.5474,
      "step": 5785
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0640056133270264,
      "learning_rate": 7.312357289222717e-06,
      "loss": 0.5855,
      "step": 5786
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9787964224815369,
      "learning_rate": 7.3114475594904e-06,
      "loss": 0.5291,
      "step": 5787
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9471306204795837,
      "learning_rate": 7.310537732431067e-06,
      "loss": 0.5816,
      "step": 5788
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9862099289894104,
      "learning_rate": 7.309627808083027e-06,
      "loss": 0.5669,
      "step": 5789
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0701364278793335,
      "learning_rate": 7.308717786484596e-06,
      "loss": 0.5144,
      "step": 5790
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0688151121139526,
      "learning_rate": 7.30780766767409e-06,
      "loss": 0.6784,
      "step": 5791
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8825664520263672,
      "learning_rate": 7.306897451689832e-06,
      "loss": 0.4798,
      "step": 5792
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9757002592086792,
      "learning_rate": 7.305987138570145e-06,
      "loss": 0.5104,
      "step": 5793
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0232665538787842,
      "learning_rate": 7.305076728353364e-06,
      "loss": 0.5211,
      "step": 5794
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0065867900848389,
      "learning_rate": 7.30416622107782e-06,
      "loss": 0.5139,
      "step": 5795
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0104619264602661,
      "learning_rate": 7.303255616781853e-06,
      "loss": 0.5353,
      "step": 5796
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0506689548492432,
      "learning_rate": 7.3023449155038016e-06,
      "loss": 0.5751,
      "step": 5797
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9864729046821594,
      "learning_rate": 7.301434117282018e-06,
      "loss": 0.5732,
      "step": 5798
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9898861050605774,
      "learning_rate": 7.300523222154848e-06,
      "loss": 0.5242,
      "step": 5799
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.04075026512146,
      "learning_rate": 7.299612230160648e-06,
      "loss": 0.5514,
      "step": 5800
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0983926057815552,
      "learning_rate": 7.298701141337778e-06,
      "loss": 0.5203,
      "step": 5801
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9918738007545471,
      "learning_rate": 7.2977899557246e-06,
      "loss": 0.5609,
      "step": 5802
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9489316940307617,
      "learning_rate": 7.2968786733594795e-06,
      "loss": 0.5345,
      "step": 5803
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.984935998916626,
      "learning_rate": 7.295967294280788e-06,
      "loss": 0.5493,
      "step": 5804
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.01011061668396,
      "learning_rate": 7.2950558185269005e-06,
      "loss": 0.5288,
      "step": 5805
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1023284196853638,
      "learning_rate": 7.294144246136198e-06,
      "loss": 0.5096,
      "step": 5806
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0046350955963135,
      "learning_rate": 7.29323257714706e-06,
      "loss": 0.58,
      "step": 5807
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0193142890930176,
      "learning_rate": 7.292320811597877e-06,
      "loss": 0.5758,
      "step": 5808
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.923527717590332,
      "learning_rate": 7.291408949527039e-06,
      "loss": 0.4869,
      "step": 5809
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0279090404510498,
      "learning_rate": 7.290496990972942e-06,
      "loss": 0.523,
      "step": 5810
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9912611842155457,
      "learning_rate": 7.2895849359739834e-06,
      "loss": 0.502,
      "step": 5811
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0363621711730957,
      "learning_rate": 7.288672784568568e-06,
      "loss": 0.5422,
      "step": 5812
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9785564541816711,
      "learning_rate": 7.2877605367951055e-06,
      "loss": 0.5127,
      "step": 5813
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9607601761817932,
      "learning_rate": 7.286848192692003e-06,
      "loss": 0.5197,
      "step": 5814
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9498841166496277,
      "learning_rate": 7.28593575229768e-06,
      "loss": 0.547,
      "step": 5815
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9618740081787109,
      "learning_rate": 7.285023215650553e-06,
      "loss": 0.5023,
      "step": 5816
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9866049289703369,
      "learning_rate": 7.2841105827890475e-06,
      "loss": 0.5586,
      "step": 5817
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0533307790756226,
      "learning_rate": 7.283197853751593e-06,
      "loss": 0.5882,
      "step": 5818
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0719332695007324,
      "learning_rate": 7.282285028576618e-06,
      "loss": 0.5941,
      "step": 5819
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0505437850952148,
      "learning_rate": 7.28137210730256e-06,
      "loss": 0.5756,
      "step": 5820
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9779177308082581,
      "learning_rate": 7.280459089967861e-06,
      "loss": 0.4963,
      "step": 5821
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9846765398979187,
      "learning_rate": 7.279545976610961e-06,
      "loss": 0.5175,
      "step": 5822
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.042790174484253,
      "learning_rate": 7.278632767270309e-06,
      "loss": 0.5478,
      "step": 5823
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9730501770973206,
      "learning_rate": 7.277719461984361e-06,
      "loss": 0.5141,
      "step": 5824
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9226754903793335,
      "learning_rate": 7.276806060791567e-06,
      "loss": 0.5409,
      "step": 5825
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.015047311782837,
      "learning_rate": 7.275892563730393e-06,
      "loss": 0.5437,
      "step": 5826
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9851754307746887,
      "learning_rate": 7.274978970839297e-06,
      "loss": 0.5081,
      "step": 5827
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9413428902626038,
      "learning_rate": 7.274065282156752e-06,
      "loss": 0.4789,
      "step": 5828
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.178230881690979,
      "learning_rate": 7.273151497721229e-06,
      "loss": 0.5777,
      "step": 5829
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9771174192428589,
      "learning_rate": 7.272237617571205e-06,
      "loss": 0.5226,
      "step": 5830
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.155280351638794,
      "learning_rate": 7.2713236417451584e-06,
      "loss": 0.5712,
      "step": 5831
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9939926266670227,
      "learning_rate": 7.2704095702815754e-06,
      "loss": 0.5274,
      "step": 5832
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9951380491256714,
      "learning_rate": 7.269495403218943e-06,
      "loss": 0.5247,
      "step": 5833
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.97422856092453,
      "learning_rate": 7.268581140595754e-06,
      "loss": 0.4867,
      "step": 5834
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0003308057785034,
      "learning_rate": 7.267666782450505e-06,
      "loss": 0.4925,
      "step": 5835
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0902355909347534,
      "learning_rate": 7.266752328821698e-06,
      "loss": 0.6134,
      "step": 5836
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9617888927459717,
      "learning_rate": 7.265837779747834e-06,
      "loss": 0.4861,
      "step": 5837
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0846583843231201,
      "learning_rate": 7.264923135267425e-06,
      "loss": 0.5432,
      "step": 5838
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1127210855484009,
      "learning_rate": 7.264008395418981e-06,
      "loss": 0.5714,
      "step": 5839
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9533615112304688,
      "learning_rate": 7.263093560241019e-06,
      "loss": 0.5281,
      "step": 5840
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9426731467247009,
      "learning_rate": 7.262178629772061e-06,
      "loss": 0.502,
      "step": 5841
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0847283601760864,
      "learning_rate": 7.261263604050628e-06,
      "loss": 0.5175,
      "step": 5842
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9640687108039856,
      "learning_rate": 7.260348483115254e-06,
      "loss": 0.4872,
      "step": 5843
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9335427284240723,
      "learning_rate": 7.259433267004466e-06,
      "loss": 0.5078,
      "step": 5844
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9635369181632996,
      "learning_rate": 7.258517955756805e-06,
      "loss": 0.5363,
      "step": 5845
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0456881523132324,
      "learning_rate": 7.257602549410808e-06,
      "loss": 0.5283,
      "step": 5846
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0658409595489502,
      "learning_rate": 7.256687048005024e-06,
      "loss": 0.5749,
      "step": 5847
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2358348369598389,
      "learning_rate": 7.255771451577996e-06,
      "loss": 0.5693,
      "step": 5848
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0727468729019165,
      "learning_rate": 7.254855760168281e-06,
      "loss": 0.5051,
      "step": 5849
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9722975492477417,
      "learning_rate": 7.2539399738144325e-06,
      "loss": 0.4984,
      "step": 5850
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1179519891738892,
      "learning_rate": 7.2530240925550145e-06,
      "loss": 0.5229,
      "step": 5851
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0390774011611938,
      "learning_rate": 7.252108116428589e-06,
      "loss": 0.52,
      "step": 5852
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1116324663162231,
      "learning_rate": 7.251192045473725e-06,
      "loss": 0.5284,
      "step": 5853
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.048357605934143,
      "learning_rate": 7.250275879728995e-06,
      "loss": 0.5232,
      "step": 5854
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9675121307373047,
      "learning_rate": 7.249359619232976e-06,
      "loss": 0.5146,
      "step": 5855
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0100194215774536,
      "learning_rate": 7.24844326402425e-06,
      "loss": 0.5491,
      "step": 5856
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.979152500629425,
      "learning_rate": 7.247526814141398e-06,
      "loss": 0.5108,
      "step": 5857
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9853137731552124,
      "learning_rate": 7.2466102696230115e-06,
      "loss": 0.5398,
      "step": 5858
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9618313312530518,
      "learning_rate": 7.24569363050768e-06,
      "loss": 0.5451,
      "step": 5859
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9729511141777039,
      "learning_rate": 7.244776896834004e-06,
      "loss": 0.5304,
      "step": 5860
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9838532209396362,
      "learning_rate": 7.243860068640581e-06,
      "loss": 0.5555,
      "step": 5861
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0647404193878174,
      "learning_rate": 7.242943145966016e-06,
      "loss": 0.5021,
      "step": 5862
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9827612638473511,
      "learning_rate": 7.242026128848918e-06,
      "loss": 0.5017,
      "step": 5863
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9512253999710083,
      "learning_rate": 7.241109017327901e-06,
      "loss": 0.5175,
      "step": 5864
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9890321493148804,
      "learning_rate": 7.240191811441577e-06,
      "loss": 0.5438,
      "step": 5865
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0261216163635254,
      "learning_rate": 7.239274511228569e-06,
      "loss": 0.5495,
      "step": 5866
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9640551209449768,
      "learning_rate": 7.238357116727502e-06,
      "loss": 0.5663,
      "step": 5867
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1027157306671143,
      "learning_rate": 7.2374396279770044e-06,
      "loss": 0.5347,
      "step": 5868
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9906113743782043,
      "learning_rate": 7.236522045015706e-06,
      "loss": 0.5146,
      "step": 5869
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0150017738342285,
      "learning_rate": 7.235604367882245e-06,
      "loss": 0.5514,
      "step": 5870
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0121136903762817,
      "learning_rate": 7.234686596615262e-06,
      "loss": 0.5075,
      "step": 5871
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.036228895187378,
      "learning_rate": 7.2337687312534e-06,
      "loss": 0.5592,
      "step": 5872
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0446488857269287,
      "learning_rate": 7.232850771835307e-06,
      "loss": 0.5495,
      "step": 5873
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.035109043121338,
      "learning_rate": 7.231932718399635e-06,
      "loss": 0.5271,
      "step": 5874
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0854578018188477,
      "learning_rate": 7.231014570985042e-06,
      "loss": 0.5426,
      "step": 5875
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.036075234413147,
      "learning_rate": 7.230096329630185e-06,
      "loss": 0.5134,
      "step": 5876
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2155336141586304,
      "learning_rate": 7.22917799437373e-06,
      "loss": 0.508,
      "step": 5877
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0050281286239624,
      "learning_rate": 7.228259565254345e-06,
      "loss": 0.5759,
      "step": 5878
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.069089651107788,
      "learning_rate": 7.227341042310702e-06,
      "loss": 0.5352,
      "step": 5879
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.133564829826355,
      "learning_rate": 7.226422425581474e-06,
      "loss": 0.5747,
      "step": 5880
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0305187702178955,
      "learning_rate": 7.225503715105344e-06,
      "loss": 0.5182,
      "step": 5881
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9435314536094666,
      "learning_rate": 7.224584910920994e-06,
      "loss": 0.5672,
      "step": 5882
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0364540815353394,
      "learning_rate": 7.223666013067113e-06,
      "loss": 0.5042,
      "step": 5883
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.054391860961914,
      "learning_rate": 7.222747021582392e-06,
      "loss": 0.5352,
      "step": 5884
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0646604299545288,
      "learning_rate": 7.221827936505524e-06,
      "loss": 0.5673,
      "step": 5885
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9911362528800964,
      "learning_rate": 7.220908757875214e-06,
      "loss": 0.5121,
      "step": 5886
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9722857475280762,
      "learning_rate": 7.21998948573016e-06,
      "loss": 0.5494,
      "step": 5887
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.931648313999176,
      "learning_rate": 7.219070120109072e-06,
      "loss": 0.4944,
      "step": 5888
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0476207733154297,
      "learning_rate": 7.2181506610506605e-06,
      "loss": 0.5319,
      "step": 5889
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0259501934051514,
      "learning_rate": 7.217231108593642e-06,
      "loss": 0.5619,
      "step": 5890
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.985495388507843,
      "learning_rate": 7.2163114627767336e-06,
      "loss": 0.5224,
      "step": 5891
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0021553039550781,
      "learning_rate": 7.21539172363866e-06,
      "loss": 0.5759,
      "step": 5892
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0142312049865723,
      "learning_rate": 7.214471891218147e-06,
      "loss": 0.5011,
      "step": 5893
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9844617247581482,
      "learning_rate": 7.213551965553927e-06,
      "loss": 0.5086,
      "step": 5894
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.048954725265503,
      "learning_rate": 7.212631946684735e-06,
      "loss": 0.5307,
      "step": 5895
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9628361463546753,
      "learning_rate": 7.211711834649308e-06,
      "loss": 0.5715,
      "step": 5896
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0532171726226807,
      "learning_rate": 7.210791629486389e-06,
      "loss": 0.5546,
      "step": 5897
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0313563346862793,
      "learning_rate": 7.209871331234727e-06,
      "loss": 0.5123,
      "step": 5898
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0224592685699463,
      "learning_rate": 7.208950939933069e-06,
      "loss": 0.5466,
      "step": 5899
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9858354330062866,
      "learning_rate": 7.208030455620172e-06,
      "loss": 0.5493,
      "step": 5900
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0029754638671875,
      "learning_rate": 7.207109878334794e-06,
      "loss": 0.4909,
      "step": 5901
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9922599196434021,
      "learning_rate": 7.206189208115697e-06,
      "loss": 0.5534,
      "step": 5902
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9229944944381714,
      "learning_rate": 7.205268445001647e-06,
      "loss": 0.5001,
      "step": 5903
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.019700050354004,
      "learning_rate": 7.204347589031413e-06,
      "loss": 0.5436,
      "step": 5904
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9846832752227783,
      "learning_rate": 7.203426640243772e-06,
      "loss": 0.5058,
      "step": 5905
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9663556814193726,
      "learning_rate": 7.2025055986775e-06,
      "loss": 0.5284,
      "step": 5906
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.962640643119812,
      "learning_rate": 7.201584464371378e-06,
      "loss": 0.5636,
      "step": 5907
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0475239753723145,
      "learning_rate": 7.200663237364195e-06,
      "loss": 0.5509,
      "step": 5908
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0490944385528564,
      "learning_rate": 7.199741917694738e-06,
      "loss": 0.5452,
      "step": 5909
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0195828676223755,
      "learning_rate": 7.198820505401801e-06,
      "loss": 0.5843,
      "step": 5910
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0507577657699585,
      "learning_rate": 7.197899000524181e-06,
      "loss": 0.5732,
      "step": 5911
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0144188404083252,
      "learning_rate": 7.196977403100681e-06,
      "loss": 0.5377,
      "step": 5912
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0264872312545776,
      "learning_rate": 7.196055713170105e-06,
      "loss": 0.5162,
      "step": 5913
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9802569150924683,
      "learning_rate": 7.195133930771263e-06,
      "loss": 0.5609,
      "step": 5914
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0005505084991455,
      "learning_rate": 7.194212055942966e-06,
      "loss": 0.5513,
      "step": 5915
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9948568940162659,
      "learning_rate": 7.193290088724034e-06,
      "loss": 0.5248,
      "step": 5916
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.050577998161316,
      "learning_rate": 7.192368029153285e-06,
      "loss": 0.5972,
      "step": 5917
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0255308151245117,
      "learning_rate": 7.191445877269548e-06,
      "loss": 0.4951,
      "step": 5918
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9985212683677673,
      "learning_rate": 7.190523633111644e-06,
      "loss": 0.5641,
      "step": 5919
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0271326303482056,
      "learning_rate": 7.189601296718413e-06,
      "loss": 0.5382,
      "step": 5920
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0106226205825806,
      "learning_rate": 7.188678868128687e-06,
      "loss": 0.5551,
      "step": 5921
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0215424299240112,
      "learning_rate": 7.18775634738131e-06,
      "loss": 0.5541,
      "step": 5922
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0225163698196411,
      "learning_rate": 7.18683373451512e-06,
      "loss": 0.5559,
      "step": 5923
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.012471318244934,
      "learning_rate": 7.185911029568972e-06,
      "loss": 0.5423,
      "step": 5924
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.021701455116272,
      "learning_rate": 7.184988232581713e-06,
      "loss": 0.5427,
      "step": 5925
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9949175715446472,
      "learning_rate": 7.184065343592203e-06,
      "loss": 0.5773,
      "step": 5926
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0874463319778442,
      "learning_rate": 7.183142362639296e-06,
      "loss": 0.5638,
      "step": 5927
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.064543604850769,
      "learning_rate": 7.18221928976186e-06,
      "loss": 0.5246,
      "step": 5928
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9457411766052246,
      "learning_rate": 7.181296124998762e-06,
      "loss": 0.543,
      "step": 5929
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0419443845748901,
      "learning_rate": 7.180372868388873e-06,
      "loss": 0.5445,
      "step": 5930
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0131248235702515,
      "learning_rate": 7.179449519971066e-06,
      "loss": 0.5167,
      "step": 5931
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0083361864089966,
      "learning_rate": 7.178526079784221e-06,
      "loss": 0.5367,
      "step": 5932
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.010665774345398,
      "learning_rate": 7.1776025478672225e-06,
      "loss": 0.5431,
      "step": 5933
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1985095739364624,
      "learning_rate": 7.176678924258955e-06,
      "loss": 0.5673,
      "step": 5934
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9250592589378357,
      "learning_rate": 7.175755208998311e-06,
      "loss": 0.4978,
      "step": 5935
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9751607775688171,
      "learning_rate": 7.174831402124184e-06,
      "loss": 0.5452,
      "step": 5936
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0652177333831787,
      "learning_rate": 7.173907503675472e-06,
      "loss": 0.5584,
      "step": 5937
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.144456148147583,
      "learning_rate": 7.172983513691076e-06,
      "loss": 0.5653,
      "step": 5938
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.030582070350647,
      "learning_rate": 7.172059432209907e-06,
      "loss": 0.5257,
      "step": 5939
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0467146635055542,
      "learning_rate": 7.171135259270868e-06,
      "loss": 0.5582,
      "step": 5940
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.044650673866272,
      "learning_rate": 7.170210994912878e-06,
      "loss": 0.5191,
      "step": 5941
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.049615740776062,
      "learning_rate": 7.169286639174852e-06,
      "loss": 0.5332,
      "step": 5942
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1635546684265137,
      "learning_rate": 7.168362192095712e-06,
      "loss": 0.562,
      "step": 5943
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.193900227546692,
      "learning_rate": 7.1674376537143845e-06,
      "loss": 0.5554,
      "step": 5944
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0751802921295166,
      "learning_rate": 7.166513024069797e-06,
      "loss": 0.5491,
      "step": 5945
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9560760259628296,
      "learning_rate": 7.16558830320088e-06,
      "loss": 0.5585,
      "step": 5946
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0157760381698608,
      "learning_rate": 7.1646634911465765e-06,
      "loss": 0.4894,
      "step": 5947
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0313069820404053,
      "learning_rate": 7.163738587945822e-06,
      "loss": 0.5303,
      "step": 5948
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0127816200256348,
      "learning_rate": 7.162813593637563e-06,
      "loss": 0.5285,
      "step": 5949
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0272399187088013,
      "learning_rate": 7.161888508260748e-06,
      "loss": 0.5537,
      "step": 5950
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.013628602027893,
      "learning_rate": 7.160963331854327e-06,
      "loss": 0.4851,
      "step": 5951
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.007154107093811,
      "learning_rate": 7.16003806445726e-06,
      "loss": 0.5287,
      "step": 5952
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0067106485366821,
      "learning_rate": 7.159112706108502e-06,
      "loss": 0.4884,
      "step": 5953
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.994032621383667,
      "learning_rate": 7.15818725684702e-06,
      "loss": 0.5397,
      "step": 5954
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0723838806152344,
      "learning_rate": 7.15726171671178e-06,
      "loss": 0.5378,
      "step": 5955
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9964460730552673,
      "learning_rate": 7.156336085741755e-06,
      "loss": 0.5543,
      "step": 5956
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.007655382156372,
      "learning_rate": 7.155410363975916e-06,
      "loss": 0.5154,
      "step": 5957
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0438203811645508,
      "learning_rate": 7.154484551453247e-06,
      "loss": 0.5611,
      "step": 5958
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.007995367050171,
      "learning_rate": 7.1535586482127284e-06,
      "loss": 0.5762,
      "step": 5959
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0440713167190552,
      "learning_rate": 7.152632654293347e-06,
      "loss": 0.5378,
      "step": 5960
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0187562704086304,
      "learning_rate": 7.151706569734091e-06,
      "loss": 0.5286,
      "step": 5961
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8911324143409729,
      "learning_rate": 7.150780394573957e-06,
      "loss": 0.5182,
      "step": 5962
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0653436183929443,
      "learning_rate": 7.149854128851945e-06,
      "loss": 0.5198,
      "step": 5963
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.015015959739685,
      "learning_rate": 7.148927772607053e-06,
      "loss": 0.5289,
      "step": 5964
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0303547382354736,
      "learning_rate": 7.148001325878287e-06,
      "loss": 0.4941,
      "step": 5965
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0739134550094604,
      "learning_rate": 7.147074788704659e-06,
      "loss": 0.552,
      "step": 5966
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0836094617843628,
      "learning_rate": 7.14614816112518e-06,
      "loss": 0.5439,
      "step": 5967
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9984629154205322,
      "learning_rate": 7.145221443178868e-06,
      "loss": 0.5378,
      "step": 5968
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0398366451263428,
      "learning_rate": 7.144294634904744e-06,
      "loss": 0.5474,
      "step": 5969
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1328619718551636,
      "learning_rate": 7.143367736341832e-06,
      "loss": 0.6143,
      "step": 5970
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1427115201950073,
      "learning_rate": 7.142440747529161e-06,
      "loss": 0.5713,
      "step": 5971
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.063494324684143,
      "learning_rate": 7.141513668505764e-06,
      "loss": 0.5325,
      "step": 5972
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0437992811203003,
      "learning_rate": 7.140586499310674e-06,
      "loss": 0.5313,
      "step": 5973
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9050129055976868,
      "learning_rate": 7.139659239982935e-06,
      "loss": 0.4815,
      "step": 5974
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9722835421562195,
      "learning_rate": 7.138731890561589e-06,
      "loss": 0.5278,
      "step": 5975
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.006160855293274,
      "learning_rate": 7.1378044510856814e-06,
      "loss": 0.5462,
      "step": 5976
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.033493995666504,
      "learning_rate": 7.136876921594267e-06,
      "loss": 0.5524,
      "step": 5977
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.030398964881897,
      "learning_rate": 7.1359493021263986e-06,
      "loss": 0.5793,
      "step": 5978
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.004825472831726,
      "learning_rate": 7.135021592721134e-06,
      "loss": 0.5153,
      "step": 5979
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9759687185287476,
      "learning_rate": 7.134093793417539e-06,
      "loss": 0.5439,
      "step": 5980
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0009828805923462,
      "learning_rate": 7.133165904254677e-06,
      "loss": 0.5331,
      "step": 5981
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.038694977760315,
      "learning_rate": 7.132237925271621e-06,
      "loss": 0.5276,
      "step": 5982
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9898457527160645,
      "learning_rate": 7.131309856507444e-06,
      "loss": 0.5934,
      "step": 5983
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0886754989624023,
      "learning_rate": 7.13038169800122e-06,
      "loss": 0.5706,
      "step": 5984
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0914766788482666,
      "learning_rate": 7.129453449792036e-06,
      "loss": 0.564,
      "step": 5985
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0608938932418823,
      "learning_rate": 7.1285251119189754e-06,
      "loss": 0.52,
      "step": 5986
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1529028415679932,
      "learning_rate": 7.127596684421127e-06,
      "loss": 0.5301,
      "step": 5987
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9991157650947571,
      "learning_rate": 7.126668167337583e-06,
      "loss": 0.5239,
      "step": 5988
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9912459850311279,
      "learning_rate": 7.12573956070744e-06,
      "loss": 0.573,
      "step": 5989
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0392462015151978,
      "learning_rate": 7.1248108645698e-06,
      "loss": 0.5405,
      "step": 5990
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0258725881576538,
      "learning_rate": 7.123882078963766e-06,
      "loss": 0.5184,
      "step": 5991
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0548800230026245,
      "learning_rate": 7.1229532039284455e-06,
      "loss": 0.5775,
      "step": 5992
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0427476167678833,
      "learning_rate": 7.122024239502951e-06,
      "loss": 0.5358,
      "step": 5993
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0223079919815063,
      "learning_rate": 7.121095185726399e-06,
      "loss": 0.5779,
      "step": 5994
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0239362716674805,
      "learning_rate": 7.120166042637906e-06,
      "loss": 0.5594,
      "step": 5995
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9494137167930603,
      "learning_rate": 7.119236810276598e-06,
      "loss": 0.4868,
      "step": 5996
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.037915825843811,
      "learning_rate": 7.118307488681598e-06,
      "loss": 0.5636,
      "step": 5997
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9847166538238525,
      "learning_rate": 7.11737807789204e-06,
      "loss": 0.5158,
      "step": 5998
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1142280101776123,
      "learning_rate": 7.116448577947057e-06,
      "loss": 0.5092,
      "step": 5999
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9346677660942078,
      "learning_rate": 7.115518988885785e-06,
      "loss": 0.5324,
      "step": 6000
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0473169088363647,
      "learning_rate": 7.114589310747371e-06,
      "loss": 0.5687,
      "step": 6001
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0970706939697266,
      "learning_rate": 7.113659543570956e-06,
      "loss": 0.5676,
      "step": 6002
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9799151420593262,
      "learning_rate": 7.11272968739569e-06,
      "loss": 0.4992,
      "step": 6003
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9896637201309204,
      "learning_rate": 7.1117997422607264e-06,
      "loss": 0.5302,
      "step": 6004
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9624154567718506,
      "learning_rate": 7.110869708205224e-06,
      "loss": 0.5075,
      "step": 6005
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9685401320457458,
      "learning_rate": 7.109939585268339e-06,
      "loss": 0.5583,
      "step": 6006
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.057396650314331,
      "learning_rate": 7.109009373489239e-06,
      "loss": 0.5373,
      "step": 6007
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.065391182899475,
      "learning_rate": 7.10807907290709e-06,
      "loss": 0.5856,
      "step": 6008
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.111362099647522,
      "learning_rate": 7.107148683561066e-06,
      "loss": 0.5181,
      "step": 6009
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9908615350723267,
      "learning_rate": 7.106218205490342e-06,
      "loss": 0.5563,
      "step": 6010
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.030978798866272,
      "learning_rate": 7.105287638734093e-06,
      "loss": 0.5814,
      "step": 6011
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0351821184158325,
      "learning_rate": 7.104356983331509e-06,
      "loss": 0.543,
      "step": 6012
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0720235109329224,
      "learning_rate": 7.1034262393217705e-06,
      "loss": 0.512,
      "step": 6013
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1268174648284912,
      "learning_rate": 7.1024954067440725e-06,
      "loss": 0.5858,
      "step": 6014
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.09067964553833,
      "learning_rate": 7.101564485637603e-06,
      "loss": 0.5348,
      "step": 6015
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9996499419212341,
      "learning_rate": 7.1006334760415674e-06,
      "loss": 0.5254,
      "step": 6016
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0276721715927124,
      "learning_rate": 7.0997023779951625e-06,
      "loss": 0.5662,
      "step": 6017
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.125401496887207,
      "learning_rate": 7.098771191537596e-06,
      "loss": 0.6017,
      "step": 6018
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0361688137054443,
      "learning_rate": 7.097839916708073e-06,
      "loss": 0.5537,
      "step": 6019
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1818232536315918,
      "learning_rate": 7.096908553545812e-06,
      "loss": 0.5483,
      "step": 6020
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9446892142295837,
      "learning_rate": 7.095977102090025e-06,
      "loss": 0.5471,
      "step": 6021
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.089293360710144,
      "learning_rate": 7.095045562379934e-06,
      "loss": 0.5183,
      "step": 6022
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9710480570793152,
      "learning_rate": 7.0941139344547605e-06,
      "loss": 0.5095,
      "step": 6023
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0657448768615723,
      "learning_rate": 7.093182218353737e-06,
      "loss": 0.5713,
      "step": 6024
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0107380151748657,
      "learning_rate": 7.092250414116091e-06,
      "loss": 0.4987,
      "step": 6025
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9729856848716736,
      "learning_rate": 7.091318521781058e-06,
      "loss": 0.5294,
      "step": 6026
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0331482887268066,
      "learning_rate": 7.090386541387878e-06,
      "loss": 0.5804,
      "step": 6027
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.917664110660553,
      "learning_rate": 7.089454472975792e-06,
      "loss": 0.5312,
      "step": 6028
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9795371294021606,
      "learning_rate": 7.088522316584048e-06,
      "loss": 0.5235,
      "step": 6029
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9877316355705261,
      "learning_rate": 7.087590072251893e-06,
      "loss": 0.5196,
      "step": 6030
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9603331089019775,
      "learning_rate": 7.086657740018582e-06,
      "loss": 0.5331,
      "step": 6031
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.926292896270752,
      "learning_rate": 7.085725319923373e-06,
      "loss": 0.5002,
      "step": 6032
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0340707302093506,
      "learning_rate": 7.084792812005528e-06,
      "loss": 0.5793,
      "step": 6033
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0735664367675781,
      "learning_rate": 7.083860216304309e-06,
      "loss": 0.5494,
      "step": 6034
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0449724197387695,
      "learning_rate": 7.082927532858985e-06,
      "loss": 0.5082,
      "step": 6035
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1143243312835693,
      "learning_rate": 7.0819947617088294e-06,
      "loss": 0.5323,
      "step": 6036
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0565412044525146,
      "learning_rate": 7.081061902893117e-06,
      "loss": 0.5603,
      "step": 6037
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1616407632827759,
      "learning_rate": 7.080128956451125e-06,
      "loss": 0.5288,
      "step": 6038
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.060279369354248,
      "learning_rate": 7.079195922422143e-06,
      "loss": 0.534,
      "step": 6039
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.040551781654358,
      "learning_rate": 7.078262800845453e-06,
      "loss": 0.5331,
      "step": 6040
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0278587341308594,
      "learning_rate": 7.0773295917603445e-06,
      "loss": 0.5345,
      "step": 6041
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9772768020629883,
      "learning_rate": 7.076396295206113e-06,
      "loss": 0.5115,
      "step": 6042
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0806363821029663,
      "learning_rate": 7.075462911222057e-06,
      "loss": 0.5101,
      "step": 6043
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0037871599197388,
      "learning_rate": 7.07452943984748e-06,
      "loss": 0.5258,
      "step": 6044
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9234239459037781,
      "learning_rate": 7.073595881121683e-06,
      "loss": 0.5169,
      "step": 6045
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0970360040664673,
      "learning_rate": 7.072662235083977e-06,
      "loss": 0.5895,
      "step": 6046
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.041857123374939,
      "learning_rate": 7.071728501773675e-06,
      "loss": 0.555,
      "step": 6047
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0813242197036743,
      "learning_rate": 7.070794681230093e-06,
      "loss": 0.5576,
      "step": 6048
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0041872262954712,
      "learning_rate": 7.06986077349255e-06,
      "loss": 0.5532,
      "step": 6049
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.91943359375,
      "learning_rate": 7.068926778600372e-06,
      "loss": 0.5303,
      "step": 6050
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9743295311927795,
      "learning_rate": 7.067992696592882e-06,
      "loss": 0.548,
      "step": 6051
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0182067155838013,
      "learning_rate": 7.067058527509416e-06,
      "loss": 0.5124,
      "step": 6052
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0509822368621826,
      "learning_rate": 7.066124271389305e-06,
      "loss": 0.483,
      "step": 6053
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9800784587860107,
      "learning_rate": 7.0651899282718896e-06,
      "loss": 0.578,
      "step": 6054
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1136858463287354,
      "learning_rate": 7.064255498196509e-06,
      "loss": 0.5482,
      "step": 6055
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0547521114349365,
      "learning_rate": 7.0633209812025116e-06,
      "loss": 0.5467,
      "step": 6056
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9685391187667847,
      "learning_rate": 7.062386377329245e-06,
      "loss": 0.5177,
      "step": 6057
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0092555284500122,
      "learning_rate": 7.061451686616062e-06,
      "loss": 0.5246,
      "step": 6058
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.109516978263855,
      "learning_rate": 7.0605169091023205e-06,
      "loss": 0.5814,
      "step": 6059
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1080363988876343,
      "learning_rate": 7.05958204482738e-06,
      "loss": 0.5113,
      "step": 6060
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1008321046829224,
      "learning_rate": 7.058647093830604e-06,
      "loss": 0.5606,
      "step": 6061
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.029272437095642,
      "learning_rate": 7.0577120561513604e-06,
      "loss": 0.5364,
      "step": 6062
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0312248468399048,
      "learning_rate": 7.056776931829021e-06,
      "loss": 0.488,
      "step": 6063
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0757166147232056,
      "learning_rate": 7.055841720902959e-06,
      "loss": 0.5122,
      "step": 6064
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0667598247528076,
      "learning_rate": 7.054906423412554e-06,
      "loss": 0.5817,
      "step": 6065
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0422972440719604,
      "learning_rate": 7.053971039397188e-06,
      "loss": 0.5785,
      "step": 6066
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0605396032333374,
      "learning_rate": 7.0530355688962484e-06,
      "loss": 0.5636,
      "step": 6067
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.014585018157959,
      "learning_rate": 7.0521000119491215e-06,
      "loss": 0.5179,
      "step": 6068
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1167665719985962,
      "learning_rate": 7.0511643685952014e-06,
      "loss": 0.5565,
      "step": 6069
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0698355436325073,
      "learning_rate": 7.050228638873886e-06,
      "loss": 0.5141,
      "step": 6070
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0831199884414673,
      "learning_rate": 7.049292822824575e-06,
      "loss": 0.5896,
      "step": 6071
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.045108437538147,
      "learning_rate": 7.048356920486672e-06,
      "loss": 0.558,
      "step": 6072
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0129737854003906,
      "learning_rate": 7.047420931899585e-06,
      "loss": 0.5345,
      "step": 6073
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0114727020263672,
      "learning_rate": 7.0464848571027246e-06,
      "loss": 0.5566,
      "step": 6074
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0448603630065918,
      "learning_rate": 7.045548696135506e-06,
      "loss": 0.5243,
      "step": 6075
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.061661958694458,
      "learning_rate": 7.044612449037348e-06,
      "loss": 0.5884,
      "step": 6076
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.001895546913147,
      "learning_rate": 7.0436761158476715e-06,
      "loss": 0.5432,
      "step": 6077
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.007230281829834,
      "learning_rate": 7.042739696605905e-06,
      "loss": 0.5211,
      "step": 6078
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9304594993591309,
      "learning_rate": 7.041803191351475e-06,
      "loss": 0.5272,
      "step": 6079
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.011073350906372,
      "learning_rate": 7.040866600123816e-06,
      "loss": 0.5678,
      "step": 6080
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0891047716140747,
      "learning_rate": 7.039929922962363e-06,
      "loss": 0.5529,
      "step": 6081
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0254418849945068,
      "learning_rate": 7.038993159906558e-06,
      "loss": 0.5572,
      "step": 6082
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9843958020210266,
      "learning_rate": 7.0380563109958445e-06,
      "loss": 0.5261,
      "step": 6083
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.045296549797058,
      "learning_rate": 7.03711937626967e-06,
      "loss": 0.547,
      "step": 6084
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9561773538589478,
      "learning_rate": 7.036182355767485e-06,
      "loss": 0.515,
      "step": 6085
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9849733114242554,
      "learning_rate": 7.0352452495287435e-06,
      "loss": 0.5254,
      "step": 6086
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0817680358886719,
      "learning_rate": 7.034308057592907e-06,
      "loss": 0.4934,
      "step": 6087
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0080195665359497,
      "learning_rate": 7.033370779999431e-06,
      "loss": 0.5738,
      "step": 6088
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9855676293373108,
      "learning_rate": 7.032433416787788e-06,
      "loss": 0.5446,
      "step": 6089
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9518822431564331,
      "learning_rate": 7.031495967997444e-06,
      "loss": 0.4807,
      "step": 6090
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9908321499824524,
      "learning_rate": 7.0305584336678715e-06,
      "loss": 0.5243,
      "step": 6091
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9574987888336182,
      "learning_rate": 7.029620813838544e-06,
      "loss": 0.502,
      "step": 6092
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.018323302268982,
      "learning_rate": 7.02868310854895e-06,
      "loss": 0.516,
      "step": 6093
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0103620290756226,
      "learning_rate": 7.027745317838564e-06,
      "loss": 0.5337,
      "step": 6094
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0037866830825806,
      "learning_rate": 7.026807441746879e-06,
      "loss": 0.5451,
      "step": 6095
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0321918725967407,
      "learning_rate": 7.025869480313381e-06,
      "loss": 0.5442,
      "step": 6096
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0174425840377808,
      "learning_rate": 7.0249314335775675e-06,
      "loss": 0.5187,
      "step": 6097
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9343465566635132,
      "learning_rate": 7.023993301578935e-06,
      "loss": 0.4886,
      "step": 6098
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0583616495132446,
      "learning_rate": 7.023055084356987e-06,
      "loss": 0.4964,
      "step": 6099
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.017966866493225,
      "learning_rate": 7.022116781951226e-06,
      "loss": 0.5475,
      "step": 6100
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9800785779953003,
      "learning_rate": 7.021178394401162e-06,
      "loss": 0.5518,
      "step": 6101
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9927045702934265,
      "learning_rate": 7.020239921746308e-06,
      "loss": 0.5416,
      "step": 6102
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0396851301193237,
      "learning_rate": 7.019301364026178e-06,
      "loss": 0.5618,
      "step": 6103
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.045152187347412,
      "learning_rate": 7.018362721280292e-06,
      "loss": 0.5206,
      "step": 6104
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0131950378417969,
      "learning_rate": 7.0174239935481735e-06,
      "loss": 0.5591,
      "step": 6105
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0114343166351318,
      "learning_rate": 7.016485180869349e-06,
      "loss": 0.5126,
      "step": 6106
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.071434736251831,
      "learning_rate": 7.015546283283346e-06,
      "loss": 0.4994,
      "step": 6107
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.11801278591156,
      "learning_rate": 7.014607300829703e-06,
      "loss": 0.5724,
      "step": 6108
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0889805555343628,
      "learning_rate": 7.013668233547955e-06,
      "loss": 0.5321,
      "step": 6109
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.009221076965332,
      "learning_rate": 7.0127290814776424e-06,
      "loss": 0.5145,
      "step": 6110
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.04290771484375,
      "learning_rate": 7.0117898446583084e-06,
      "loss": 0.5851,
      "step": 6111
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0475056171417236,
      "learning_rate": 7.010850523129504e-06,
      "loss": 0.5237,
      "step": 6112
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.01203453540802,
      "learning_rate": 7.009911116930779e-06,
      "loss": 0.5497,
      "step": 6113
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1558852195739746,
      "learning_rate": 7.00897162610169e-06,
      "loss": 0.5615,
      "step": 6114
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0910730361938477,
      "learning_rate": 7.0080320506817926e-06,
      "loss": 0.5076,
      "step": 6115
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9866907000541687,
      "learning_rate": 7.007092390710652e-06,
      "loss": 0.525,
      "step": 6116
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9702544808387756,
      "learning_rate": 7.006152646227833e-06,
      "loss": 0.4603,
      "step": 6117
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9692726731300354,
      "learning_rate": 7.005212817272905e-06,
      "loss": 0.5746,
      "step": 6118
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0775811672210693,
      "learning_rate": 7.0042729038854405e-06,
      "loss": 0.5711,
      "step": 6119
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9793890118598938,
      "learning_rate": 7.003332906105016e-06,
      "loss": 0.5655,
      "step": 6120
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0329787731170654,
      "learning_rate": 7.002392823971214e-06,
      "loss": 0.5754,
      "step": 6121
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9337785840034485,
      "learning_rate": 7.001452657523614e-06,
      "loss": 0.4502,
      "step": 6122
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0856430530548096,
      "learning_rate": 7.000512406801805e-06,
      "loss": 0.5455,
      "step": 6123
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9802913069725037,
      "learning_rate": 6.9995720718453786e-06,
      "loss": 0.518,
      "step": 6124
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1211453676223755,
      "learning_rate": 6.998631652693928e-06,
      "loss": 0.5777,
      "step": 6125
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.989659309387207,
      "learning_rate": 6.997691149387052e-06,
      "loss": 0.5355,
      "step": 6126
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9833707809448242,
      "learning_rate": 6.99675056196435e-06,
      "loss": 0.528,
      "step": 6127
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9955346584320068,
      "learning_rate": 6.995809890465428e-06,
      "loss": 0.5341,
      "step": 6128
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9848581552505493,
      "learning_rate": 6.994869134929895e-06,
      "loss": 0.56,
      "step": 6129
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0137369632720947,
      "learning_rate": 6.993928295397363e-06,
      "loss": 0.5487,
      "step": 6130
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0635900497436523,
      "learning_rate": 6.992987371907446e-06,
      "loss": 0.5587,
      "step": 6131
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9602803587913513,
      "learning_rate": 6.992046364499764e-06,
      "loss": 0.5159,
      "step": 6132
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9842841029167175,
      "learning_rate": 6.991105273213939e-06,
      "loss": 0.491,
      "step": 6133
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0112354755401611,
      "learning_rate": 6.990164098089598e-06,
      "loss": 0.539,
      "step": 6134
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9850631356239319,
      "learning_rate": 6.9892228391663694e-06,
      "loss": 0.5184,
      "step": 6135
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9878668189048767,
      "learning_rate": 6.988281496483888e-06,
      "loss": 0.5472,
      "step": 6136
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0139793157577515,
      "learning_rate": 6.987340070081789e-06,
      "loss": 0.5217,
      "step": 6137
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.041130781173706,
      "learning_rate": 6.986398559999714e-06,
      "loss": 0.5372,
      "step": 6138
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9763038158416748,
      "learning_rate": 6.9854569662773044e-06,
      "loss": 0.4826,
      "step": 6139
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0625923871994019,
      "learning_rate": 6.984515288954211e-06,
      "loss": 0.5403,
      "step": 6140
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0266550779342651,
      "learning_rate": 6.98357352807008e-06,
      "loss": 0.4992,
      "step": 6141
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9099301099777222,
      "learning_rate": 6.982631683664569e-06,
      "loss": 0.4971,
      "step": 6142
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9947066903114319,
      "learning_rate": 6.981689755777335e-06,
      "loss": 0.48,
      "step": 6143
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9713681936264038,
      "learning_rate": 6.98074774444804e-06,
      "loss": 0.4948,
      "step": 6144
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9177002310752869,
      "learning_rate": 6.979805649716347e-06,
      "loss": 0.4484,
      "step": 6145
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0028992891311646,
      "learning_rate": 6.978863471621925e-06,
      "loss": 0.6121,
      "step": 6146
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.065828561782837,
      "learning_rate": 6.977921210204446e-06,
      "loss": 0.5806,
      "step": 6147
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9567848443984985,
      "learning_rate": 6.9769788655035875e-06,
      "loss": 0.5258,
      "step": 6148
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0322932004928589,
      "learning_rate": 6.976036437559024e-06,
      "loss": 0.6068,
      "step": 6149
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9327954053878784,
      "learning_rate": 6.975093926410441e-06,
      "loss": 0.5294,
      "step": 6150
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0788025856018066,
      "learning_rate": 6.974151332097525e-06,
      "loss": 0.5618,
      "step": 6151
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0738791227340698,
      "learning_rate": 6.973208654659962e-06,
      "loss": 0.5739,
      "step": 6152
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1076157093048096,
      "learning_rate": 6.9722658941374475e-06,
      "loss": 0.5123,
      "step": 6153
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8980148434638977,
      "learning_rate": 6.971323050569677e-06,
      "loss": 0.5308,
      "step": 6154
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9834483861923218,
      "learning_rate": 6.970380123996352e-06,
      "loss": 0.5193,
      "step": 6155
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1332507133483887,
      "learning_rate": 6.969437114457174e-06,
      "loss": 0.5675,
      "step": 6156
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9124565124511719,
      "learning_rate": 6.968494021991848e-06,
      "loss": 0.5158,
      "step": 6157
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0040991306304932,
      "learning_rate": 6.967550846640089e-06,
      "loss": 0.4864,
      "step": 6158
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.040360689163208,
      "learning_rate": 6.966607588441609e-06,
      "loss": 0.5617,
      "step": 6159
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9898555278778076,
      "learning_rate": 6.9656642474361225e-06,
      "loss": 0.4797,
      "step": 6160
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9709599614143372,
      "learning_rate": 6.964720823663353e-06,
      "loss": 0.5402,
      "step": 6161
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0226396322250366,
      "learning_rate": 6.963777317163025e-06,
      "loss": 0.5439,
      "step": 6162
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.963808000087738,
      "learning_rate": 6.962833727974867e-06,
      "loss": 0.5229,
      "step": 6163
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0142179727554321,
      "learning_rate": 6.961890056138607e-06,
      "loss": 0.5408,
      "step": 6164
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0173602104187012,
      "learning_rate": 6.9609463016939816e-06,
      "loss": 0.5459,
      "step": 6165
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.948390781879425,
      "learning_rate": 6.960002464680731e-06,
      "loss": 0.4954,
      "step": 6166
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9785526990890503,
      "learning_rate": 6.959058545138593e-06,
      "loss": 0.512,
      "step": 6167
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1291639804840088,
      "learning_rate": 6.958114543107315e-06,
      "loss": 0.5671,
      "step": 6168
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9646851420402527,
      "learning_rate": 6.957170458626645e-06,
      "loss": 0.4931,
      "step": 6169
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0027590990066528,
      "learning_rate": 6.956226291736338e-06,
      "loss": 0.5538,
      "step": 6170
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0033183097839355,
      "learning_rate": 6.955282042476144e-06,
      "loss": 0.5133,
      "step": 6171
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.014385461807251,
      "learning_rate": 6.9543377108858265e-06,
      "loss": 0.5415,
      "step": 6172
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0565842390060425,
      "learning_rate": 6.9533932970051465e-06,
      "loss": 0.5239,
      "step": 6173
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0331782102584839,
      "learning_rate": 6.952448800873871e-06,
      "loss": 0.5959,
      "step": 6174
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0689690113067627,
      "learning_rate": 6.951504222531768e-06,
      "loss": 0.5552,
      "step": 6175
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9959222078323364,
      "learning_rate": 6.950559562018611e-06,
      "loss": 0.5421,
      "step": 6176
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9821951985359192,
      "learning_rate": 6.949614819374175e-06,
      "loss": 0.5326,
      "step": 6177
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0054478645324707,
      "learning_rate": 6.948669994638243e-06,
      "loss": 0.543,
      "step": 6178
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9514632821083069,
      "learning_rate": 6.947725087850595e-06,
      "loss": 0.4761,
      "step": 6179
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0094581842422485,
      "learning_rate": 6.94678009905102e-06,
      "loss": 0.5214,
      "step": 6180
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0191506147384644,
      "learning_rate": 6.945835028279308e-06,
      "loss": 0.5829,
      "step": 6181
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0834375619888306,
      "learning_rate": 6.944889875575251e-06,
      "loss": 0.5345,
      "step": 6182
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9999750852584839,
      "learning_rate": 6.943944640978648e-06,
      "loss": 0.5565,
      "step": 6183
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0202056169509888,
      "learning_rate": 6.942999324529297e-06,
      "loss": 0.5142,
      "step": 6184
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.000248908996582,
      "learning_rate": 6.942053926267005e-06,
      "loss": 0.5268,
      "step": 6185
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0070136785507202,
      "learning_rate": 6.941108446231578e-06,
      "loss": 0.533,
      "step": 6186
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9712101221084595,
      "learning_rate": 6.940162884462828e-06,
      "loss": 0.5313,
      "step": 6187
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9556286334991455,
      "learning_rate": 6.9392172410005656e-06,
      "loss": 0.5484,
      "step": 6188
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0260111093521118,
      "learning_rate": 6.9382715158846135e-06,
      "loss": 0.5298,
      "step": 6189
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.06009840965271,
      "learning_rate": 6.93732570915479e-06,
      "loss": 0.5274,
      "step": 6190
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.088383674621582,
      "learning_rate": 6.93637982085092e-06,
      "loss": 0.5408,
      "step": 6191
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.039387583732605,
      "learning_rate": 6.9354338510128315e-06,
      "loss": 0.5236,
      "step": 6192
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0569809675216675,
      "learning_rate": 6.934487799680357e-06,
      "loss": 0.5541,
      "step": 6193
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9675583243370056,
      "learning_rate": 6.933541666893331e-06,
      "loss": 0.5471,
      "step": 6194
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0061856508255005,
      "learning_rate": 6.932595452691592e-06,
      "loss": 0.5677,
      "step": 6195
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9990823268890381,
      "learning_rate": 6.9316491571149815e-06,
      "loss": 0.5449,
      "step": 6196
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0470032691955566,
      "learning_rate": 6.930702780203344e-06,
      "loss": 0.5731,
      "step": 6197
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2246379852294922,
      "learning_rate": 6.929756321996529e-06,
      "loss": 0.5755,
      "step": 6198
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0149028301239014,
      "learning_rate": 6.928809782534388e-06,
      "loss": 0.545,
      "step": 6199
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0562412738800049,
      "learning_rate": 6.927863161856778e-06,
      "loss": 0.5309,
      "step": 6200
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.009595274925232,
      "learning_rate": 6.9269164600035555e-06,
      "loss": 0.5182,
      "step": 6201
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0458571910858154,
      "learning_rate": 6.925969677014585e-06,
      "loss": 0.5708,
      "step": 6202
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0821754932403564,
      "learning_rate": 6.92502281292973e-06,
      "loss": 0.5318,
      "step": 6203
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9944037795066833,
      "learning_rate": 6.924075867788863e-06,
      "loss": 0.5421,
      "step": 6204
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0907917022705078,
      "learning_rate": 6.923128841631854e-06,
      "loss": 0.4908,
      "step": 6205
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0934239625930786,
      "learning_rate": 6.92218173449858e-06,
      "loss": 0.5125,
      "step": 6206
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0188615322113037,
      "learning_rate": 6.921234546428918e-06,
      "loss": 0.5283,
      "step": 6207
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9371731281280518,
      "learning_rate": 6.920287277462755e-06,
      "loss": 0.5093,
      "step": 6208
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9355340003967285,
      "learning_rate": 6.9193399276399745e-06,
      "loss": 0.4871,
      "step": 6209
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0003212690353394,
      "learning_rate": 6.918392497000466e-06,
      "loss": 0.554,
      "step": 6210
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.91237872838974,
      "learning_rate": 6.917444985584122e-06,
      "loss": 0.4917,
      "step": 6211
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.011143445968628,
      "learning_rate": 6.916497393430841e-06,
      "loss": 0.5489,
      "step": 6212
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9575095772743225,
      "learning_rate": 6.915549720580523e-06,
      "loss": 0.5222,
      "step": 6213
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0624295473098755,
      "learning_rate": 6.914601967073068e-06,
      "loss": 0.4989,
      "step": 6214
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0963515043258667,
      "learning_rate": 6.913654132948385e-06,
      "loss": 0.5305,
      "step": 6215
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9703376293182373,
      "learning_rate": 6.912706218246384e-06,
      "loss": 0.481,
      "step": 6216
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9639180898666382,
      "learning_rate": 6.911758223006979e-06,
      "loss": 0.5314,
      "step": 6217
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0230478048324585,
      "learning_rate": 6.910810147270084e-06,
      "loss": 0.4934,
      "step": 6218
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0991566181182861,
      "learning_rate": 6.909861991075622e-06,
      "loss": 0.5647,
      "step": 6219
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0460660457611084,
      "learning_rate": 6.908913754463514e-06,
      "loss": 0.5524,
      "step": 6220
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0755608081817627,
      "learning_rate": 6.90796543747369e-06,
      "loss": 0.5864,
      "step": 6221
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.017244815826416,
      "learning_rate": 6.907017040146078e-06,
      "loss": 0.5585,
      "step": 6222
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0522984266281128,
      "learning_rate": 6.906068562520613e-06,
      "loss": 0.5338,
      "step": 6223
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9456668496131897,
      "learning_rate": 6.905120004637232e-06,
      "loss": 0.4864,
      "step": 6224
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0045950412750244,
      "learning_rate": 6.904171366535873e-06,
      "loss": 0.5382,
      "step": 6225
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9706365466117859,
      "learning_rate": 6.9032226482564835e-06,
      "loss": 0.5119,
      "step": 6226
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0931531190872192,
      "learning_rate": 6.9022738498390084e-06,
      "loss": 0.5717,
      "step": 6227
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0188103914260864,
      "learning_rate": 6.9013249713234e-06,
      "loss": 0.5502,
      "step": 6228
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9862955212593079,
      "learning_rate": 6.900376012749611e-06,
      "loss": 0.5273,
      "step": 6229
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.095741629600525,
      "learning_rate": 6.899426974157598e-06,
      "loss": 0.545,
      "step": 6230
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9772562384605408,
      "learning_rate": 6.898477855587323e-06,
      "loss": 0.5392,
      "step": 6231
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.002846360206604,
      "learning_rate": 6.897528657078752e-06,
      "loss": 0.5283,
      "step": 6232
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0799317359924316,
      "learning_rate": 6.8965793786718484e-06,
      "loss": 0.5303,
      "step": 6233
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9295328259468079,
      "learning_rate": 6.895630020406584e-06,
      "loss": 0.5266,
      "step": 6234
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0229262113571167,
      "learning_rate": 6.894680582322934e-06,
      "loss": 0.5467,
      "step": 6235
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9982120990753174,
      "learning_rate": 6.893731064460878e-06,
      "loss": 0.5524,
      "step": 6236
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0514452457427979,
      "learning_rate": 6.892781466860393e-06,
      "loss": 0.5273,
      "step": 6237
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9818599820137024,
      "learning_rate": 6.891831789561465e-06,
      "loss": 0.545,
      "step": 6238
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0682156085968018,
      "learning_rate": 6.8908820326040815e-06,
      "loss": 0.5374,
      "step": 6239
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.98082035779953,
      "learning_rate": 6.889932196028235e-06,
      "loss": 0.5561,
      "step": 6240
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0529831647872925,
      "learning_rate": 6.888982279873917e-06,
      "loss": 0.5152,
      "step": 6241
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.096386432647705,
      "learning_rate": 6.888032284181127e-06,
      "loss": 0.514,
      "step": 6242
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1044929027557373,
      "learning_rate": 6.887082208989865e-06,
      "loss": 0.5502,
      "step": 6243
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0671653747558594,
      "learning_rate": 6.886132054340136e-06,
      "loss": 0.5492,
      "step": 6244
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9476322531700134,
      "learning_rate": 6.885181820271947e-06,
      "loss": 0.5347,
      "step": 6245
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.03954017162323,
      "learning_rate": 6.88423150682531e-06,
      "loss": 0.5653,
      "step": 6246
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0172086954116821,
      "learning_rate": 6.88328111404024e-06,
      "loss": 0.502,
      "step": 6247
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0906752347946167,
      "learning_rate": 6.882330641956752e-06,
      "loss": 0.5648,
      "step": 6248
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9495219588279724,
      "learning_rate": 6.881380090614871e-06,
      "loss": 0.5056,
      "step": 6249
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.050155520439148,
      "learning_rate": 6.8804294600546175e-06,
      "loss": 0.5382,
      "step": 6250
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0673195123672485,
      "learning_rate": 6.879478750316022e-06,
      "loss": 0.5611,
      "step": 6251
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0704740285873413,
      "learning_rate": 6.878527961439113e-06,
      "loss": 0.5519,
      "step": 6252
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.04393470287323,
      "learning_rate": 6.877577093463927e-06,
      "loss": 0.5399,
      "step": 6253
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0113558769226074,
      "learning_rate": 6.876626146430502e-06,
      "loss": 0.5227,
      "step": 6254
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0777956247329712,
      "learning_rate": 6.875675120378878e-06,
      "loss": 0.5607,
      "step": 6255
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9694634675979614,
      "learning_rate": 6.8747240153491e-06,
      "loss": 0.5241,
      "step": 6256
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9827180504798889,
      "learning_rate": 6.873772831381214e-06,
      "loss": 0.5364,
      "step": 6257
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1433156728744507,
      "learning_rate": 6.872821568515275e-06,
      "loss": 0.592,
      "step": 6258
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0650519132614136,
      "learning_rate": 6.8718702267913325e-06,
      "loss": 0.5413,
      "step": 6259
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0455180406570435,
      "learning_rate": 6.870918806249449e-06,
      "loss": 0.546,
      "step": 6260
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9944682717323303,
      "learning_rate": 6.8699673069296806e-06,
      "loss": 0.5242,
      "step": 6261
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9604764580726624,
      "learning_rate": 6.869015728872095e-06,
      "loss": 0.5627,
      "step": 6262
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.064626932144165,
      "learning_rate": 6.868064072116758e-06,
      "loss": 0.5855,
      "step": 6263
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9918391704559326,
      "learning_rate": 6.867112336703743e-06,
      "loss": 0.5058,
      "step": 6264
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.035409927368164,
      "learning_rate": 6.866160522673121e-06,
      "loss": 0.5655,
      "step": 6265
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.93015056848526,
      "learning_rate": 6.865208630064973e-06,
      "loss": 0.5595,
      "step": 6266
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0266493558883667,
      "learning_rate": 6.864256658919377e-06,
      "loss": 0.4955,
      "step": 6267
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.025432825088501,
      "learning_rate": 6.8633046092764174e-06,
      "loss": 0.5654,
      "step": 6268
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0468451976776123,
      "learning_rate": 6.862352481176184e-06,
      "loss": 0.5787,
      "step": 6269
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1604524850845337,
      "learning_rate": 6.861400274658767e-06,
      "loss": 0.5261,
      "step": 6270
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0462700128555298,
      "learning_rate": 6.860447989764259e-06,
      "loss": 0.5194,
      "step": 6271
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.015401840209961,
      "learning_rate": 6.8594956265327585e-06,
      "loss": 0.5136,
      "step": 6272
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0995614528656006,
      "learning_rate": 6.858543185004365e-06,
      "loss": 0.5547,
      "step": 6273
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0032482147216797,
      "learning_rate": 6.857590665219185e-06,
      "loss": 0.5629,
      "step": 6274
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9372909665107727,
      "learning_rate": 6.856638067217324e-06,
      "loss": 0.4823,
      "step": 6275
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.068374752998352,
      "learning_rate": 6.85568539103889e-06,
      "loss": 0.5269,
      "step": 6276
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.093066692352295,
      "learning_rate": 6.854732636724002e-06,
      "loss": 0.5849,
      "step": 6277
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9396721720695496,
      "learning_rate": 6.853779804312775e-06,
      "loss": 0.4918,
      "step": 6278
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1168854236602783,
      "learning_rate": 6.8528268938453295e-06,
      "loss": 0.4996,
      "step": 6279
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0356525182724,
      "learning_rate": 6.851873905361786e-06,
      "loss": 0.5363,
      "step": 6280
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0906376838684082,
      "learning_rate": 6.850920838902278e-06,
      "loss": 0.5743,
      "step": 6281
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.061261773109436,
      "learning_rate": 6.84996769450693e-06,
      "loss": 0.4891,
      "step": 6282
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.131471872329712,
      "learning_rate": 6.84901447221588e-06,
      "loss": 0.5819,
      "step": 6283
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0375972986221313,
      "learning_rate": 6.84806117206926e-06,
      "loss": 0.5819,
      "step": 6284
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0283234119415283,
      "learning_rate": 6.847107794107216e-06,
      "loss": 0.4991,
      "step": 6285
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0950448513031006,
      "learning_rate": 6.846154338369887e-06,
      "loss": 0.5785,
      "step": 6286
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9550922513008118,
      "learning_rate": 6.845200804897421e-06,
      "loss": 0.529,
      "step": 6287
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0447999238967896,
      "learning_rate": 6.844247193729968e-06,
      "loss": 0.5736,
      "step": 6288
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9329219460487366,
      "learning_rate": 6.843293504907682e-06,
      "loss": 0.482,
      "step": 6289
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.013639211654663,
      "learning_rate": 6.84233973847072e-06,
      "loss": 0.5383,
      "step": 6290
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.048012375831604,
      "learning_rate": 6.8413858944592385e-06,
      "loss": 0.5864,
      "step": 6291
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0305635929107666,
      "learning_rate": 6.840431972913404e-06,
      "loss": 0.5534,
      "step": 6292
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0479077100753784,
      "learning_rate": 6.83947797387338e-06,
      "loss": 0.5109,
      "step": 6293
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9554098844528198,
      "learning_rate": 6.838523897379339e-06,
      "loss": 0.5521,
      "step": 6294
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9382658004760742,
      "learning_rate": 6.837569743471451e-06,
      "loss": 0.5413,
      "step": 6295
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0406569242477417,
      "learning_rate": 6.836615512189895e-06,
      "loss": 0.4994,
      "step": 6296
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0103338956832886,
      "learning_rate": 6.835661203574848e-06,
      "loss": 0.5333,
      "step": 6297
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0147366523742676,
      "learning_rate": 6.834706817666495e-06,
      "loss": 0.5447,
      "step": 6298
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9934282302856445,
      "learning_rate": 6.833752354505019e-06,
      "loss": 0.5232,
      "step": 6299
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0535008907318115,
      "learning_rate": 6.832797814130611e-06,
      "loss": 0.5237,
      "step": 6300
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9556391835212708,
      "learning_rate": 6.831843196583462e-06,
      "loss": 0.5341,
      "step": 6301
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9664854407310486,
      "learning_rate": 6.8308885019037695e-06,
      "loss": 0.524,
      "step": 6302
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0064958333969116,
      "learning_rate": 6.82993373013173e-06,
      "loss": 0.5478,
      "step": 6303
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0353161096572876,
      "learning_rate": 6.8289788813075485e-06,
      "loss": 0.5263,
      "step": 6304
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0607702732086182,
      "learning_rate": 6.82802395547143e-06,
      "loss": 0.5079,
      "step": 6305
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9825518131256104,
      "learning_rate": 6.82706895266358e-06,
      "loss": 0.4533,
      "step": 6306
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.044676423072815,
      "learning_rate": 6.826113872924213e-06,
      "loss": 0.5691,
      "step": 6307
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0349632501602173,
      "learning_rate": 6.825158716293543e-06,
      "loss": 0.5567,
      "step": 6308
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9872351884841919,
      "learning_rate": 6.824203482811788e-06,
      "loss": 0.5289,
      "step": 6309
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9518697261810303,
      "learning_rate": 6.823248172519173e-06,
      "loss": 0.5069,
      "step": 6310
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.042443871498108,
      "learning_rate": 6.8222927854559175e-06,
      "loss": 0.5468,
      "step": 6311
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.993451714515686,
      "learning_rate": 6.8213373216622514e-06,
      "loss": 0.5224,
      "step": 6312
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.061415672302246,
      "learning_rate": 6.820381781178409e-06,
      "loss": 0.5395,
      "step": 6313
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.051388144493103,
      "learning_rate": 6.819426164044622e-06,
      "loss": 0.5393,
      "step": 6314
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0344637632369995,
      "learning_rate": 6.818470470301128e-06,
      "loss": 0.5154,
      "step": 6315
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0458735227584839,
      "learning_rate": 6.817514699988168e-06,
      "loss": 0.5319,
      "step": 6316
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0880186557769775,
      "learning_rate": 6.8165588531459885e-06,
      "loss": 0.5491,
      "step": 6317
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0239330530166626,
      "learning_rate": 6.815602929814833e-06,
      "loss": 0.5305,
      "step": 6318
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0305886268615723,
      "learning_rate": 6.814646930034954e-06,
      "loss": 0.58,
      "step": 6319
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9396848082542419,
      "learning_rate": 6.813690853846606e-06,
      "loss": 0.5338,
      "step": 6320
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0301604270935059,
      "learning_rate": 6.8127347012900465e-06,
      "loss": 0.5256,
      "step": 6321
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0537376403808594,
      "learning_rate": 6.811778472405534e-06,
      "loss": 0.5524,
      "step": 6322
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9582347273826599,
      "learning_rate": 6.810822167233333e-06,
      "loss": 0.5207,
      "step": 6323
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0056788921356201,
      "learning_rate": 6.80986578581371e-06,
      "loss": 0.5508,
      "step": 6324
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.061269760131836,
      "learning_rate": 6.808909328186934e-06,
      "loss": 0.5671,
      "step": 6325
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9477720260620117,
      "learning_rate": 6.80795279439328e-06,
      "loss": 0.4916,
      "step": 6326
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0195330381393433,
      "learning_rate": 6.806996184473023e-06,
      "loss": 0.5624,
      "step": 6327
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0175155401229858,
      "learning_rate": 6.806039498466444e-06,
      "loss": 0.5314,
      "step": 6328
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9943006038665771,
      "learning_rate": 6.805082736413822e-06,
      "loss": 0.562,
      "step": 6329
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1058787107467651,
      "learning_rate": 6.804125898355447e-06,
      "loss": 0.5336,
      "step": 6330
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9576951265335083,
      "learning_rate": 6.8031689843316054e-06,
      "loss": 0.4966,
      "step": 6331
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.948913037776947,
      "learning_rate": 6.802211994382591e-06,
      "loss": 0.5148,
      "step": 6332
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9891049265861511,
      "learning_rate": 6.8012549285487e-06,
      "loss": 0.5436,
      "step": 6333
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0424014329910278,
      "learning_rate": 6.800297786870228e-06,
      "loss": 0.5378,
      "step": 6334
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9966731071472168,
      "learning_rate": 6.799340569387481e-06,
      "loss": 0.4889,
      "step": 6335
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0427660942077637,
      "learning_rate": 6.798383276140761e-06,
      "loss": 0.5412,
      "step": 6336
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0151562690734863,
      "learning_rate": 6.797425907170378e-06,
      "loss": 0.5129,
      "step": 6337
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9394248723983765,
      "learning_rate": 6.796468462516642e-06,
      "loss": 0.5167,
      "step": 6338
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.140164852142334,
      "learning_rate": 6.79551094221987e-06,
      "loss": 0.5403,
      "step": 6339
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0196154117584229,
      "learning_rate": 6.794553346320376e-06,
      "loss": 0.4968,
      "step": 6340
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9896392226219177,
      "learning_rate": 6.7935956748584855e-06,
      "loss": 0.4752,
      "step": 6341
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9792375564575195,
      "learning_rate": 6.792637927874519e-06,
      "loss": 0.5354,
      "step": 6342
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9903654456138611,
      "learning_rate": 6.791680105408807e-06,
      "loss": 0.5889,
      "step": 6343
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0461080074310303,
      "learning_rate": 6.790722207501678e-06,
      "loss": 0.5864,
      "step": 6344
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0412373542785645,
      "learning_rate": 6.789764234193465e-06,
      "loss": 0.5023,
      "step": 6345
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0502227544784546,
      "learning_rate": 6.788806185524508e-06,
      "loss": 0.566,
      "step": 6346
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0583373308181763,
      "learning_rate": 6.787848061535145e-06,
      "loss": 0.5976,
      "step": 6347
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0030248165130615,
      "learning_rate": 6.786889862265719e-06,
      "loss": 0.5051,
      "step": 6348
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.01585853099823,
      "learning_rate": 6.7859315877565775e-06,
      "loss": 0.4992,
      "step": 6349
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0228488445281982,
      "learning_rate": 6.784973238048069e-06,
      "loss": 0.5788,
      "step": 6350
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9880842566490173,
      "learning_rate": 6.7840148131805485e-06,
      "loss": 0.5526,
      "step": 6351
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9535727500915527,
      "learning_rate": 6.783056313194369e-06,
      "loss": 0.5104,
      "step": 6352
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8931145668029785,
      "learning_rate": 6.7820977381298915e-06,
      "loss": 0.5571,
      "step": 6353
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.989362895488739,
      "learning_rate": 6.781139088027477e-06,
      "loss": 0.5485,
      "step": 6354
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.040856122970581,
      "learning_rate": 6.780180362927492e-06,
      "loss": 0.5125,
      "step": 6355
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.996364414691925,
      "learning_rate": 6.779221562870306e-06,
      "loss": 0.514,
      "step": 6356
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1184418201446533,
      "learning_rate": 6.778262687896287e-06,
      "loss": 0.5479,
      "step": 6357
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1050505638122559,
      "learning_rate": 6.777303738045814e-06,
      "loss": 0.5934,
      "step": 6358
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9124274253845215,
      "learning_rate": 6.776344713359263e-06,
      "loss": 0.5294,
      "step": 6359
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0172024965286255,
      "learning_rate": 6.775385613877016e-06,
      "loss": 0.5622,
      "step": 6360
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.023863673210144,
      "learning_rate": 6.774426439639455e-06,
      "loss": 0.5271,
      "step": 6361
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0649785995483398,
      "learning_rate": 6.773467190686972e-06,
      "loss": 0.5539,
      "step": 6362
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.064691424369812,
      "learning_rate": 6.772507867059953e-06,
      "loss": 0.5705,
      "step": 6363
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0454376935958862,
      "learning_rate": 6.771548468798796e-06,
      "loss": 0.4621,
      "step": 6364
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0180542469024658,
      "learning_rate": 6.770588995943893e-06,
      "loss": 0.496,
      "step": 6365
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0939666032791138,
      "learning_rate": 6.769629448535648e-06,
      "loss": 0.5496,
      "step": 6366
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0308752059936523,
      "learning_rate": 6.768669826614464e-06,
      "loss": 0.5346,
      "step": 6367
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0306459665298462,
      "learning_rate": 6.767710130220745e-06,
      "loss": 0.5722,
      "step": 6368
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9302669167518616,
      "learning_rate": 6.766750359394904e-06,
      "loss": 0.5109,
      "step": 6369
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0190123319625854,
      "learning_rate": 6.76579051417735e-06,
      "loss": 0.6004,
      "step": 6370
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0275017023086548,
      "learning_rate": 6.7648305946085e-06,
      "loss": 0.5317,
      "step": 6371
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1311765909194946,
      "learning_rate": 6.763870600728772e-06,
      "loss": 0.5979,
      "step": 6372
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0098353624343872,
      "learning_rate": 6.76291053257859e-06,
      "loss": 0.5055,
      "step": 6373
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1389307975769043,
      "learning_rate": 6.761950390198378e-06,
      "loss": 0.5548,
      "step": 6374
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0219448804855347,
      "learning_rate": 6.760990173628566e-06,
      "loss": 0.5356,
      "step": 6375
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0062955617904663,
      "learning_rate": 6.760029882909582e-06,
      "loss": 0.5336,
      "step": 6376
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9788837432861328,
      "learning_rate": 6.759069518081863e-06,
      "loss": 0.5296,
      "step": 6377
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0798200368881226,
      "learning_rate": 6.758109079185846e-06,
      "loss": 0.526,
      "step": 6378
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0861631631851196,
      "learning_rate": 6.757148566261973e-06,
      "loss": 0.5039,
      "step": 6379
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0507569313049316,
      "learning_rate": 6.756187979350684e-06,
      "loss": 0.5441,
      "step": 6380
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0213924646377563,
      "learning_rate": 6.75522731849243e-06,
      "loss": 0.5038,
      "step": 6381
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9637886881828308,
      "learning_rate": 6.754266583727659e-06,
      "loss": 0.5594,
      "step": 6382
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9614035487174988,
      "learning_rate": 6.753305775096826e-06,
      "loss": 0.5076,
      "step": 6383
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0359253883361816,
      "learning_rate": 6.752344892640384e-06,
      "loss": 0.49,
      "step": 6384
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9942283034324646,
      "learning_rate": 6.751383936398796e-06,
      "loss": 0.5262,
      "step": 6385
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0773903131484985,
      "learning_rate": 6.750422906412523e-06,
      "loss": 0.5882,
      "step": 6386
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0718711614608765,
      "learning_rate": 6.749461802722032e-06,
      "loss": 0.5418,
      "step": 6387
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0906800031661987,
      "learning_rate": 6.7485006253677875e-06,
      "loss": 0.5531,
      "step": 6388
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1717183589935303,
      "learning_rate": 6.747539374390266e-06,
      "loss": 0.6119,
      "step": 6389
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0094630718231201,
      "learning_rate": 6.746578049829942e-06,
      "loss": 0.5017,
      "step": 6390
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9954185485839844,
      "learning_rate": 6.745616651727289e-06,
      "loss": 0.5181,
      "step": 6391
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.052478551864624,
      "learning_rate": 6.744655180122793e-06,
      "loss": 0.5333,
      "step": 6392
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.080421805381775,
      "learning_rate": 6.743693635056936e-06,
      "loss": 0.572,
      "step": 6393
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0238027572631836,
      "learning_rate": 6.742732016570207e-06,
      "loss": 0.5601,
      "step": 6394
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0315519571304321,
      "learning_rate": 6.741770324703095e-06,
      "loss": 0.5492,
      "step": 6395
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9174897074699402,
      "learning_rate": 6.740808559496093e-06,
      "loss": 0.5001,
      "step": 6396
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0448737144470215,
      "learning_rate": 6.739846720989699e-06,
      "loss": 0.5869,
      "step": 6397
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9965710639953613,
      "learning_rate": 6.738884809224413e-06,
      "loss": 0.5069,
      "step": 6398
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.11588454246521,
      "learning_rate": 6.7379228242407345e-06,
      "loss": 0.5498,
      "step": 6399
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1629881858825684,
      "learning_rate": 6.736960766079173e-06,
      "loss": 0.5415,
      "step": 6400
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.061481237411499,
      "learning_rate": 6.735998634780238e-06,
      "loss": 0.5179,
      "step": 6401
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0699317455291748,
      "learning_rate": 6.735036430384436e-06,
      "loss": 0.5091,
      "step": 6402
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0474531650543213,
      "learning_rate": 6.7340741529322875e-06,
      "loss": 0.5528,
      "step": 6403
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.034765601158142,
      "learning_rate": 6.733111802464308e-06,
      "loss": 0.5094,
      "step": 6404
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9787458777427673,
      "learning_rate": 6.732149379021022e-06,
      "loss": 0.5054,
      "step": 6405
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9702959656715393,
      "learning_rate": 6.7311868826429485e-06,
      "loss": 0.5422,
      "step": 6406
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0975489616394043,
      "learning_rate": 6.730224313370619e-06,
      "loss": 0.546,
      "step": 6407
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0134038925170898,
      "learning_rate": 6.729261671244563e-06,
      "loss": 0.5146,
      "step": 6408
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0336501598358154,
      "learning_rate": 6.728298956305313e-06,
      "loss": 0.4965,
      "step": 6409
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0559171438217163,
      "learning_rate": 6.727336168593406e-06,
      "loss": 0.5324,
      "step": 6410
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0390537977218628,
      "learning_rate": 6.726373308149382e-06,
      "loss": 0.5387,
      "step": 6411
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9847078323364258,
      "learning_rate": 6.725410375013783e-06,
      "loss": 0.4994,
      "step": 6412
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.972402036190033,
      "learning_rate": 6.724447369227159e-06,
      "loss": 0.5271,
      "step": 6413
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0684046745300293,
      "learning_rate": 6.723484290830051e-06,
      "loss": 0.5295,
      "step": 6414
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1181683540344238,
      "learning_rate": 6.722521139863017e-06,
      "loss": 0.556,
      "step": 6415
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.022675633430481,
      "learning_rate": 6.72155791636661e-06,
      "loss": 0.5411,
      "step": 6416
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9986287355422974,
      "learning_rate": 6.720594620381387e-06,
      "loss": 0.4959,
      "step": 6417
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.113417148590088,
      "learning_rate": 6.71963125194791e-06,
      "loss": 0.5561,
      "step": 6418
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1068001985549927,
      "learning_rate": 6.718667811106744e-06,
      "loss": 0.5329,
      "step": 6419
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.092809796333313,
      "learning_rate": 6.717704297898455e-06,
      "loss": 0.5233,
      "step": 6420
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0115886926651,
      "learning_rate": 6.716740712363614e-06,
      "loss": 0.5103,
      "step": 6421
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1021331548690796,
      "learning_rate": 6.715777054542793e-06,
      "loss": 0.5401,
      "step": 6422
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0838817358016968,
      "learning_rate": 6.714813324476569e-06,
      "loss": 0.5178,
      "step": 6423
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.07334566116333,
      "learning_rate": 6.713849522205522e-06,
      "loss": 0.5002,
      "step": 6424
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0014665126800537,
      "learning_rate": 6.712885647770233e-06,
      "loss": 0.5331,
      "step": 6425
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0466426610946655,
      "learning_rate": 6.711921701211288e-06,
      "loss": 0.5419,
      "step": 6426
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0850886106491089,
      "learning_rate": 6.710957682569276e-06,
      "loss": 0.585,
      "step": 6427
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9901554584503174,
      "learning_rate": 6.709993591884788e-06,
      "loss": 0.5246,
      "step": 6428
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9305109977722168,
      "learning_rate": 6.709029429198418e-06,
      "loss": 0.5141,
      "step": 6429
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9888066649436951,
      "learning_rate": 6.7080651945507645e-06,
      "loss": 0.518,
      "step": 6430
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9896424412727356,
      "learning_rate": 6.707100887982427e-06,
      "loss": 0.5147,
      "step": 6431
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0180577039718628,
      "learning_rate": 6.7061365095340105e-06,
      "loss": 0.5327,
      "step": 6432
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.014597773551941,
      "learning_rate": 6.70517205924612e-06,
      "loss": 0.5363,
      "step": 6433
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9559026956558228,
      "learning_rate": 6.7042075371593665e-06,
      "loss": 0.5206,
      "step": 6434
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.031485676765442,
      "learning_rate": 6.703242943314362e-06,
      "loss": 0.5737,
      "step": 6435
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.142580509185791,
      "learning_rate": 6.702278277751722e-06,
      "loss": 0.4991,
      "step": 6436
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1274057626724243,
      "learning_rate": 6.701313540512065e-06,
      "loss": 0.5713,
      "step": 6437
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0787616968154907,
      "learning_rate": 6.700348731636014e-06,
      "loss": 0.5213,
      "step": 6438
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0522346496582031,
      "learning_rate": 6.699383851164194e-06,
      "loss": 0.5902,
      "step": 6439
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0166659355163574,
      "learning_rate": 6.6984188991372305e-06,
      "loss": 0.5382,
      "step": 6440
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.933194100856781,
      "learning_rate": 6.697453875595755e-06,
      "loss": 0.5033,
      "step": 6441
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8746803998947144,
      "learning_rate": 6.696488780580403e-06,
      "loss": 0.49,
      "step": 6442
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9358901381492615,
      "learning_rate": 6.69552361413181e-06,
      "loss": 0.5203,
      "step": 6443
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9617918729782104,
      "learning_rate": 6.694558376290615e-06,
      "loss": 0.5361,
      "step": 6444
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9649971127510071,
      "learning_rate": 6.693593067097462e-06,
      "loss": 0.5358,
      "step": 6445
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0535650253295898,
      "learning_rate": 6.692627686592998e-06,
      "loss": 0.5792,
      "step": 6446
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.090075969696045,
      "learning_rate": 6.691662234817869e-06,
      "loss": 0.542,
      "step": 6447
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9936316609382629,
      "learning_rate": 6.690696711812729e-06,
      "loss": 0.4804,
      "step": 6448
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9206106662750244,
      "learning_rate": 6.68973111761823e-06,
      "loss": 0.5067,
      "step": 6449
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0001904964447021,
      "learning_rate": 6.688765452275033e-06,
      "loss": 0.5969,
      "step": 6450
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9626813530921936,
      "learning_rate": 6.687799715823798e-06,
      "loss": 0.486,
      "step": 6451
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9665706753730774,
      "learning_rate": 6.686833908305188e-06,
      "loss": 0.548,
      "step": 6452
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9423041343688965,
      "learning_rate": 6.68586802975987e-06,
      "loss": 0.5228,
      "step": 6453
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0378499031066895,
      "learning_rate": 6.684902080228514e-06,
      "loss": 0.5171,
      "step": 6454
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9767656922340393,
      "learning_rate": 6.6839360597517935e-06,
      "loss": 0.4995,
      "step": 6455
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0101171731948853,
      "learning_rate": 6.682969968370383e-06,
      "loss": 0.5397,
      "step": 6456
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.02167546749115,
      "learning_rate": 6.68200380612496e-06,
      "loss": 0.5212,
      "step": 6457
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0230549573898315,
      "learning_rate": 6.681037573056211e-06,
      "loss": 0.5128,
      "step": 6458
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.034674882888794,
      "learning_rate": 6.6800712692048164e-06,
      "loss": 0.5311,
      "step": 6459
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0055681467056274,
      "learning_rate": 6.679104894611466e-06,
      "loss": 0.559,
      "step": 6460
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0654057264328003,
      "learning_rate": 6.678138449316848e-06,
      "loss": 0.5788,
      "step": 6461
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0350961685180664,
      "learning_rate": 6.6771719333616584e-06,
      "loss": 0.5117,
      "step": 6462
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9289602041244507,
      "learning_rate": 6.676205346786594e-06,
      "loss": 0.527,
      "step": 6463
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9366391897201538,
      "learning_rate": 6.6752386896323526e-06,
      "loss": 0.5414,
      "step": 6464
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.154829502105713,
      "learning_rate": 6.674271961939638e-06,
      "loss": 0.5883,
      "step": 6465
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9293048977851868,
      "learning_rate": 6.673305163749155e-06,
      "loss": 0.4986,
      "step": 6466
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9967445731163025,
      "learning_rate": 6.672338295101614e-06,
      "loss": 0.5039,
      "step": 6467
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9883416295051575,
      "learning_rate": 6.671371356037723e-06,
      "loss": 0.5462,
      "step": 6468
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0531284809112549,
      "learning_rate": 6.670404346598199e-06,
      "loss": 0.5444,
      "step": 6469
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0140522718429565,
      "learning_rate": 6.669437266823759e-06,
      "loss": 0.5388,
      "step": 6470
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0097755193710327,
      "learning_rate": 6.668470116755125e-06,
      "loss": 0.5297,
      "step": 6471
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9846504330635071,
      "learning_rate": 6.6675028964330156e-06,
      "loss": 0.5286,
      "step": 6472
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0103787183761597,
      "learning_rate": 6.666535605898162e-06,
      "loss": 0.5903,
      "step": 6473
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0471034049987793,
      "learning_rate": 6.6655682451912915e-06,
      "loss": 0.5307,
      "step": 6474
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0226188898086548,
      "learning_rate": 6.664600814353137e-06,
      "loss": 0.5333,
      "step": 6475
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0311198234558105,
      "learning_rate": 6.6636333134244305e-06,
      "loss": 0.5282,
      "step": 6476
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0073610544204712,
      "learning_rate": 6.662665742445914e-06,
      "loss": 0.5265,
      "step": 6477
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0142841339111328,
      "learning_rate": 6.661698101458327e-06,
      "loss": 0.5301,
      "step": 6478
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9951909184455872,
      "learning_rate": 6.660730390502414e-06,
      "loss": 0.5186,
      "step": 6479
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9450570344924927,
      "learning_rate": 6.6597626096189206e-06,
      "loss": 0.5458,
      "step": 6480
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9907074570655823,
      "learning_rate": 6.658794758848598e-06,
      "loss": 0.5399,
      "step": 6481
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0250133275985718,
      "learning_rate": 6.6578268382322e-06,
      "loss": 0.5148,
      "step": 6482
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9800702929496765,
      "learning_rate": 6.656858847810479e-06,
      "loss": 0.5195,
      "step": 6483
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9772721529006958,
      "learning_rate": 6.655890787624195e-06,
      "loss": 0.5012,
      "step": 6484
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9839169383049011,
      "learning_rate": 6.654922657714112e-06,
      "loss": 0.4937,
      "step": 6485
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.044286847114563,
      "learning_rate": 6.6539544581209935e-06,
      "loss": 0.566,
      "step": 6486
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9918443560600281,
      "learning_rate": 6.652986188885605e-06,
      "loss": 0.5345,
      "step": 6487
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0191620588302612,
      "learning_rate": 6.652017850048719e-06,
      "loss": 0.5222,
      "step": 6488
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0628665685653687,
      "learning_rate": 6.651049441651107e-06,
      "loss": 0.5547,
      "step": 6489
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9881481528282166,
      "learning_rate": 6.65008096373355e-06,
      "loss": 0.5246,
      "step": 6490
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.997631847858429,
      "learning_rate": 6.6491124163368215e-06,
      "loss": 0.533,
      "step": 6491
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9483758211135864,
      "learning_rate": 6.648143799501705e-06,
      "loss": 0.5201,
      "step": 6492
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9615177512168884,
      "learning_rate": 6.647175113268989e-06,
      "loss": 0.534,
      "step": 6493
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0250009298324585,
      "learning_rate": 6.646206357679458e-06,
      "loss": 0.552,
      "step": 6494
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0113991498947144,
      "learning_rate": 6.645237532773902e-06,
      "loss": 0.573,
      "step": 6495
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.03082275390625,
      "learning_rate": 6.64426863859312e-06,
      "loss": 0.5649,
      "step": 6496
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0717030763626099,
      "learning_rate": 6.643299675177906e-06,
      "loss": 0.5586,
      "step": 6497
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.072690725326538,
      "learning_rate": 6.642330642569056e-06,
      "loss": 0.5275,
      "step": 6498
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9866908192634583,
      "learning_rate": 6.641361540807377e-06,
      "loss": 0.5133,
      "step": 6499
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0255568027496338,
      "learning_rate": 6.640392369933675e-06,
      "loss": 0.5299,
      "step": 6500
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0390933752059937,
      "learning_rate": 6.639423129988756e-06,
      "loss": 0.5288,
      "step": 6501
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.038448691368103,
      "learning_rate": 6.638453821013431e-06,
      "loss": 0.6081,
      "step": 6502
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9224210381507874,
      "learning_rate": 6.637484443048516e-06,
      "loss": 0.5248,
      "step": 6503
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.966993510723114,
      "learning_rate": 6.636514996134828e-06,
      "loss": 0.4875,
      "step": 6504
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0495386123657227,
      "learning_rate": 6.635545480313187e-06,
      "loss": 0.5672,
      "step": 6505
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0295976400375366,
      "learning_rate": 6.634575895624414e-06,
      "loss": 0.5866,
      "step": 6506
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0618761777877808,
      "learning_rate": 6.6336062421093374e-06,
      "loss": 0.562,
      "step": 6507
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0179448127746582,
      "learning_rate": 6.632636519808785e-06,
      "loss": 0.5375,
      "step": 6508
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.019084095954895,
      "learning_rate": 6.6316667287635875e-06,
      "loss": 0.5497,
      "step": 6509
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.040083885192871,
      "learning_rate": 6.63069686901458e-06,
      "loss": 0.5661,
      "step": 6510
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0258654356002808,
      "learning_rate": 6.629726940602601e-06,
      "loss": 0.5391,
      "step": 6511
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.982742965221405,
      "learning_rate": 6.62875694356849e-06,
      "loss": 0.5586,
      "step": 6512
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.039351224899292,
      "learning_rate": 6.62778687795309e-06,
      "loss": 0.5062,
      "step": 6513
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0537571907043457,
      "learning_rate": 6.626816743797246e-06,
      "loss": 0.4956,
      "step": 6514
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.928530216217041,
      "learning_rate": 6.62584654114181e-06,
      "loss": 0.4869,
      "step": 6515
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0642585754394531,
      "learning_rate": 6.6248762700276315e-06,
      "loss": 0.5522,
      "step": 6516
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9538542628288269,
      "learning_rate": 6.623905930495565e-06,
      "loss": 0.5304,
      "step": 6517
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0592854022979736,
      "learning_rate": 6.622935522586469e-06,
      "loss": 0.4979,
      "step": 6518
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9506630897521973,
      "learning_rate": 6.6219650463412034e-06,
      "loss": 0.5171,
      "step": 6519
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0356582403182983,
      "learning_rate": 6.620994501800634e-06,
      "loss": 0.5231,
      "step": 6520
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0392316579818726,
      "learning_rate": 6.620023889005624e-06,
      "loss": 0.5735,
      "step": 6521
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0876318216323853,
      "learning_rate": 6.619053207997043e-06,
      "loss": 0.5476,
      "step": 6522
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0757724046707153,
      "learning_rate": 6.618082458815765e-06,
      "loss": 0.52,
      "step": 6523
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0079342126846313,
      "learning_rate": 6.617111641502664e-06,
      "loss": 0.5142,
      "step": 6524
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9872135519981384,
      "learning_rate": 6.616140756098617e-06,
      "loss": 0.4967,
      "step": 6525
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1018986701965332,
      "learning_rate": 6.615169802644503e-06,
      "loss": 0.5744,
      "step": 6526
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1604558229446411,
      "learning_rate": 6.614198781181209e-06,
      "loss": 0.5501,
      "step": 6527
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.044912576675415,
      "learning_rate": 6.613227691749619e-06,
      "loss": 0.5811,
      "step": 6528
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0152037143707275,
      "learning_rate": 6.612256534390624e-06,
      "loss": 0.5524,
      "step": 6529
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0281046628952026,
      "learning_rate": 6.611285309145113e-06,
      "loss": 0.5104,
      "step": 6530
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9736826419830322,
      "learning_rate": 6.610314016053986e-06,
      "loss": 0.5092,
      "step": 6531
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9732024669647217,
      "learning_rate": 6.609342655158135e-06,
      "loss": 0.5532,
      "step": 6532
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0911118984222412,
      "learning_rate": 6.608371226498464e-06,
      "loss": 0.5026,
      "step": 6533
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.047197937965393,
      "learning_rate": 6.607399730115875e-06,
      "loss": 0.5424,
      "step": 6534
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0095125436782837,
      "learning_rate": 6.6064281660512775e-06,
      "loss": 0.4706,
      "step": 6535
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0289307832717896,
      "learning_rate": 6.6054565343455765e-06,
      "loss": 0.5548,
      "step": 6536
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9875046610832214,
      "learning_rate": 6.604484835039686e-06,
      "loss": 0.5222,
      "step": 6537
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0515408515930176,
      "learning_rate": 6.603513068174521e-06,
      "loss": 0.4744,
      "step": 6538
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9239807724952698,
      "learning_rate": 6.602541233790999e-06,
      "loss": 0.5288,
      "step": 6539
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.121701955795288,
      "learning_rate": 6.601569331930041e-06,
      "loss": 0.5923,
      "step": 6540
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9341360926628113,
      "learning_rate": 6.600597362632568e-06,
      "loss": 0.5167,
      "step": 6541
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.015208125114441,
      "learning_rate": 6.599625325939509e-06,
      "loss": 0.5704,
      "step": 6542
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0524418354034424,
      "learning_rate": 6.598653221891793e-06,
      "loss": 0.558,
      "step": 6543
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9874768853187561,
      "learning_rate": 6.597681050530351e-06,
      "loss": 0.5702,
      "step": 6544
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0313066244125366,
      "learning_rate": 6.596708811896116e-06,
      "loss": 0.5523,
      "step": 6545
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.067104458808899,
      "learning_rate": 6.595736506030029e-06,
      "loss": 0.5437,
      "step": 6546
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9491031765937805,
      "learning_rate": 6.59476413297303e-06,
      "loss": 0.5008,
      "step": 6547
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0489870309829712,
      "learning_rate": 6.59379169276606e-06,
      "loss": 0.5247,
      "step": 6548
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9969553351402283,
      "learning_rate": 6.5928191854500644e-06,
      "loss": 0.5564,
      "step": 6549
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0018967390060425,
      "learning_rate": 6.591846611065997e-06,
      "loss": 0.5337,
      "step": 6550
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9852742552757263,
      "learning_rate": 6.590873969654805e-06,
      "loss": 0.5623,
      "step": 6551
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.039914846420288,
      "learning_rate": 6.589901261257445e-06,
      "loss": 0.5445,
      "step": 6552
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0345009565353394,
      "learning_rate": 6.588928485914871e-06,
      "loss": 0.5891,
      "step": 6553
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9844979643821716,
      "learning_rate": 6.587955643668049e-06,
      "loss": 0.5386,
      "step": 6554
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.042885661125183,
      "learning_rate": 6.58698273455794e-06,
      "loss": 0.4922,
      "step": 6555
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0444130897521973,
      "learning_rate": 6.586009758625507e-06,
      "loss": 0.5699,
      "step": 6556
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0053129196166992,
      "learning_rate": 6.585036715911719e-06,
      "loss": 0.5309,
      "step": 6557
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0225436687469482,
      "learning_rate": 6.58406360645755e-06,
      "loss": 0.5656,
      "step": 6558
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.930986225605011,
      "learning_rate": 6.583090430303975e-06,
      "loss": 0.5037,
      "step": 6559
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.007002353668213,
      "learning_rate": 6.582117187491967e-06,
      "loss": 0.5977,
      "step": 6560
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0585647821426392,
      "learning_rate": 6.581143878062507e-06,
      "loss": 0.5395,
      "step": 6561
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0320481061935425,
      "learning_rate": 6.58017050205658e-06,
      "loss": 0.4973,
      "step": 6562
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0310266017913818,
      "learning_rate": 6.5791970595151714e-06,
      "loss": 0.5637,
      "step": 6563
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9598348140716553,
      "learning_rate": 6.578223550479266e-06,
      "loss": 0.5487,
      "step": 6564
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8893693685531616,
      "learning_rate": 6.5772499749898585e-06,
      "loss": 0.4807,
      "step": 6565
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0530903339385986,
      "learning_rate": 6.576276333087941e-06,
      "loss": 0.5382,
      "step": 6566
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0158112049102783,
      "learning_rate": 6.575302624814512e-06,
      "loss": 0.5204,
      "step": 6567
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.030490756034851,
      "learning_rate": 6.5743288502105675e-06,
      "loss": 0.5139,
      "step": 6568
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9897223114967346,
      "learning_rate": 6.5733550093171115e-06,
      "loss": 0.5287,
      "step": 6569
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.047285795211792,
      "learning_rate": 6.572381102175151e-06,
      "loss": 0.526,
      "step": 6570
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0202997922897339,
      "learning_rate": 6.571407128825692e-06,
      "loss": 0.5615,
      "step": 6571
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0218029022216797,
      "learning_rate": 6.570433089309745e-06,
      "loss": 0.4885,
      "step": 6572
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0741409063339233,
      "learning_rate": 6.569458983668323e-06,
      "loss": 0.4896,
      "step": 6573
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.016112208366394,
      "learning_rate": 6.5684848119424435e-06,
      "loss": 0.567,
      "step": 6574
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.040443778038025,
      "learning_rate": 6.567510574173126e-06,
      "loss": 0.5531,
      "step": 6575
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.054648518562317,
      "learning_rate": 6.566536270401389e-06,
      "loss": 0.5775,
      "step": 6576
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0508983135223389,
      "learning_rate": 6.5655619006682604e-06,
      "loss": 0.5625,
      "step": 6577
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1195893287658691,
      "learning_rate": 6.5645874650147676e-06,
      "loss": 0.5335,
      "step": 6578
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0068633556365967,
      "learning_rate": 6.563612963481938e-06,
      "loss": 0.5265,
      "step": 6579
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0301257371902466,
      "learning_rate": 6.562638396110805e-06,
      "loss": 0.5061,
      "step": 6580
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1142112016677856,
      "learning_rate": 6.561663762942407e-06,
      "loss": 0.5522,
      "step": 6581
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1140917539596558,
      "learning_rate": 6.560689064017781e-06,
      "loss": 0.4971,
      "step": 6582
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0406494140625,
      "learning_rate": 6.559714299377966e-06,
      "loss": 0.5766,
      "step": 6583
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9920594096183777,
      "learning_rate": 6.558739469064008e-06,
      "loss": 0.6039,
      "step": 6584
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0681092739105225,
      "learning_rate": 6.5577645731169535e-06,
      "loss": 0.5815,
      "step": 6585
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2725310325622559,
      "learning_rate": 6.556789611577854e-06,
      "loss": 0.5283,
      "step": 6586
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0315043926239014,
      "learning_rate": 6.555814584487757e-06,
      "loss": 0.5367,
      "step": 6587
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.012886881828308,
      "learning_rate": 6.5548394918877216e-06,
      "loss": 0.5401,
      "step": 6588
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9161368608474731,
      "learning_rate": 6.553864333818803e-06,
      "loss": 0.508,
      "step": 6589
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9886636137962341,
      "learning_rate": 6.552889110322062e-06,
      "loss": 0.5091,
      "step": 6590
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0155935287475586,
      "learning_rate": 6.551913821438565e-06,
      "loss": 0.4738,
      "step": 6591
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0011165142059326,
      "learning_rate": 6.550938467209375e-06,
      "loss": 0.5253,
      "step": 6592
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9297583103179932,
      "learning_rate": 6.5499630476755616e-06,
      "loss": 0.4809,
      "step": 6593
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0220762491226196,
      "learning_rate": 6.548987562878195e-06,
      "loss": 0.5204,
      "step": 6594
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0843340158462524,
      "learning_rate": 6.548012012858352e-06,
      "loss": 0.5195,
      "step": 6595
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0520986318588257,
      "learning_rate": 6.547036397657106e-06,
      "loss": 0.5212,
      "step": 6596
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0433666706085205,
      "learning_rate": 6.546060717315542e-06,
      "loss": 0.5474,
      "step": 6597
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0284672975540161,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.5432,
      "step": 6598
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0332967042922974,
      "learning_rate": 6.5441091613757805e-06,
      "loss": 0.5188,
      "step": 6599
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9910998344421387,
      "learning_rate": 6.543133285859758e-06,
      "loss": 0.5183,
      "step": 6600
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9519909620285034,
      "learning_rate": 6.542157345367763e-06,
      "loss": 0.4884,
      "step": 6601
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.069887399673462,
      "learning_rate": 6.5411813399408845e-06,
      "loss": 0.5841,
      "step": 6602
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.022975206375122,
      "learning_rate": 6.540205269620221e-06,
      "loss": 0.5542,
      "step": 6603
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9461351037025452,
      "learning_rate": 6.539229134446874e-06,
      "loss": 0.5331,
      "step": 6604
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0331028699874878,
      "learning_rate": 6.538252934461941e-06,
      "loss": 0.5288,
      "step": 6605
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0788092613220215,
      "learning_rate": 6.537276669706527e-06,
      "loss": 0.5742,
      "step": 6606
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0447759628295898,
      "learning_rate": 6.536300340221742e-06,
      "loss": 0.573,
      "step": 6607
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9823431968688965,
      "learning_rate": 6.535323946048695e-06,
      "loss": 0.5166,
      "step": 6608
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9577629566192627,
      "learning_rate": 6.534347487228495e-06,
      "loss": 0.4816,
      "step": 6609
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9827210307121277,
      "learning_rate": 6.533370963802261e-06,
      "loss": 0.497,
      "step": 6610
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1767102479934692,
      "learning_rate": 6.532394375811111e-06,
      "loss": 0.5633,
      "step": 6611
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1166398525238037,
      "learning_rate": 6.531417723296164e-06,
      "loss": 0.5354,
      "step": 6612
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1115363836288452,
      "learning_rate": 6.530441006298544e-06,
      "loss": 0.5402,
      "step": 6613
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0333431959152222,
      "learning_rate": 6.5294642248593765e-06,
      "loss": 0.5907,
      "step": 6614
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0875381231307983,
      "learning_rate": 6.528487379019791e-06,
      "loss": 0.529,
      "step": 6615
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9464609026908875,
      "learning_rate": 6.5275104688209215e-06,
      "loss": 0.4968,
      "step": 6616
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.094282627105713,
      "learning_rate": 6.526533494303898e-06,
      "loss": 0.582,
      "step": 6617
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.061200737953186,
      "learning_rate": 6.525556455509858e-06,
      "loss": 0.5653,
      "step": 6618
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9713898301124573,
      "learning_rate": 6.5245793524799465e-06,
      "loss": 0.5051,
      "step": 6619
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0049147605895996,
      "learning_rate": 6.5236021852553e-06,
      "loss": 0.5664,
      "step": 6620
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.095810890197754,
      "learning_rate": 6.522624953877066e-06,
      "loss": 0.5445,
      "step": 6621
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9827590584754944,
      "learning_rate": 6.52164765838639e-06,
      "loss": 0.4835,
      "step": 6622
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.028319239616394,
      "learning_rate": 6.520670298824428e-06,
      "loss": 0.5277,
      "step": 6623
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0088140964508057,
      "learning_rate": 6.519692875232328e-06,
      "loss": 0.4855,
      "step": 6624
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0765272378921509,
      "learning_rate": 6.518715387651249e-06,
      "loss": 0.5564,
      "step": 6625
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0901261568069458,
      "learning_rate": 6.517737836122345e-06,
      "loss": 0.5494,
      "step": 6626
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9907105565071106,
      "learning_rate": 6.516760220686783e-06,
      "loss": 0.5482,
      "step": 6627
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0758845806121826,
      "learning_rate": 6.515782541385725e-06,
      "loss": 0.595,
      "step": 6628
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0081164836883545,
      "learning_rate": 6.514804798260337e-06,
      "loss": 0.5667,
      "step": 6629
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9572669863700867,
      "learning_rate": 6.513826991351786e-06,
      "loss": 0.558,
      "step": 6630
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9977998733520508,
      "learning_rate": 6.512849120701249e-06,
      "loss": 0.5219,
      "step": 6631
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9644599556922913,
      "learning_rate": 6.511871186349897e-06,
      "loss": 0.5111,
      "step": 6632
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0376458168029785,
      "learning_rate": 6.510893188338911e-06,
      "loss": 0.5563,
      "step": 6633
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.039550542831421,
      "learning_rate": 6.509915126709467e-06,
      "loss": 0.5022,
      "step": 6634
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0308173894882202,
      "learning_rate": 6.50893700150275e-06,
      "loss": 0.5156,
      "step": 6635
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.015334963798523,
      "learning_rate": 6.5079588127599455e-06,
      "loss": 0.5134,
      "step": 6636
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9615833163261414,
      "learning_rate": 6.50698056052224e-06,
      "loss": 0.5199,
      "step": 6637
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9502135515213013,
      "learning_rate": 6.506002244830827e-06,
      "loss": 0.5224,
      "step": 6638
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.999001681804657,
      "learning_rate": 6.505023865726898e-06,
      "loss": 0.5313,
      "step": 6639
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9371646046638489,
      "learning_rate": 6.50404542325165e-06,
      "loss": 0.5288,
      "step": 6640
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0646824836730957,
      "learning_rate": 6.503066917446279e-06,
      "loss": 0.5263,
      "step": 6641
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9830496311187744,
      "learning_rate": 6.502088348351992e-06,
      "loss": 0.5154,
      "step": 6642
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0101431608200073,
      "learning_rate": 6.501109716009988e-06,
      "loss": 0.4874,
      "step": 6643
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0016448497772217,
      "learning_rate": 6.500131020461477e-06,
      "loss": 0.5067,
      "step": 6644
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9707490801811218,
      "learning_rate": 6.4991522617476666e-06,
      "loss": 0.4988,
      "step": 6645
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.03791081905365,
      "learning_rate": 6.498173439909771e-06,
      "loss": 0.5594,
      "step": 6646
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0289043188095093,
      "learning_rate": 6.497194554989001e-06,
      "loss": 0.5021,
      "step": 6647
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0819637775421143,
      "learning_rate": 6.496215607026579e-06,
      "loss": 0.4839,
      "step": 6648
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1179145574569702,
      "learning_rate": 6.495236596063722e-06,
      "loss": 0.5231,
      "step": 6649
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.049654483795166,
      "learning_rate": 6.494257522141654e-06,
      "loss": 0.5223,
      "step": 6650
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.221232533454895,
      "learning_rate": 6.4932783853016005e-06,
      "loss": 0.5759,
      "step": 6651
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9750707149505615,
      "learning_rate": 6.492299185584787e-06,
      "loss": 0.4607,
      "step": 6652
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0411572456359863,
      "learning_rate": 6.491319923032446e-06,
      "loss": 0.5288,
      "step": 6653
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9910756945610046,
      "learning_rate": 6.490340597685811e-06,
      "loss": 0.5358,
      "step": 6654
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9476546049118042,
      "learning_rate": 6.48936120958612e-06,
      "loss": 0.512,
      "step": 6655
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9512740969657898,
      "learning_rate": 6.488381758774609e-06,
      "loss": 0.5522,
      "step": 6656
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9980294108390808,
      "learning_rate": 6.487402245292518e-06,
      "loss": 0.5226,
      "step": 6657
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0905872583389282,
      "learning_rate": 6.486422669181094e-06,
      "loss": 0.5611,
      "step": 6658
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9635381698608398,
      "learning_rate": 6.485443030481583e-06,
      "loss": 0.5382,
      "step": 6659
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0294065475463867,
      "learning_rate": 6.4844633292352335e-06,
      "loss": 0.5563,
      "step": 6660
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9263328909873962,
      "learning_rate": 6.483483565483295e-06,
      "loss": 0.4946,
      "step": 6661
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9932078719139099,
      "learning_rate": 6.482503739267026e-06,
      "loss": 0.5044,
      "step": 6662
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.065634846687317,
      "learning_rate": 6.481523850627682e-06,
      "loss": 0.5581,
      "step": 6663
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.054053544998169,
      "learning_rate": 6.4805438996065215e-06,
      "loss": 0.6028,
      "step": 6664
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9770582318305969,
      "learning_rate": 6.479563886244809e-06,
      "loss": 0.5487,
      "step": 6665
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0783045291900635,
      "learning_rate": 6.478583810583807e-06,
      "loss": 0.5277,
      "step": 6666
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0439814329147339,
      "learning_rate": 6.477603672664785e-06,
      "loss": 0.5342,
      "step": 6667
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9825044274330139,
      "learning_rate": 6.476623472529012e-06,
      "loss": 0.5235,
      "step": 6668
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9862465858459473,
      "learning_rate": 6.475643210217762e-06,
      "loss": 0.5693,
      "step": 6669
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1262718439102173,
      "learning_rate": 6.47466288577231e-06,
      "loss": 0.5738,
      "step": 6670
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9301985502243042,
      "learning_rate": 6.473682499233934e-06,
      "loss": 0.4966,
      "step": 6671
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0108258724212646,
      "learning_rate": 6.472702050643913e-06,
      "loss": 0.5076,
      "step": 6672
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0300981998443604,
      "learning_rate": 6.471721540043533e-06,
      "loss": 0.5227,
      "step": 6673
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1556057929992676,
      "learning_rate": 6.47074096747408e-06,
      "loss": 0.5453,
      "step": 6674
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9517860412597656,
      "learning_rate": 6.469760332976839e-06,
      "loss": 0.4722,
      "step": 6675
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0921064615249634,
      "learning_rate": 6.4687796365931035e-06,
      "loss": 0.5293,
      "step": 6676
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0534489154815674,
      "learning_rate": 6.467798878364168e-06,
      "loss": 0.5218,
      "step": 6677
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9986158013343811,
      "learning_rate": 6.466818058331328e-06,
      "loss": 0.5382,
      "step": 6678
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.045770287513733,
      "learning_rate": 6.465837176535881e-06,
      "loss": 0.5067,
      "step": 6679
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9914131164550781,
      "learning_rate": 6.46485623301913e-06,
      "loss": 0.504,
      "step": 6680
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9861920475959778,
      "learning_rate": 6.46387522782238e-06,
      "loss": 0.548,
      "step": 6681
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9626903533935547,
      "learning_rate": 6.462894160986937e-06,
      "loss": 0.532,
      "step": 6682
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0459203720092773,
      "learning_rate": 6.461913032554108e-06,
      "loss": 0.5218,
      "step": 6683
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9998902082443237,
      "learning_rate": 6.460931842565207e-06,
      "loss": 0.5367,
      "step": 6684
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9615074396133423,
      "learning_rate": 6.4599505910615505e-06,
      "loss": 0.5265,
      "step": 6685
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0468496084213257,
      "learning_rate": 6.45896927808445e-06,
      "loss": 0.5384,
      "step": 6686
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9850332140922546,
      "learning_rate": 6.4579879036752315e-06,
      "loss": 0.5649,
      "step": 6687
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.007083535194397,
      "learning_rate": 6.457006467875213e-06,
      "loss": 0.5188,
      "step": 6688
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0500504970550537,
      "learning_rate": 6.456024970725722e-06,
      "loss": 0.502,
      "step": 6689
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0205175876617432,
      "learning_rate": 6.455043412268083e-06,
      "loss": 0.5195,
      "step": 6690
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0550191402435303,
      "learning_rate": 6.4540617925436275e-06,
      "loss": 0.4939,
      "step": 6691
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0376819372177124,
      "learning_rate": 6.453080111593689e-06,
      "loss": 0.5356,
      "step": 6692
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9219244718551636,
      "learning_rate": 6.4520983694596025e-06,
      "loss": 0.4873,
      "step": 6693
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0686750411987305,
      "learning_rate": 6.451116566182704e-06,
      "loss": 0.5296,
      "step": 6694
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0167797803878784,
      "learning_rate": 6.4501347018043356e-06,
      "loss": 0.5431,
      "step": 6695
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0186437368392944,
      "learning_rate": 6.44915277636584e-06,
      "loss": 0.5459,
      "step": 6696
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0310328006744385,
      "learning_rate": 6.4481707899085624e-06,
      "loss": 0.5266,
      "step": 6697
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0500410795211792,
      "learning_rate": 6.447188742473849e-06,
      "loss": 0.5378,
      "step": 6698
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.064555048942566,
      "learning_rate": 6.446206634103053e-06,
      "loss": 0.5527,
      "step": 6699
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9763587713241577,
      "learning_rate": 6.445224464837527e-06,
      "loss": 0.564,
      "step": 6700
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9487929344177246,
      "learning_rate": 6.444242234718626e-06,
      "loss": 0.4782,
      "step": 6701
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0297126770019531,
      "learning_rate": 6.443259943787708e-06,
      "loss": 0.5551,
      "step": 6702
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0050936937332153,
      "learning_rate": 6.442277592086135e-06,
      "loss": 0.5399,
      "step": 6703
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0384033918380737,
      "learning_rate": 6.4412951796552715e-06,
      "loss": 0.5786,
      "step": 6704
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0282918214797974,
      "learning_rate": 6.44031270653648e-06,
      "loss": 0.544,
      "step": 6705
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.083512306213379,
      "learning_rate": 6.4393301727711296e-06,
      "loss": 0.6097,
      "step": 6706
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0070874691009521,
      "learning_rate": 6.438347578400595e-06,
      "loss": 0.5758,
      "step": 6707
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0564383268356323,
      "learning_rate": 6.437364923466247e-06,
      "loss": 0.5719,
      "step": 6708
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0171891450881958,
      "learning_rate": 6.436382208009463e-06,
      "loss": 0.5656,
      "step": 6709
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1269818544387817,
      "learning_rate": 6.43539943207162e-06,
      "loss": 0.54,
      "step": 6710
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9396026134490967,
      "learning_rate": 6.434416595694102e-06,
      "loss": 0.4855,
      "step": 6711
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9992225170135498,
      "learning_rate": 6.433433698918289e-06,
      "loss": 0.5318,
      "step": 6712
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0550096035003662,
      "learning_rate": 6.432450741785571e-06,
      "loss": 0.5177,
      "step": 6713
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1212995052337646,
      "learning_rate": 6.431467724337332e-06,
      "loss": 0.5223,
      "step": 6714
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9880232810974121,
      "learning_rate": 6.430484646614971e-06,
      "loss": 0.6145,
      "step": 6715
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9042444229125977,
      "learning_rate": 6.429501508659877e-06,
      "loss": 0.4942,
      "step": 6716
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.032116174697876,
      "learning_rate": 6.428518310513446e-06,
      "loss": 0.5615,
      "step": 6717
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0590288639068604,
      "learning_rate": 6.427535052217078e-06,
      "loss": 0.5699,
      "step": 6718
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9387075304985046,
      "learning_rate": 6.4265517338121764e-06,
      "loss": 0.5472,
      "step": 6719
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0132668018341064,
      "learning_rate": 6.4255683553401435e-06,
      "loss": 0.5519,
      "step": 6720
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9345173239707947,
      "learning_rate": 6.424584916842387e-06,
      "loss": 0.5051,
      "step": 6721
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9697826504707336,
      "learning_rate": 6.423601418360314e-06,
      "loss": 0.5309,
      "step": 6722
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0445016622543335,
      "learning_rate": 6.4226178599353385e-06,
      "loss": 0.5459,
      "step": 6723
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0668346881866455,
      "learning_rate": 6.421634241608874e-06,
      "loss": 0.5311,
      "step": 6724
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.009395956993103,
      "learning_rate": 6.420650563422337e-06,
      "loss": 0.5064,
      "step": 6725
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0648528337478638,
      "learning_rate": 6.419666825417147e-06,
      "loss": 0.5421,
      "step": 6726
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.070974588394165,
      "learning_rate": 6.4186830276347246e-06,
      "loss": 0.523,
      "step": 6727
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0476374626159668,
      "learning_rate": 6.417699170116497e-06,
      "loss": 0.5602,
      "step": 6728
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0144678354263306,
      "learning_rate": 6.416715252903888e-06,
      "loss": 0.5607,
      "step": 6729
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0093867778778076,
      "learning_rate": 6.415731276038327e-06,
      "loss": 0.5283,
      "step": 6730
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.955757737159729,
      "learning_rate": 6.414747239561249e-06,
      "loss": 0.5502,
      "step": 6731
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0207679271697998,
      "learning_rate": 6.413763143514086e-06,
      "loss": 0.5735,
      "step": 6732
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0862566232681274,
      "learning_rate": 6.412778987938273e-06,
      "loss": 0.5535,
      "step": 6733
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0303869247436523,
      "learning_rate": 6.411794772875253e-06,
      "loss": 0.5531,
      "step": 6734
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9645281434059143,
      "learning_rate": 6.4108104983664665e-06,
      "loss": 0.4851,
      "step": 6735
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0301026105880737,
      "learning_rate": 6.409826164453359e-06,
      "loss": 0.4651,
      "step": 6736
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0346624851226807,
      "learning_rate": 6.408841771177373e-06,
      "loss": 0.582,
      "step": 6737
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9744212627410889,
      "learning_rate": 6.407857318579963e-06,
      "loss": 0.4727,
      "step": 6738
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9972878098487854,
      "learning_rate": 6.4068728067025785e-06,
      "loss": 0.5673,
      "step": 6739
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.027344822883606,
      "learning_rate": 6.405888235586676e-06,
      "loss": 0.5011,
      "step": 6740
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0794671773910522,
      "learning_rate": 6.4049036052737065e-06,
      "loss": 0.5509,
      "step": 6741
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0829273462295532,
      "learning_rate": 6.403918915805138e-06,
      "loss": 0.5341,
      "step": 6742
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0934503078460693,
      "learning_rate": 6.402934167222427e-06,
      "loss": 0.5575,
      "step": 6743
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0320874452590942,
      "learning_rate": 6.4019493595670365e-06,
      "loss": 0.4917,
      "step": 6744
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9831816554069519,
      "learning_rate": 6.400964492880437e-06,
      "loss": 0.5485,
      "step": 6745
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0676835775375366,
      "learning_rate": 6.399979567204096e-06,
      "loss": 0.5663,
      "step": 6746
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0714167356491089,
      "learning_rate": 6.398994582579485e-06,
      "loss": 0.5501,
      "step": 6747
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.889264702796936,
      "learning_rate": 6.39800953904808e-06,
      "loss": 0.5037,
      "step": 6748
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9774421453475952,
      "learning_rate": 6.397024436651356e-06,
      "loss": 0.4996,
      "step": 6749
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0261812210083008,
      "learning_rate": 6.396039275430792e-06,
      "loss": 0.5452,
      "step": 6750
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1411161422729492,
      "learning_rate": 6.395054055427872e-06,
      "loss": 0.5884,
      "step": 6751
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9426940679550171,
      "learning_rate": 6.394068776684078e-06,
      "loss": 0.5264,
      "step": 6752
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9334375262260437,
      "learning_rate": 6.393083439240897e-06,
      "loss": 0.4932,
      "step": 6753
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9701194763183594,
      "learning_rate": 6.39209804313982e-06,
      "loss": 0.5225,
      "step": 6754
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9619608521461487,
      "learning_rate": 6.391112588422337e-06,
      "loss": 0.496,
      "step": 6755
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0969892740249634,
      "learning_rate": 6.390127075129941e-06,
      "loss": 0.5008,
      "step": 6756
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0912171602249146,
      "learning_rate": 6.38914150330413e-06,
      "loss": 0.5472,
      "step": 6757
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0904109477996826,
      "learning_rate": 6.388155872986404e-06,
      "loss": 0.5307,
      "step": 6758
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9923735857009888,
      "learning_rate": 6.3871701842182625e-06,
      "loss": 0.5192,
      "step": 6759
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9856535196304321,
      "learning_rate": 6.386184437041208e-06,
      "loss": 0.5701,
      "step": 6760
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0405540466308594,
      "learning_rate": 6.385198631496752e-06,
      "loss": 0.5626,
      "step": 6761
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9822201728820801,
      "learning_rate": 6.3842127676263995e-06,
      "loss": 0.5156,
      "step": 6762
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0651607513427734,
      "learning_rate": 6.383226845471663e-06,
      "loss": 0.5882,
      "step": 6763
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9878702163696289,
      "learning_rate": 6.382240865074055e-06,
      "loss": 0.5313,
      "step": 6764
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0117194652557373,
      "learning_rate": 6.381254826475093e-06,
      "loss": 0.5126,
      "step": 6765
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.964489758014679,
      "learning_rate": 6.380268729716296e-06,
      "loss": 0.5657,
      "step": 6766
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0411863327026367,
      "learning_rate": 6.379282574839184e-06,
      "loss": 0.5442,
      "step": 6767
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9609441161155701,
      "learning_rate": 6.37829636188528e-06,
      "loss": 0.5049,
      "step": 6768
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0357297658920288,
      "learning_rate": 6.377310090896112e-06,
      "loss": 0.5303,
      "step": 6769
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0132602453231812,
      "learning_rate": 6.376323761913208e-06,
      "loss": 0.5077,
      "step": 6770
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.058708667755127,
      "learning_rate": 6.375337374978097e-06,
      "loss": 0.5762,
      "step": 6771
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0159242153167725,
      "learning_rate": 6.374350930132313e-06,
      "loss": 0.5666,
      "step": 6772
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0299150943756104,
      "learning_rate": 6.373364427417395e-06,
      "loss": 0.5648,
      "step": 6773
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.998892605304718,
      "learning_rate": 6.372377866874876e-06,
      "loss": 0.483,
      "step": 6774
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9861887693405151,
      "learning_rate": 6.371391248546299e-06,
      "loss": 0.5156,
      "step": 6775
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9344152808189392,
      "learning_rate": 6.370404572473209e-06,
      "loss": 0.5683,
      "step": 6776
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0953096151351929,
      "learning_rate": 6.36941783869715e-06,
      "loss": 0.609,
      "step": 6777
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9250005483627319,
      "learning_rate": 6.368431047259668e-06,
      "loss": 0.4682,
      "step": 6778
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.148626685142517,
      "learning_rate": 6.367444198202315e-06,
      "loss": 0.5514,
      "step": 6779
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.086868166923523,
      "learning_rate": 6.366457291566645e-06,
      "loss": 0.575,
      "step": 6780
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0445231199264526,
      "learning_rate": 6.365470327394212e-06,
      "loss": 0.5424,
      "step": 6781
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9501314163208008,
      "learning_rate": 6.3644833057265735e-06,
      "loss": 0.5533,
      "step": 6782
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.951246976852417,
      "learning_rate": 6.363496226605289e-06,
      "loss": 0.528,
      "step": 6783
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1135523319244385,
      "learning_rate": 6.362509090071922e-06,
      "loss": 0.5157,
      "step": 6784
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0375339984893799,
      "learning_rate": 6.361521896168037e-06,
      "loss": 0.5384,
      "step": 6785
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9648322463035583,
      "learning_rate": 6.360534644935201e-06,
      "loss": 0.5344,
      "step": 6786
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0026075839996338,
      "learning_rate": 6.359547336414985e-06,
      "loss": 0.5257,
      "step": 6787
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.063840627670288,
      "learning_rate": 6.358559970648958e-06,
      "loss": 0.5589,
      "step": 6788
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9657617211341858,
      "learning_rate": 6.357572547678701e-06,
      "loss": 0.5318,
      "step": 6789
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0497090816497803,
      "learning_rate": 6.356585067545784e-06,
      "loss": 0.5196,
      "step": 6790
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9832949638366699,
      "learning_rate": 6.355597530291788e-06,
      "loss": 0.5398,
      "step": 6791
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.04899001121521,
      "learning_rate": 6.354609935958298e-06,
      "loss": 0.5216,
      "step": 6792
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1187092065811157,
      "learning_rate": 6.3536222845868934e-06,
      "loss": 0.5698,
      "step": 6793
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0921283960342407,
      "learning_rate": 6.3526345762191656e-06,
      "loss": 0.5633,
      "step": 6794
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9514185190200806,
      "learning_rate": 6.351646810896699e-06,
      "loss": 0.5154,
      "step": 6795
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9936259984970093,
      "learning_rate": 6.350658988661089e-06,
      "loss": 0.519,
      "step": 6796
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.969613254070282,
      "learning_rate": 6.349671109553928e-06,
      "loss": 0.4938,
      "step": 6797
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0424509048461914,
      "learning_rate": 6.348683173616811e-06,
      "loss": 0.5087,
      "step": 6798
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9708645939826965,
      "learning_rate": 6.347695180891337e-06,
      "loss": 0.5396,
      "step": 6799
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9806487560272217,
      "learning_rate": 6.346707131419108e-06,
      "loss": 0.5133,
      "step": 6800
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.017426609992981,
      "learning_rate": 6.345719025241725e-06,
      "loss": 0.5211,
      "step": 6801
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.030496597290039,
      "learning_rate": 6.3447308624007964e-06,
      "loss": 0.5496,
      "step": 6802
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0744706392288208,
      "learning_rate": 6.343742642937929e-06,
      "loss": 0.4844,
      "step": 6803
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0421864986419678,
      "learning_rate": 6.342754366894735e-06,
      "loss": 0.5664,
      "step": 6804
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0138890743255615,
      "learning_rate": 6.341766034312824e-06,
      "loss": 0.5302,
      "step": 6805
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0514960289001465,
      "learning_rate": 6.340777645233811e-06,
      "loss": 0.5616,
      "step": 6806
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.040167212486267,
      "learning_rate": 6.339789199699319e-06,
      "loss": 0.5204,
      "step": 6807
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0476608276367188,
      "learning_rate": 6.338800697750963e-06,
      "loss": 0.5165,
      "step": 6808
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9987275004386902,
      "learning_rate": 6.337812139430368e-06,
      "loss": 0.4884,
      "step": 6809
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1082664728164673,
      "learning_rate": 6.336823524779155e-06,
      "loss": 0.5455,
      "step": 6810
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9753262996673584,
      "learning_rate": 6.335834853838957e-06,
      "loss": 0.4923,
      "step": 6811
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0573155879974365,
      "learning_rate": 6.334846126651399e-06,
      "loss": 0.5637,
      "step": 6812
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.116104006767273,
      "learning_rate": 6.333857343258115e-06,
      "loss": 0.5546,
      "step": 6813
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0665115118026733,
      "learning_rate": 6.3328685037007365e-06,
      "loss": 0.5216,
      "step": 6814
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1459187269210815,
      "learning_rate": 6.331879608020905e-06,
      "loss": 0.536,
      "step": 6815
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.086716651916504,
      "learning_rate": 6.330890656260253e-06,
      "loss": 0.5948,
      "step": 6816
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0453628301620483,
      "learning_rate": 6.329901648460428e-06,
      "loss": 0.5709,
      "step": 6817
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0266021490097046,
      "learning_rate": 6.32891258466307e-06,
      "loss": 0.5359,
      "step": 6818
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9942297339439392,
      "learning_rate": 6.3279234649098265e-06,
      "loss": 0.5165,
      "step": 6819
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.034275770187378,
      "learning_rate": 6.326934289242346e-06,
      "loss": 0.4993,
      "step": 6820
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0777255296707153,
      "learning_rate": 6.325945057702276e-06,
      "loss": 0.5784,
      "step": 6821
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9982571601867676,
      "learning_rate": 6.324955770331274e-06,
      "loss": 0.5562,
      "step": 6822
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9832544922828674,
      "learning_rate": 6.323966427170993e-06,
      "loss": 0.5012,
      "step": 6823
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1062462329864502,
      "learning_rate": 6.322977028263093e-06,
      "loss": 0.5411,
      "step": 6824
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1453473567962646,
      "learning_rate": 6.321987573649232e-06,
      "loss": 0.5131,
      "step": 6825
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9943939447402954,
      "learning_rate": 6.320998063371072e-06,
      "loss": 0.5,
      "step": 6826
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.993712842464447,
      "learning_rate": 6.320008497470281e-06,
      "loss": 0.5696,
      "step": 6827
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0750510692596436,
      "learning_rate": 6.319018875988523e-06,
      "loss": 0.5619,
      "step": 6828
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.065926194190979,
      "learning_rate": 6.318029198967468e-06,
      "loss": 0.5357,
      "step": 6829
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0730857849121094,
      "learning_rate": 6.317039466448789e-06,
      "loss": 0.5626,
      "step": 6830
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9925942420959473,
      "learning_rate": 6.316049678474159e-06,
      "loss": 0.5374,
      "step": 6831
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1061160564422607,
      "learning_rate": 6.315059835085257e-06,
      "loss": 0.5149,
      "step": 6832
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0566545724868774,
      "learning_rate": 6.314069936323759e-06,
      "loss": 0.5216,
      "step": 6833
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9762769341468811,
      "learning_rate": 6.313079982231347e-06,
      "loss": 0.5119,
      "step": 6834
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0013434886932373,
      "learning_rate": 6.312089972849707e-06,
      "loss": 0.546,
      "step": 6835
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9725648760795593,
      "learning_rate": 6.31109990822052e-06,
      "loss": 0.5382,
      "step": 6836
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9421480894088745,
      "learning_rate": 6.3101097883854765e-06,
      "loss": 0.5517,
      "step": 6837
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1255269050598145,
      "learning_rate": 6.30911961338627e-06,
      "loss": 0.5163,
      "step": 6838
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0309721231460571,
      "learning_rate": 6.3081293832645896e-06,
      "loss": 0.5258,
      "step": 6839
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0831512212753296,
      "learning_rate": 6.30713909806213e-06,
      "loss": 0.486,
      "step": 6840
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9742302298545837,
      "learning_rate": 6.306148757820591e-06,
      "loss": 0.5228,
      "step": 6841
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.971137523651123,
      "learning_rate": 6.3051583625816725e-06,
      "loss": 0.4978,
      "step": 6842
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9712395668029785,
      "learning_rate": 6.304167912387076e-06,
      "loss": 0.4797,
      "step": 6843
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0512776374816895,
      "learning_rate": 6.303177407278504e-06,
      "loss": 0.5436,
      "step": 6844
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9775992631912231,
      "learning_rate": 6.302186847297666e-06,
      "loss": 0.4938,
      "step": 6845
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0109450817108154,
      "learning_rate": 6.301196232486269e-06,
      "loss": 0.5052,
      "step": 6846
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.024448275566101,
      "learning_rate": 6.300205562886026e-06,
      "loss": 0.4826,
      "step": 6847
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0678684711456299,
      "learning_rate": 6.29921483853865e-06,
      "loss": 0.5418,
      "step": 6848
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9749902486801147,
      "learning_rate": 6.298224059485856e-06,
      "loss": 0.5008,
      "step": 6849
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9713776111602783,
      "learning_rate": 6.297233225769363e-06,
      "loss": 0.5541,
      "step": 6850
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0189266204833984,
      "learning_rate": 6.296242337430892e-06,
      "loss": 0.4715,
      "step": 6851
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.994621753692627,
      "learning_rate": 6.2952513945121654e-06,
      "loss": 0.5534,
      "step": 6852
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.974395215511322,
      "learning_rate": 6.2942603970549075e-06,
      "loss": 0.514,
      "step": 6853
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9735104441642761,
      "learning_rate": 6.293269345100849e-06,
      "loss": 0.5116,
      "step": 6854
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9252398610115051,
      "learning_rate": 6.292278238691715e-06,
      "loss": 0.5395,
      "step": 6855
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9296538829803467,
      "learning_rate": 6.29128707786924e-06,
      "loss": 0.5244,
      "step": 6856
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0363876819610596,
      "learning_rate": 6.29029586267516e-06,
      "loss": 0.5,
      "step": 6857
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0339961051940918,
      "learning_rate": 6.289304593151209e-06,
      "loss": 0.5435,
      "step": 6858
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9722663760185242,
      "learning_rate": 6.288313269339126e-06,
      "loss": 0.5263,
      "step": 6859
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9953331351280212,
      "learning_rate": 6.287321891280653e-06,
      "loss": 0.5067,
      "step": 6860
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.011223316192627,
      "learning_rate": 6.2863304590175335e-06,
      "loss": 0.522,
      "step": 6861
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9478493928909302,
      "learning_rate": 6.2853389725915146e-06,
      "loss": 0.5295,
      "step": 6862
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.960415780544281,
      "learning_rate": 6.284347432044342e-06,
      "loss": 0.538,
      "step": 6863
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0387394428253174,
      "learning_rate": 6.2833558374177664e-06,
      "loss": 0.5375,
      "step": 6864
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9971448183059692,
      "learning_rate": 6.282364188753541e-06,
      "loss": 0.5221,
      "step": 6865
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9863330721855164,
      "learning_rate": 6.281372486093422e-06,
      "loss": 0.557,
      "step": 6866
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0580681562423706,
      "learning_rate": 6.280380729479164e-06,
      "loss": 0.5088,
      "step": 6867
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0422565937042236,
      "learning_rate": 6.279388918952527e-06,
      "loss": 0.544,
      "step": 6868
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0941506624221802,
      "learning_rate": 6.278397054555275e-06,
      "loss": 0.5481,
      "step": 6869
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0862221717834473,
      "learning_rate": 6.277405136329169e-06,
      "loss": 0.5898,
      "step": 6870
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.996271550655365,
      "learning_rate": 6.276413164315978e-06,
      "loss": 0.4989,
      "step": 6871
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9512807726860046,
      "learning_rate": 6.2754211385574674e-06,
      "loss": 0.5268,
      "step": 6872
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0293751955032349,
      "learning_rate": 6.274429059095411e-06,
      "loss": 0.5562,
      "step": 6873
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.046460509300232,
      "learning_rate": 6.273436925971578e-06,
      "loss": 0.537,
      "step": 6874
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9507560729980469,
      "learning_rate": 6.272444739227748e-06,
      "loss": 0.4754,
      "step": 6875
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.004612684249878,
      "learning_rate": 6.2714524989056945e-06,
      "loss": 0.5451,
      "step": 6876
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1527137756347656,
      "learning_rate": 6.270460205047202e-06,
      "loss": 0.5796,
      "step": 6877
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9972601532936096,
      "learning_rate": 6.269467857694047e-06,
      "loss": 0.4971,
      "step": 6878
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0107241868972778,
      "learning_rate": 6.268475456888019e-06,
      "loss": 0.5368,
      "step": 6879
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9781653881072998,
      "learning_rate": 6.2674830026709014e-06,
      "loss": 0.5635,
      "step": 6880
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0893415212631226,
      "learning_rate": 6.266490495084484e-06,
      "loss": 0.5249,
      "step": 6881
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.066577672958374,
      "learning_rate": 6.265497934170559e-06,
      "loss": 0.5358,
      "step": 6882
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0110509395599365,
      "learning_rate": 6.264505319970915e-06,
      "loss": 0.5135,
      "step": 6883
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8866680264472961,
      "learning_rate": 6.263512652527353e-06,
      "loss": 0.5053,
      "step": 6884
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0273590087890625,
      "learning_rate": 6.262519931881669e-06,
      "loss": 0.5519,
      "step": 6885
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0803920030593872,
      "learning_rate": 6.261527158075662e-06,
      "loss": 0.5404,
      "step": 6886
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9233646392822266,
      "learning_rate": 6.260534331151133e-06,
      "loss": 0.4459,
      "step": 6887
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0693546533584595,
      "learning_rate": 6.259541451149892e-06,
      "loss": 0.4995,
      "step": 6888
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0313082933425903,
      "learning_rate": 6.258548518113741e-06,
      "loss": 0.5383,
      "step": 6889
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9983394145965576,
      "learning_rate": 6.257555532084489e-06,
      "loss": 0.4719,
      "step": 6890
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0172432661056519,
      "learning_rate": 6.2565624931039485e-06,
      "loss": 0.5472,
      "step": 6891
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0359172821044922,
      "learning_rate": 6.255569401213933e-06,
      "loss": 0.5452,
      "step": 6892
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0232447385787964,
      "learning_rate": 6.254576256456257e-06,
      "loss": 0.5078,
      "step": 6893
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8931428790092468,
      "learning_rate": 6.253583058872741e-06,
      "loss": 0.5009,
      "step": 6894
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9559874534606934,
      "learning_rate": 6.2525898085052005e-06,
      "loss": 0.5176,
      "step": 6895
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0395197868347168,
      "learning_rate": 6.251596505395463e-06,
      "loss": 0.5279,
      "step": 6896
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0660779476165771,
      "learning_rate": 6.25060314958535e-06,
      "loss": 0.5436,
      "step": 6897
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0156817436218262,
      "learning_rate": 6.249609741116689e-06,
      "loss": 0.5115,
      "step": 6898
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0072088241577148,
      "learning_rate": 6.24861628003131e-06,
      "loss": 0.5693,
      "step": 6899
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0126852989196777,
      "learning_rate": 6.247622766371041e-06,
      "loss": 0.5165,
      "step": 6900
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0487877130508423,
      "learning_rate": 6.246629200177718e-06,
      "loss": 0.571,
      "step": 6901
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0167573690414429,
      "learning_rate": 6.245635581493176e-06,
      "loss": 0.5,
      "step": 6902
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.094506025314331,
      "learning_rate": 6.244641910359254e-06,
      "loss": 0.6057,
      "step": 6903
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0023902654647827,
      "learning_rate": 6.24364818681779e-06,
      "loss": 0.5305,
      "step": 6904
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9998495578765869,
      "learning_rate": 6.242654410910628e-06,
      "loss": 0.5083,
      "step": 6905
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.012750267982483,
      "learning_rate": 6.2416605826796095e-06,
      "loss": 0.5704,
      "step": 6906
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.027174472808838,
      "learning_rate": 6.240666702166587e-06,
      "loss": 0.5447,
      "step": 6907
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9723641276359558,
      "learning_rate": 6.239672769413403e-06,
      "loss": 0.5385,
      "step": 6908
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.960335910320282,
      "learning_rate": 6.238678784461913e-06,
      "loss": 0.4809,
      "step": 6909
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0493924617767334,
      "learning_rate": 6.237684747353965e-06,
      "loss": 0.5192,
      "step": 6910
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9411200881004333,
      "learning_rate": 6.23669065813142e-06,
      "loss": 0.5224,
      "step": 6911
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9179388284683228,
      "learning_rate": 6.235696516836134e-06,
      "loss": 0.5309,
      "step": 6912
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0288889408111572,
      "learning_rate": 6.234702323509967e-06,
      "loss": 0.5611,
      "step": 6913
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.042685627937317,
      "learning_rate": 6.233708078194778e-06,
      "loss": 0.5618,
      "step": 6914
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0587626695632935,
      "learning_rate": 6.232713780932434e-06,
      "loss": 0.5725,
      "step": 6915
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0170725584030151,
      "learning_rate": 6.231719431764804e-06,
      "loss": 0.5366,
      "step": 6916
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0194566249847412,
      "learning_rate": 6.230725030733751e-06,
      "loss": 0.562,
      "step": 6917
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0699495077133179,
      "learning_rate": 6.229730577881148e-06,
      "loss": 0.5892,
      "step": 6918
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9561670422554016,
      "learning_rate": 6.2287360732488685e-06,
      "loss": 0.474,
      "step": 6919
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.077793002128601,
      "learning_rate": 6.227741516878789e-06,
      "loss": 0.5547,
      "step": 6920
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0175807476043701,
      "learning_rate": 6.226746908812784e-06,
      "loss": 0.5531,
      "step": 6921
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2047051191329956,
      "learning_rate": 6.225752249092734e-06,
      "loss": 0.5856,
      "step": 6922
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1163352727890015,
      "learning_rate": 6.224757537760521e-06,
      "loss": 0.5526,
      "step": 6923
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0539308786392212,
      "learning_rate": 6.2237627748580294e-06,
      "loss": 0.5748,
      "step": 6924
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.031957745552063,
      "learning_rate": 6.222767960427144e-06,
      "loss": 0.5798,
      "step": 6925
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0769531726837158,
      "learning_rate": 6.221773094509753e-06,
      "loss": 0.5501,
      "step": 6926
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.992579996585846,
      "learning_rate": 6.220778177147747e-06,
      "loss": 0.5198,
      "step": 6927
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0876187086105347,
      "learning_rate": 6.219783208383021e-06,
      "loss": 0.5047,
      "step": 6928
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.100014090538025,
      "learning_rate": 6.218788188257465e-06,
      "loss": 0.5257,
      "step": 6929
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9707440137863159,
      "learning_rate": 6.217793116812979e-06,
      "loss": 0.492,
      "step": 6930
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0497099161148071,
      "learning_rate": 6.216797994091462e-06,
      "loss": 0.5627,
      "step": 6931
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.900424063205719,
      "learning_rate": 6.215802820134814e-06,
      "loss": 0.4742,
      "step": 6932
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9930451512336731,
      "learning_rate": 6.214807594984939e-06,
      "loss": 0.5533,
      "step": 6933
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0777264833450317,
      "learning_rate": 6.213812318683741e-06,
      "loss": 0.5664,
      "step": 6934
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.008448839187622,
      "learning_rate": 6.2128169912731295e-06,
      "loss": 0.562,
      "step": 6935
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.102477788925171,
      "learning_rate": 6.211821612795014e-06,
      "loss": 0.53,
      "step": 6936
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0119174718856812,
      "learning_rate": 6.210826183291305e-06,
      "loss": 0.5147,
      "step": 6937
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9260329008102417,
      "learning_rate": 6.209830702803918e-06,
      "loss": 0.4912,
      "step": 6938
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0289298295974731,
      "learning_rate": 6.208835171374769e-06,
      "loss": 0.5524,
      "step": 6939
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0635002851486206,
      "learning_rate": 6.207839589045777e-06,
      "loss": 0.5207,
      "step": 6940
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.086357593536377,
      "learning_rate": 6.20684395585886e-06,
      "loss": 0.5053,
      "step": 6941
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.067955493927002,
      "learning_rate": 6.205848271855943e-06,
      "loss": 0.5373,
      "step": 6942
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1029582023620605,
      "learning_rate": 6.204852537078952e-06,
      "loss": 0.5356,
      "step": 6943
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0627195835113525,
      "learning_rate": 6.203856751569809e-06,
      "loss": 0.6011,
      "step": 6944
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0238324403762817,
      "learning_rate": 6.202860915370447e-06,
      "loss": 0.5398,
      "step": 6945
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9644851684570312,
      "learning_rate": 6.201865028522798e-06,
      "loss": 0.4662,
      "step": 6946
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0515786409378052,
      "learning_rate": 6.200869091068791e-06,
      "loss": 0.5293,
      "step": 6947
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0306024551391602,
      "learning_rate": 6.1998731030503655e-06,
      "loss": 0.4727,
      "step": 6948
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0533422231674194,
      "learning_rate": 6.198877064509458e-06,
      "loss": 0.5211,
      "step": 6949
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1155869960784912,
      "learning_rate": 6.1978809754880076e-06,
      "loss": 0.565,
      "step": 6950
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0015124082565308,
      "learning_rate": 6.196884836027957e-06,
      "loss": 0.4974,
      "step": 6951
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9570860862731934,
      "learning_rate": 6.195888646171247e-06,
      "loss": 0.4626,
      "step": 6952
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9626646637916565,
      "learning_rate": 6.194892405959829e-06,
      "loss": 0.5341,
      "step": 6953
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0123460292816162,
      "learning_rate": 6.193896115435648e-06,
      "loss": 0.5306,
      "step": 6954
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9832209348678589,
      "learning_rate": 6.192899774640655e-06,
      "loss": 0.5549,
      "step": 6955
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0607764720916748,
      "learning_rate": 6.191903383616801e-06,
      "loss": 0.5334,
      "step": 6956
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.986181378364563,
      "learning_rate": 6.190906942406043e-06,
      "loss": 0.4941,
      "step": 6957
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0732365846633911,
      "learning_rate": 6.189910451050336e-06,
      "loss": 0.5171,
      "step": 6958
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1563911437988281,
      "learning_rate": 6.1889139095916395e-06,
      "loss": 0.5898,
      "step": 6959
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0473549365997314,
      "learning_rate": 6.187917318071914e-06,
      "loss": 0.5564,
      "step": 6960
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9854289889335632,
      "learning_rate": 6.1869206765331234e-06,
      "loss": 0.5256,
      "step": 6961
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9489195942878723,
      "learning_rate": 6.1859239850172325e-06,
      "loss": 0.5141,
      "step": 6962
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9980160593986511,
      "learning_rate": 6.1849272435662065e-06,
      "loss": 0.5251,
      "step": 6963
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0156224966049194,
      "learning_rate": 6.183930452222017e-06,
      "loss": 0.5046,
      "step": 6964
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0200015306472778,
      "learning_rate": 6.1829336110266356e-06,
      "loss": 0.5237,
      "step": 6965
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9782587289810181,
      "learning_rate": 6.181936720022033e-06,
      "loss": 0.5378,
      "step": 6966
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.067857027053833,
      "learning_rate": 6.180939779250188e-06,
      "loss": 0.5529,
      "step": 6967
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0193872451782227,
      "learning_rate": 6.179942788753077e-06,
      "loss": 0.5597,
      "step": 6968
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2080793380737305,
      "learning_rate": 6.178945748572681e-06,
      "loss": 0.5478,
      "step": 6969
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0767008066177368,
      "learning_rate": 6.177948658750979e-06,
      "loss": 0.465,
      "step": 6970
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0378178358078003,
      "learning_rate": 6.176951519329958e-06,
      "loss": 0.5356,
      "step": 6971
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.094118595123291,
      "learning_rate": 6.1759543303516025e-06,
      "loss": 0.5179,
      "step": 6972
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1048043966293335,
      "learning_rate": 6.174957091857901e-06,
      "loss": 0.5683,
      "step": 6973
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9769123792648315,
      "learning_rate": 6.173959803890843e-06,
      "loss": 0.5302,
      "step": 6974
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1159851551055908,
      "learning_rate": 6.172962466492423e-06,
      "loss": 0.5453,
      "step": 6975
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9689269661903381,
      "learning_rate": 6.171965079704634e-06,
      "loss": 0.5562,
      "step": 6976
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0246952772140503,
      "learning_rate": 6.17096764356947e-06,
      "loss": 0.5288,
      "step": 6977
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9824046492576599,
      "learning_rate": 6.169970158128935e-06,
      "loss": 0.5636,
      "step": 6978
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.055438756942749,
      "learning_rate": 6.168972623425023e-06,
      "loss": 0.5349,
      "step": 6979
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0025379657745361,
      "learning_rate": 6.167975039499744e-06,
      "loss": 0.5189,
      "step": 6980
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9753031134605408,
      "learning_rate": 6.1669774063950985e-06,
      "loss": 0.6042,
      "step": 6981
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.338631510734558,
      "learning_rate": 6.165979724153094e-06,
      "loss": 0.5728,
      "step": 6982
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0687596797943115,
      "learning_rate": 6.164981992815737e-06,
      "loss": 0.5211,
      "step": 6983
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9963400363922119,
      "learning_rate": 6.163984212425043e-06,
      "loss": 0.5205,
      "step": 6984
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1149332523345947,
      "learning_rate": 6.162986383023023e-06,
      "loss": 0.5527,
      "step": 6985
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0832549333572388,
      "learning_rate": 6.161988504651692e-06,
      "loss": 0.5406,
      "step": 6986
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0374681949615479,
      "learning_rate": 6.160990577353066e-06,
      "loss": 0.5424,
      "step": 6987
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9844253659248352,
      "learning_rate": 6.1599926011691695e-06,
      "loss": 0.5301,
      "step": 6988
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1734834909439087,
      "learning_rate": 6.1589945761420166e-06,
      "loss": 0.5422,
      "step": 6989
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9257333278656006,
      "learning_rate": 6.157996502313635e-06,
      "loss": 0.5301,
      "step": 6990
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9583601355552673,
      "learning_rate": 6.156998379726048e-06,
      "loss": 0.4917,
      "step": 6991
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.039122462272644,
      "learning_rate": 6.1560002084212845e-06,
      "loss": 0.5045,
      "step": 6992
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9951550364494324,
      "learning_rate": 6.155001988441375e-06,
      "loss": 0.513,
      "step": 6993
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.099072813987732,
      "learning_rate": 6.154003719828349e-06,
      "loss": 0.5187,
      "step": 6994
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0065536499023438,
      "learning_rate": 6.1530054026242405e-06,
      "loss": 0.466,
      "step": 6995
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9675000905990601,
      "learning_rate": 6.152007036871085e-06,
      "loss": 0.4686,
      "step": 6996
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9593127369880676,
      "learning_rate": 6.151008622610921e-06,
      "loss": 0.5441,
      "step": 6997
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.01827871799469,
      "learning_rate": 6.150010159885789e-06,
      "loss": 0.4729,
      "step": 6998
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.112960934638977,
      "learning_rate": 6.149011648737728e-06,
      "loss": 0.5966,
      "step": 6999
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0497690439224243,
      "learning_rate": 6.148013089208784e-06,
      "loss": 0.5707,
      "step": 7000
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0520048141479492,
      "learning_rate": 6.1470144813410045e-06,
      "loss": 0.5515,
      "step": 7001
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0527669191360474,
      "learning_rate": 6.146015825176432e-06,
      "loss": 0.5006,
      "step": 7002
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.03863525390625,
      "learning_rate": 6.145017120757123e-06,
      "loss": 0.5391,
      "step": 7003
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9999639987945557,
      "learning_rate": 6.144018368125124e-06,
      "loss": 0.5037,
      "step": 7004
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0270698070526123,
      "learning_rate": 6.143019567322493e-06,
      "loss": 0.4914,
      "step": 7005
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.092484951019287,
      "learning_rate": 6.1420207183912824e-06,
      "loss": 0.5923,
      "step": 7006
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9754437804222107,
      "learning_rate": 6.141021821373555e-06,
      "loss": 0.5353,
      "step": 7007
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.034129023551941,
      "learning_rate": 6.140022876311367e-06,
      "loss": 0.5276,
      "step": 7008
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9061996936798096,
      "learning_rate": 6.139023883246781e-06,
      "loss": 0.4921,
      "step": 7009
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0841065645217896,
      "learning_rate": 6.1380248422218604e-06,
      "loss": 0.465,
      "step": 7010
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0482650995254517,
      "learning_rate": 6.137025753278673e-06,
      "loss": 0.5501,
      "step": 7011
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.004610538482666,
      "learning_rate": 6.1360266164592886e-06,
      "loss": 0.5599,
      "step": 7012
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0636749267578125,
      "learning_rate": 6.135027431805774e-06,
      "loss": 0.561,
      "step": 7013
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9683583378791809,
      "learning_rate": 6.134028199360203e-06,
      "loss": 0.4905,
      "step": 7014
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.003348708152771,
      "learning_rate": 6.133028919164647e-06,
      "loss": 0.5758,
      "step": 7015
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9529774785041809,
      "learning_rate": 6.132029591261188e-06,
      "loss": 0.5214,
      "step": 7016
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1036036014556885,
      "learning_rate": 6.1310302156919e-06,
      "loss": 0.5496,
      "step": 7017
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0264195203781128,
      "learning_rate": 6.130030792498865e-06,
      "loss": 0.5335,
      "step": 7018
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0804702043533325,
      "learning_rate": 6.129031321724163e-06,
      "loss": 0.5226,
      "step": 7019
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.05149245262146,
      "learning_rate": 6.128031803409881e-06,
      "loss": 0.5137,
      "step": 7020
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.036487340927124,
      "learning_rate": 6.127032237598102e-06,
      "loss": 0.5587,
      "step": 7021
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9598145484924316,
      "learning_rate": 6.126032624330917e-06,
      "loss": 0.5237,
      "step": 7022
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0859782695770264,
      "learning_rate": 6.125032963650417e-06,
      "loss": 0.5699,
      "step": 7023
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0361154079437256,
      "learning_rate": 6.124033255598691e-06,
      "loss": 0.5027,
      "step": 7024
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0015485286712646,
      "learning_rate": 6.1230335002178345e-06,
      "loss": 0.5339,
      "step": 7025
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0793373584747314,
      "learning_rate": 6.1220336975499435e-06,
      "loss": 0.5159,
      "step": 7026
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.102267861366272,
      "learning_rate": 6.121033847637119e-06,
      "loss": 0.5646,
      "step": 7027
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0097570419311523,
      "learning_rate": 6.120033950521458e-06,
      "loss": 0.5047,
      "step": 7028
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9865574836730957,
      "learning_rate": 6.119034006245063e-06,
      "loss": 0.5083,
      "step": 7029
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0643231868743896,
      "learning_rate": 6.118034014850039e-06,
      "loss": 0.5583,
      "step": 7030
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9546524882316589,
      "learning_rate": 6.117033976378493e-06,
      "loss": 0.4621,
      "step": 7031
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.061911940574646,
      "learning_rate": 6.116033890872531e-06,
      "loss": 0.5029,
      "step": 7032
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0499768257141113,
      "learning_rate": 6.115033758374265e-06,
      "loss": 0.5618,
      "step": 7033
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9878824353218079,
      "learning_rate": 6.114033578925805e-06,
      "loss": 0.5288,
      "step": 7034
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0290721654891968,
      "learning_rate": 6.1130333525692684e-06,
      "loss": 0.5213,
      "step": 7035
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9989311099052429,
      "learning_rate": 6.112033079346767e-06,
      "loss": 0.5193,
      "step": 7036
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9303186535835266,
      "learning_rate": 6.111032759300423e-06,
      "loss": 0.5508,
      "step": 7037
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9461025595664978,
      "learning_rate": 6.110032392472354e-06,
      "loss": 0.5139,
      "step": 7038
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9654031991958618,
      "learning_rate": 6.109031978904683e-06,
      "loss": 0.5021,
      "step": 7039
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9956388473510742,
      "learning_rate": 6.108031518639532e-06,
      "loss": 0.5524,
      "step": 7040
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.084328532218933,
      "learning_rate": 6.107031011719029e-06,
      "loss": 0.6076,
      "step": 7041
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9746370911598206,
      "learning_rate": 6.106030458185303e-06,
      "loss": 0.5185,
      "step": 7042
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0461184978485107,
      "learning_rate": 6.105029858080479e-06,
      "loss": 0.5268,
      "step": 7043
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9482165575027466,
      "learning_rate": 6.1040292114466935e-06,
      "loss": 0.5076,
      "step": 7044
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.044372797012329,
      "learning_rate": 6.103028518326077e-06,
      "loss": 0.5106,
      "step": 7045
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.982097864151001,
      "learning_rate": 6.102027778760769e-06,
      "loss": 0.5403,
      "step": 7046
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0360002517700195,
      "learning_rate": 6.101026992792904e-06,
      "loss": 0.5853,
      "step": 7047
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9550833702087402,
      "learning_rate": 6.100026160464621e-06,
      "loss": 0.511,
      "step": 7048
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9916879534721375,
      "learning_rate": 6.099025281818065e-06,
      "loss": 0.5023,
      "step": 7049
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0016727447509766,
      "learning_rate": 6.098024356895378e-06,
      "loss": 0.514,
      "step": 7050
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.995428740978241,
      "learning_rate": 6.097023385738704e-06,
      "loss": 0.4829,
      "step": 7051
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0760793685913086,
      "learning_rate": 6.096022368390191e-06,
      "loss": 0.5491,
      "step": 7052
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0174927711486816,
      "learning_rate": 6.0950213048919895e-06,
      "loss": 0.5249,
      "step": 7053
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9960131049156189,
      "learning_rate": 6.094020195286251e-06,
      "loss": 0.5026,
      "step": 7054
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0475268363952637,
      "learning_rate": 6.093019039615128e-06,
      "loss": 0.5887,
      "step": 7055
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9496270418167114,
      "learning_rate": 6.092017837920773e-06,
      "loss": 0.5053,
      "step": 7056
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.999090313911438,
      "learning_rate": 6.091016590245347e-06,
      "loss": 0.51,
      "step": 7057
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0329222679138184,
      "learning_rate": 6.090015296631009e-06,
      "loss": 0.4875,
      "step": 7058
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0270658731460571,
      "learning_rate": 6.089013957119918e-06,
      "loss": 0.5859,
      "step": 7059
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9140265583992004,
      "learning_rate": 6.088012571754236e-06,
      "loss": 0.4633,
      "step": 7060
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1130242347717285,
      "learning_rate": 6.087011140576132e-06,
      "loss": 0.5435,
      "step": 7061
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9865022897720337,
      "learning_rate": 6.086009663627769e-06,
      "loss": 0.5155,
      "step": 7062
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0083192586898804,
      "learning_rate": 6.085008140951318e-06,
      "loss": 0.5141,
      "step": 7063
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1197551488876343,
      "learning_rate": 6.084006572588947e-06,
      "loss": 0.5497,
      "step": 7064
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0250638723373413,
      "learning_rate": 6.083004958582832e-06,
      "loss": 0.4772,
      "step": 7065
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0011177062988281,
      "learning_rate": 6.082003298975144e-06,
      "loss": 0.5462,
      "step": 7066
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0027698278427124,
      "learning_rate": 6.081001593808063e-06,
      "loss": 0.5069,
      "step": 7067
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0362638235092163,
      "learning_rate": 6.079999843123763e-06,
      "loss": 0.5839,
      "step": 7068
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9713705778121948,
      "learning_rate": 6.07899804696443e-06,
      "loss": 0.5555,
      "step": 7069
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.039381980895996,
      "learning_rate": 6.077996205372241e-06,
      "loss": 0.4965,
      "step": 7070
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9748886823654175,
      "learning_rate": 6.07699431838938e-06,
      "loss": 0.5145,
      "step": 7071
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0257819890975952,
      "learning_rate": 6.075992386058037e-06,
      "loss": 0.5321,
      "step": 7072
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.067957878112793,
      "learning_rate": 6.074990408420397e-06,
      "loss": 0.5104,
      "step": 7073
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9645175337791443,
      "learning_rate": 6.073988385518652e-06,
      "loss": 0.4861,
      "step": 7074
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9836180210113525,
      "learning_rate": 6.07298631739499e-06,
      "loss": 0.5501,
      "step": 7075
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9671414494514465,
      "learning_rate": 6.071984204091608e-06,
      "loss": 0.5687,
      "step": 7076
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.02313232421875,
      "learning_rate": 6.0709820456507e-06,
      "loss": 0.5469,
      "step": 7077
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0085055828094482,
      "learning_rate": 6.069979842114465e-06,
      "loss": 0.5109,
      "step": 7078
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1000394821166992,
      "learning_rate": 6.068977593525098e-06,
      "loss": 0.5624,
      "step": 7079
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9950181245803833,
      "learning_rate": 6.067975299924806e-06,
      "loss": 0.5516,
      "step": 7080
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0332417488098145,
      "learning_rate": 6.066972961355788e-06,
      "loss": 0.5637,
      "step": 7081
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9457080364227295,
      "learning_rate": 6.065970577860252e-06,
      "loss": 0.5312,
      "step": 7082
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.035883903503418,
      "learning_rate": 6.0649681494804014e-06,
      "loss": 0.5549,
      "step": 7083
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9677972197532654,
      "learning_rate": 6.063965676258448e-06,
      "loss": 0.5325,
      "step": 7084
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9882851839065552,
      "learning_rate": 6.0629631582366015e-06,
      "loss": 0.549,
      "step": 7085
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9976891875267029,
      "learning_rate": 6.0619605954570726e-06,
      "loss": 0.5047,
      "step": 7086
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9844744801521301,
      "learning_rate": 6.060957987962077e-06,
      "loss": 0.5322,
      "step": 7087
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0253148078918457,
      "learning_rate": 6.059955335793832e-06,
      "loss": 0.5551,
      "step": 7088
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0289077758789062,
      "learning_rate": 6.0589526389945576e-06,
      "loss": 0.5105,
      "step": 7089
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9292216897010803,
      "learning_rate": 6.057949897606469e-06,
      "loss": 0.5282,
      "step": 7090
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0359158515930176,
      "learning_rate": 6.05694711167179e-06,
      "loss": 0.5276,
      "step": 7091
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0398133993148804,
      "learning_rate": 6.055944281232746e-06,
      "loss": 0.5545,
      "step": 7092
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0795741081237793,
      "learning_rate": 6.0549414063315625e-06,
      "loss": 0.5894,
      "step": 7093
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0921592712402344,
      "learning_rate": 6.053938487010464e-06,
      "loss": 0.5194,
      "step": 7094
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.942608654499054,
      "learning_rate": 6.052935523311684e-06,
      "loss": 0.516,
      "step": 7095
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9816745519638062,
      "learning_rate": 6.0519325152774515e-06,
      "loss": 0.5125,
      "step": 7096
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.950702428817749,
      "learning_rate": 6.05092946295e-06,
      "loss": 0.5427,
      "step": 7097
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9388405084609985,
      "learning_rate": 6.049926366371565e-06,
      "loss": 0.5388,
      "step": 7098
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9361859560012817,
      "learning_rate": 6.048923225584383e-06,
      "loss": 0.5181,
      "step": 7099
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0264475345611572,
      "learning_rate": 6.047920040630692e-06,
      "loss": 0.5171,
      "step": 7100
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9536789059638977,
      "learning_rate": 6.046916811552735e-06,
      "loss": 0.5023,
      "step": 7101
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.930137574672699,
      "learning_rate": 6.045913538392754e-06,
      "loss": 0.5208,
      "step": 7102
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9892581701278687,
      "learning_rate": 6.04491022119299e-06,
      "loss": 0.5434,
      "step": 7103
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9390099048614502,
      "learning_rate": 6.043906859995693e-06,
      "loss": 0.4808,
      "step": 7104
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.090390920639038,
      "learning_rate": 6.042903454843109e-06,
      "loss": 0.5112,
      "step": 7105
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1024433374404907,
      "learning_rate": 6.041900005777488e-06,
      "loss": 0.6206,
      "step": 7106
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0256569385528564,
      "learning_rate": 6.040896512841083e-06,
      "loss": 0.5461,
      "step": 7107
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.032071828842163,
      "learning_rate": 6.039892976076147e-06,
      "loss": 0.5021,
      "step": 7108
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.000108003616333,
      "learning_rate": 6.038889395524935e-06,
      "loss": 0.5312,
      "step": 7109
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0488094091415405,
      "learning_rate": 6.037885771229703e-06,
      "loss": 0.5545,
      "step": 7110
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9266911745071411,
      "learning_rate": 6.036882103232714e-06,
      "loss": 0.5378,
      "step": 7111
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0361651182174683,
      "learning_rate": 6.0358783915762265e-06,
      "loss": 0.5222,
      "step": 7112
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0582817792892456,
      "learning_rate": 6.034874636302502e-06,
      "loss": 0.5234,
      "step": 7113
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1131682395935059,
      "learning_rate": 6.033870837453808e-06,
      "loss": 0.5866,
      "step": 7114
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0257360935211182,
      "learning_rate": 6.0328669950724096e-06,
      "loss": 0.5094,
      "step": 7115
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0267966985702515,
      "learning_rate": 6.031863109200575e-06,
      "loss": 0.5952,
      "step": 7116
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0467355251312256,
      "learning_rate": 6.030859179880574e-06,
      "loss": 0.5453,
      "step": 7117
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.02724289894104,
      "learning_rate": 6.029855207154679e-06,
      "loss": 0.5244,
      "step": 7118
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0115382671356201,
      "learning_rate": 6.0288511910651644e-06,
      "loss": 0.5273,
      "step": 7119
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9595180749893188,
      "learning_rate": 6.027847131654305e-06,
      "loss": 0.518,
      "step": 7120
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.017401933670044,
      "learning_rate": 6.026843028964378e-06,
      "loss": 0.6056,
      "step": 7121
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0124104022979736,
      "learning_rate": 6.025838883037664e-06,
      "loss": 0.5109,
      "step": 7122
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.096415400505066,
      "learning_rate": 6.024834693916443e-06,
      "loss": 0.5399,
      "step": 7123
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0635879039764404,
      "learning_rate": 6.023830461642998e-06,
      "loss": 0.5239,
      "step": 7124
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9495733380317688,
      "learning_rate": 6.022826186259614e-06,
      "loss": 0.5079,
      "step": 7125
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0496259927749634,
      "learning_rate": 6.021821867808576e-06,
      "loss": 0.5161,
      "step": 7126
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0454121828079224,
      "learning_rate": 6.0208175063321765e-06,
      "loss": 0.5236,
      "step": 7127
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9280383586883545,
      "learning_rate": 6.019813101872701e-06,
      "loss": 0.4974,
      "step": 7128
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.033594012260437,
      "learning_rate": 6.018808654472445e-06,
      "loss": 0.5289,
      "step": 7129
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0081610679626465,
      "learning_rate": 6.017804164173698e-06,
      "loss": 0.5561,
      "step": 7130
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1945390701293945,
      "learning_rate": 6.0167996310187615e-06,
      "loss": 0.5433,
      "step": 7131
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9437111616134644,
      "learning_rate": 6.015795055049929e-06,
      "loss": 0.483,
      "step": 7132
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.036550521850586,
      "learning_rate": 6.014790436309499e-06,
      "loss": 0.5601,
      "step": 7133
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9895676970481873,
      "learning_rate": 6.013785774839776e-06,
      "loss": 0.4948,
      "step": 7134
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9519324898719788,
      "learning_rate": 6.012781070683058e-06,
      "loss": 0.559,
      "step": 7135
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0611015558242798,
      "learning_rate": 6.011776323881654e-06,
      "loss": 0.5345,
      "step": 7136
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.933640718460083,
      "learning_rate": 6.0107715344778684e-06,
      "loss": 0.4938,
      "step": 7137
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0362071990966797,
      "learning_rate": 6.00976670251401e-06,
      "loss": 0.5398,
      "step": 7138
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0097980499267578,
      "learning_rate": 6.008761828032389e-06,
      "loss": 0.5134,
      "step": 7139
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9700465798377991,
      "learning_rate": 6.007756911075315e-06,
      "loss": 0.4799,
      "step": 7140
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9842071533203125,
      "learning_rate": 6.006751951685104e-06,
      "loss": 0.5089,
      "step": 7141
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0734883546829224,
      "learning_rate": 6.005746949904072e-06,
      "loss": 0.5458,
      "step": 7142
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0110918283462524,
      "learning_rate": 6.004741905774533e-06,
      "loss": 0.5478,
      "step": 7143
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.040485143661499,
      "learning_rate": 6.003736819338808e-06,
      "loss": 0.5477,
      "step": 7144
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.105627417564392,
      "learning_rate": 6.0027316906392165e-06,
      "loss": 0.4995,
      "step": 7145
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0592191219329834,
      "learning_rate": 6.001726519718083e-06,
      "loss": 0.5105,
      "step": 7146
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0756666660308838,
      "learning_rate": 6.000721306617731e-06,
      "loss": 0.5461,
      "step": 7147
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0701828002929688,
      "learning_rate": 5.999716051380484e-06,
      "loss": 0.523,
      "step": 7148
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9982286691665649,
      "learning_rate": 5.998710754048674e-06,
      "loss": 0.5411,
      "step": 7149
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1478774547576904,
      "learning_rate": 5.997705414664627e-06,
      "loss": 0.5197,
      "step": 7150
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0219048261642456,
      "learning_rate": 5.996700033270676e-06,
      "loss": 0.474,
      "step": 7151
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0214338302612305,
      "learning_rate": 5.995694609909153e-06,
      "loss": 0.5173,
      "step": 7152
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.961108386516571,
      "learning_rate": 5.9946891446223955e-06,
      "loss": 0.513,
      "step": 7153
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0029499530792236,
      "learning_rate": 5.993683637452736e-06,
      "loss": 0.5055,
      "step": 7154
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0898581743240356,
      "learning_rate": 5.992678088442518e-06,
      "loss": 0.5773,
      "step": 7155
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9855949282646179,
      "learning_rate": 5.991672497634076e-06,
      "loss": 0.5103,
      "step": 7156
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9675732851028442,
      "learning_rate": 5.990666865069759e-06,
      "loss": 0.5209,
      "step": 7157
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9956784844398499,
      "learning_rate": 5.9896611907919034e-06,
      "loss": 0.5619,
      "step": 7158
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9192211031913757,
      "learning_rate": 5.98865547484286e-06,
      "loss": 0.4706,
      "step": 7159
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.973389744758606,
      "learning_rate": 5.9876497172649704e-06,
      "loss": 0.5267,
      "step": 7160
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0324655771255493,
      "learning_rate": 5.986643918100591e-06,
      "loss": 0.5635,
      "step": 7161
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.014541745185852,
      "learning_rate": 5.985638077392066e-06,
      "loss": 0.5196,
      "step": 7162
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9843694567680359,
      "learning_rate": 5.984632195181752e-06,
      "loss": 0.528,
      "step": 7163
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0023690462112427,
      "learning_rate": 5.983626271512e-06,
      "loss": 0.5346,
      "step": 7164
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1338468790054321,
      "learning_rate": 5.982620306425167e-06,
      "loss": 0.5567,
      "step": 7165
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0319536924362183,
      "learning_rate": 5.981614299963614e-06,
      "loss": 0.5341,
      "step": 7166
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0526453256607056,
      "learning_rate": 5.9806082521696936e-06,
      "loss": 0.5511,
      "step": 7167
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0752928256988525,
      "learning_rate": 5.979602163085775e-06,
      "loss": 0.5323,
      "step": 7168
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.059898018836975,
      "learning_rate": 5.978596032754215e-06,
      "loss": 0.4771,
      "step": 7169
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9587810635566711,
      "learning_rate": 5.977589861217381e-06,
      "loss": 0.5201,
      "step": 7170
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0029340982437134,
      "learning_rate": 5.9765836485176376e-06,
      "loss": 0.5188,
      "step": 7171
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9671736359596252,
      "learning_rate": 5.9755773946973546e-06,
      "loss": 0.5466,
      "step": 7172
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.058545470237732,
      "learning_rate": 5.974571099798902e-06,
      "loss": 0.5128,
      "step": 7173
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9857040047645569,
      "learning_rate": 5.973564763864651e-06,
      "loss": 0.5702,
      "step": 7174
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9996533989906311,
      "learning_rate": 5.972558386936973e-06,
      "loss": 0.5469,
      "step": 7175
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0640591382980347,
      "learning_rate": 5.971551969058246e-06,
      "loss": 0.5474,
      "step": 7176
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0704646110534668,
      "learning_rate": 5.970545510270845e-06,
      "loss": 0.518,
      "step": 7177
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.013759732246399,
      "learning_rate": 5.969539010617149e-06,
      "loss": 0.4987,
      "step": 7178
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.059359073638916,
      "learning_rate": 5.968532470139537e-06,
      "loss": 0.5226,
      "step": 7179
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9865264892578125,
      "learning_rate": 5.967525888880392e-06,
      "loss": 0.5028,
      "step": 7180
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.984788179397583,
      "learning_rate": 5.966519266882099e-06,
      "loss": 0.5146,
      "step": 7181
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.011838436126709,
      "learning_rate": 5.965512604187041e-06,
      "loss": 0.4846,
      "step": 7182
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.053436279296875,
      "learning_rate": 5.964505900837606e-06,
      "loss": 0.5144,
      "step": 7183
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0706701278686523,
      "learning_rate": 5.963499156876182e-06,
      "loss": 0.5005,
      "step": 7184
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0562870502471924,
      "learning_rate": 5.962492372345163e-06,
      "loss": 0.5236,
      "step": 7185
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.06119704246521,
      "learning_rate": 5.961485547286936e-06,
      "loss": 0.5663,
      "step": 7186
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9982765913009644,
      "learning_rate": 5.960478681743897e-06,
      "loss": 0.4916,
      "step": 7187
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.040130615234375,
      "learning_rate": 5.959471775758444e-06,
      "loss": 0.5832,
      "step": 7188
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0240367650985718,
      "learning_rate": 5.9584648293729715e-06,
      "loss": 0.5293,
      "step": 7189
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9615424275398254,
      "learning_rate": 5.957457842629879e-06,
      "loss": 0.5175,
      "step": 7190
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9359433650970459,
      "learning_rate": 5.956450815571567e-06,
      "loss": 0.484,
      "step": 7191
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.009809970855713,
      "learning_rate": 5.955443748240439e-06,
      "loss": 0.4816,
      "step": 7192
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9725993275642395,
      "learning_rate": 5.9544366406789e-06,
      "loss": 0.5377,
      "step": 7193
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.033837914466858,
      "learning_rate": 5.953429492929352e-06,
      "loss": 0.5023,
      "step": 7194
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.032569169998169,
      "learning_rate": 5.952422305034206e-06,
      "loss": 0.4974,
      "step": 7195
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3495159149169922,
      "learning_rate": 5.95141507703587e-06,
      "loss": 0.5209,
      "step": 7196
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.002882480621338,
      "learning_rate": 5.9504078089767545e-06,
      "loss": 0.5489,
      "step": 7197
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0131049156188965,
      "learning_rate": 5.949400500899272e-06,
      "loss": 0.5326,
      "step": 7198
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9826775193214417,
      "learning_rate": 5.948393152845837e-06,
      "loss": 0.4392,
      "step": 7199
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0567611455917358,
      "learning_rate": 5.9473857648588665e-06,
      "loss": 0.5934,
      "step": 7200
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9829743504524231,
      "learning_rate": 5.9463783369807775e-06,
      "loss": 0.5364,
      "step": 7201
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9536810517311096,
      "learning_rate": 5.945370869253987e-06,
      "loss": 0.5174,
      "step": 7202
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.981810986995697,
      "learning_rate": 5.944363361720919e-06,
      "loss": 0.5436,
      "step": 7203
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.965626060962677,
      "learning_rate": 5.943355814423996e-06,
      "loss": 0.4884,
      "step": 7204
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0625500679016113,
      "learning_rate": 5.94234822740564e-06,
      "loss": 0.549,
      "step": 7205
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0152928829193115,
      "learning_rate": 5.941340600708279e-06,
      "loss": 0.5234,
      "step": 7206
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0407637357711792,
      "learning_rate": 5.9403329343743385e-06,
      "loss": 0.5577,
      "step": 7207
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0825738906860352,
      "learning_rate": 5.939325228446251e-06,
      "loss": 0.6029,
      "step": 7208
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9441272616386414,
      "learning_rate": 5.938317482966446e-06,
      "loss": 0.5153,
      "step": 7209
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9823698997497559,
      "learning_rate": 5.937309697977355e-06,
      "loss": 0.4753,
      "step": 7210
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9696682691574097,
      "learning_rate": 5.936301873521414e-06,
      "loss": 0.5201,
      "step": 7211
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.059671401977539,
      "learning_rate": 5.935294009641057e-06,
      "loss": 0.5362,
      "step": 7212
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0727226734161377,
      "learning_rate": 5.934286106378724e-06,
      "loss": 0.5466,
      "step": 7213
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0253998041152954,
      "learning_rate": 5.933278163776852e-06,
      "loss": 0.5549,
      "step": 7214
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.10127592086792,
      "learning_rate": 5.932270181877886e-06,
      "loss": 0.5668,
      "step": 7215
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9771760106086731,
      "learning_rate": 5.9312621607242625e-06,
      "loss": 0.5037,
      "step": 7216
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9859318733215332,
      "learning_rate": 5.93025410035843e-06,
      "loss": 0.5106,
      "step": 7217
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9642918109893799,
      "learning_rate": 5.929246000822835e-06,
      "loss": 0.5447,
      "step": 7218
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0542140007019043,
      "learning_rate": 5.928237862159922e-06,
      "loss": 0.537,
      "step": 7219
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0037953853607178,
      "learning_rate": 5.927229684412143e-06,
      "loss": 0.5388,
      "step": 7220
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9469515681266785,
      "learning_rate": 5.926221467621945e-06,
      "loss": 0.4964,
      "step": 7221
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9464107751846313,
      "learning_rate": 5.925213211831785e-06,
      "loss": 0.4892,
      "step": 7222
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0472491979599,
      "learning_rate": 5.924204917084116e-06,
      "loss": 0.5375,
      "step": 7223
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.16463303565979,
      "learning_rate": 5.923196583421392e-06,
      "loss": 0.5273,
      "step": 7224
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9915179014205933,
      "learning_rate": 5.922188210886071e-06,
      "loss": 0.5008,
      "step": 7225
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9429498314857483,
      "learning_rate": 5.921179799520613e-06,
      "loss": 0.5507,
      "step": 7226
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0695574283599854,
      "learning_rate": 5.920171349367478e-06,
      "loss": 0.54,
      "step": 7227
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0468008518218994,
      "learning_rate": 5.919162860469129e-06,
      "loss": 0.5241,
      "step": 7228
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.993760883808136,
      "learning_rate": 5.9181543328680295e-06,
      "loss": 0.5781,
      "step": 7229
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0547833442687988,
      "learning_rate": 5.917145766606645e-06,
      "loss": 0.5714,
      "step": 7230
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0336589813232422,
      "learning_rate": 5.9161371617274425e-06,
      "loss": 0.5129,
      "step": 7231
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0273765325546265,
      "learning_rate": 5.91512851827289e-06,
      "loss": 0.5227,
      "step": 7232
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9871826767921448,
      "learning_rate": 5.914119836285461e-06,
      "loss": 0.5027,
      "step": 7233
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9362433552742004,
      "learning_rate": 5.913111115807626e-06,
      "loss": 0.5302,
      "step": 7234
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0491071939468384,
      "learning_rate": 5.912102356881857e-06,
      "loss": 0.5021,
      "step": 7235
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0819073915481567,
      "learning_rate": 5.91109355955063e-06,
      "loss": 0.5088,
      "step": 7236
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9606463313102722,
      "learning_rate": 5.910084723856424e-06,
      "loss": 0.5218,
      "step": 7237
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9783573150634766,
      "learning_rate": 5.909075849841717e-06,
      "loss": 0.5265,
      "step": 7238
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9417604207992554,
      "learning_rate": 5.908066937548987e-06,
      "loss": 0.4648,
      "step": 7239
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.942327618598938,
      "learning_rate": 5.907057987020717e-06,
      "loss": 0.5158,
      "step": 7240
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9482901692390442,
      "learning_rate": 5.906048998299392e-06,
      "loss": 0.5049,
      "step": 7241
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9977984428405762,
      "learning_rate": 5.905039971427494e-06,
      "loss": 0.5542,
      "step": 7242
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9763760566711426,
      "learning_rate": 5.9040309064475136e-06,
      "loss": 0.5426,
      "step": 7243
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.97144615650177,
      "learning_rate": 5.903021803401933e-06,
      "loss": 0.5248,
      "step": 7244
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9501352310180664,
      "learning_rate": 5.902012662333248e-06,
      "loss": 0.4832,
      "step": 7245
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0225166082382202,
      "learning_rate": 5.9010034832839466e-06,
      "loss": 0.5706,
      "step": 7246
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0085699558258057,
      "learning_rate": 5.899994266296525e-06,
      "loss": 0.5263,
      "step": 7247
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0352619886398315,
      "learning_rate": 5.898985011413473e-06,
      "loss": 0.4869,
      "step": 7248
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9713915586471558,
      "learning_rate": 5.897975718677291e-06,
      "loss": 0.4974,
      "step": 7249
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.03853178024292,
      "learning_rate": 5.896966388130475e-06,
      "loss": 0.5162,
      "step": 7250
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9080546498298645,
      "learning_rate": 5.895957019815526e-06,
      "loss": 0.4928,
      "step": 7251
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.152125597000122,
      "learning_rate": 5.894947613774942e-06,
      "loss": 0.5041,
      "step": 7252
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8981519341468811,
      "learning_rate": 5.8939381700512275e-06,
      "loss": 0.4771,
      "step": 7253
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9719518423080444,
      "learning_rate": 5.892928688686887e-06,
      "loss": 0.5585,
      "step": 7254
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9432595372200012,
      "learning_rate": 5.891919169724426e-06,
      "loss": 0.5286,
      "step": 7255
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9516439437866211,
      "learning_rate": 5.890909613206351e-06,
      "loss": 0.5506,
      "step": 7256
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9617924094200134,
      "learning_rate": 5.889900019175171e-06,
      "loss": 0.4861,
      "step": 7257
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0408291816711426,
      "learning_rate": 5.888890387673398e-06,
      "loss": 0.5119,
      "step": 7258
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9240466952323914,
      "learning_rate": 5.887880718743541e-06,
      "loss": 0.4916,
      "step": 7259
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.952886164188385,
      "learning_rate": 5.886871012428117e-06,
      "loss": 0.5074,
      "step": 7260
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.010315179824829,
      "learning_rate": 5.885861268769641e-06,
      "loss": 0.5255,
      "step": 7261
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0000734329223633,
      "learning_rate": 5.8848514878106275e-06,
      "loss": 0.4968,
      "step": 7262
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0513924360275269,
      "learning_rate": 5.883841669593595e-06,
      "loss": 0.5587,
      "step": 7263
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9953838586807251,
      "learning_rate": 5.882831814161065e-06,
      "loss": 0.5463,
      "step": 7264
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9913728833198547,
      "learning_rate": 5.881821921555559e-06,
      "loss": 0.4985,
      "step": 7265
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0422443151474,
      "learning_rate": 5.880811991819601e-06,
      "loss": 0.5383,
      "step": 7266
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0038896799087524,
      "learning_rate": 5.879802024995712e-06,
      "loss": 0.5339,
      "step": 7267
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.005843162536621,
      "learning_rate": 5.878792021126421e-06,
      "loss": 0.5532,
      "step": 7268
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9732350707054138,
      "learning_rate": 5.877781980254255e-06,
      "loss": 0.5583,
      "step": 7269
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9197762608528137,
      "learning_rate": 5.876771902421743e-06,
      "loss": 0.4758,
      "step": 7270
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0640990734100342,
      "learning_rate": 5.875761787671416e-06,
      "loss": 0.539,
      "step": 7271
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0230790376663208,
      "learning_rate": 5.874751636045808e-06,
      "loss": 0.4872,
      "step": 7272
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9654695391654968,
      "learning_rate": 5.873741447587451e-06,
      "loss": 0.5178,
      "step": 7273
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.006279468536377,
      "learning_rate": 5.8727312223388814e-06,
      "loss": 0.5642,
      "step": 7274
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.96772301197052,
      "learning_rate": 5.871720960342635e-06,
      "loss": 0.527,
      "step": 7275
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0234496593475342,
      "learning_rate": 5.870710661641252e-06,
      "loss": 0.4944,
      "step": 7276
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9897413849830627,
      "learning_rate": 5.869700326277273e-06,
      "loss": 0.5446,
      "step": 7277
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9591410756111145,
      "learning_rate": 5.868689954293239e-06,
      "loss": 0.522,
      "step": 7278
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9390556812286377,
      "learning_rate": 5.86767954573169e-06,
      "loss": 0.4907,
      "step": 7279
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0197323560714722,
      "learning_rate": 5.866669100635176e-06,
      "loss": 0.5545,
      "step": 7280
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.019065260887146,
      "learning_rate": 5.865658619046242e-06,
      "loss": 0.5217,
      "step": 7281
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9386879801750183,
      "learning_rate": 5.864648101007433e-06,
      "loss": 0.4943,
      "step": 7282
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0120941400527954,
      "learning_rate": 5.863637546561301e-06,
      "loss": 0.4953,
      "step": 7283
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0714585781097412,
      "learning_rate": 5.862626955750397e-06,
      "loss": 0.5644,
      "step": 7284
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9985573887825012,
      "learning_rate": 5.8616163286172726e-06,
      "loss": 0.5713,
      "step": 7285
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.022886037826538,
      "learning_rate": 5.8606056652044805e-06,
      "loss": 0.5039,
      "step": 7286
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0849807262420654,
      "learning_rate": 5.859594965554579e-06,
      "loss": 0.5748,
      "step": 7287
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0356742143630981,
      "learning_rate": 5.858584229710124e-06,
      "loss": 0.5417,
      "step": 7288
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9849649667739868,
      "learning_rate": 5.857573457713674e-06,
      "loss": 0.508,
      "step": 7289
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9887516498565674,
      "learning_rate": 5.856562649607788e-06,
      "loss": 0.5373,
      "step": 7290
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0872023105621338,
      "learning_rate": 5.855551805435028e-06,
      "loss": 0.5253,
      "step": 7291
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0221582651138306,
      "learning_rate": 5.854540925237959e-06,
      "loss": 0.5486,
      "step": 7292
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1867270469665527,
      "learning_rate": 5.853530009059144e-06,
      "loss": 0.5338,
      "step": 7293
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0727027654647827,
      "learning_rate": 5.852519056941149e-06,
      "loss": 0.5335,
      "step": 7294
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.959506094455719,
      "learning_rate": 5.851508068926542e-06,
      "loss": 0.5079,
      "step": 7295
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.063923716545105,
      "learning_rate": 5.850497045057895e-06,
      "loss": 0.5529,
      "step": 7296
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0106937885284424,
      "learning_rate": 5.849485985377774e-06,
      "loss": 0.5366,
      "step": 7297
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0616973638534546,
      "learning_rate": 5.848474889928753e-06,
      "loss": 0.5117,
      "step": 7298
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.140915036201477,
      "learning_rate": 5.8474637587534065e-06,
      "loss": 0.5163,
      "step": 7299
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.064170479774475,
      "learning_rate": 5.84645259189431e-06,
      "loss": 0.5552,
      "step": 7300
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0307254791259766,
      "learning_rate": 5.845441389394039e-06,
      "loss": 0.5654,
      "step": 7301
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9613641500473022,
      "learning_rate": 5.844430151295171e-06,
      "loss": 0.5264,
      "step": 7302
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0288296937942505,
      "learning_rate": 5.843418877640289e-06,
      "loss": 0.5845,
      "step": 7303
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0588910579681396,
      "learning_rate": 5.842407568471971e-06,
      "loss": 0.534,
      "step": 7304
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.153749942779541,
      "learning_rate": 5.8413962238328e-06,
      "loss": 0.5142,
      "step": 7305
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.014904260635376,
      "learning_rate": 5.840384843765361e-06,
      "loss": 0.5243,
      "step": 7306
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.179085612297058,
      "learning_rate": 5.839373428312242e-06,
      "loss": 0.5021,
      "step": 7307
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0122723579406738,
      "learning_rate": 5.838361977516026e-06,
      "loss": 0.5477,
      "step": 7308
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.081276535987854,
      "learning_rate": 5.837350491419304e-06,
      "loss": 0.6057,
      "step": 7309
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0447216033935547,
      "learning_rate": 5.836338970064664e-06,
      "loss": 0.5004,
      "step": 7310
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0738158226013184,
      "learning_rate": 5.835327413494702e-06,
      "loss": 0.5689,
      "step": 7311
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9571483731269836,
      "learning_rate": 5.834315821752008e-06,
      "loss": 0.5389,
      "step": 7312
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1208528280258179,
      "learning_rate": 5.833304194879176e-06,
      "loss": 0.5823,
      "step": 7313
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.013647437095642,
      "learning_rate": 5.832292532918804e-06,
      "loss": 0.5109,
      "step": 7314
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9659610986709595,
      "learning_rate": 5.831280835913489e-06,
      "loss": 0.5205,
      "step": 7315
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9982751607894897,
      "learning_rate": 5.83026910390583e-06,
      "loss": 0.5645,
      "step": 7316
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9619269967079163,
      "learning_rate": 5.829257336938427e-06,
      "loss": 0.5456,
      "step": 7317
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9727155566215515,
      "learning_rate": 5.8282455350538815e-06,
      "loss": 0.5522,
      "step": 7318
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9817243814468384,
      "learning_rate": 5.827233698294799e-06,
      "loss": 0.5309,
      "step": 7319
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0565499067306519,
      "learning_rate": 5.826221826703783e-06,
      "loss": 0.6129,
      "step": 7320
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9987887740135193,
      "learning_rate": 5.825209920323438e-06,
      "loss": 0.5099,
      "step": 7321
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0380562543869019,
      "learning_rate": 5.824197979196377e-06,
      "loss": 0.5281,
      "step": 7322
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0254791975021362,
      "learning_rate": 5.823186003365205e-06,
      "loss": 0.5485,
      "step": 7323
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.097720980644226,
      "learning_rate": 5.822173992872534e-06,
      "loss": 0.4919,
      "step": 7324
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0166734457015991,
      "learning_rate": 5.821161947760975e-06,
      "loss": 0.4991,
      "step": 7325
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1378370523452759,
      "learning_rate": 5.820149868073145e-06,
      "loss": 0.633,
      "step": 7326
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.127633810043335,
      "learning_rate": 5.819137753851656e-06,
      "loss": 0.5529,
      "step": 7327
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0835357904434204,
      "learning_rate": 5.8181256051391276e-06,
      "loss": 0.5081,
      "step": 7328
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1418222188949585,
      "learning_rate": 5.817113421978173e-06,
      "loss": 0.5587,
      "step": 7329
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1440740823745728,
      "learning_rate": 5.816101204411417e-06,
      "loss": 0.5128,
      "step": 7330
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9324057102203369,
      "learning_rate": 5.815088952481478e-06,
      "loss": 0.4532,
      "step": 7331
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9553789496421814,
      "learning_rate": 5.814076666230978e-06,
      "loss": 0.5435,
      "step": 7332
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9789353609085083,
      "learning_rate": 5.813064345702542e-06,
      "loss": 0.5286,
      "step": 7333
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0638878345489502,
      "learning_rate": 5.812051990938794e-06,
      "loss": 0.5014,
      "step": 7334
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0748344659805298,
      "learning_rate": 5.811039601982363e-06,
      "loss": 0.5408,
      "step": 7335
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0366030931472778,
      "learning_rate": 5.810027178875875e-06,
      "loss": 0.5314,
      "step": 7336
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0968186855316162,
      "learning_rate": 5.809014721661961e-06,
      "loss": 0.5518,
      "step": 7337
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0389659404754639,
      "learning_rate": 5.808002230383249e-06,
      "loss": 0.5479,
      "step": 7338
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0004786252975464,
      "learning_rate": 5.806989705082377e-06,
      "loss": 0.4869,
      "step": 7339
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.078975796699524,
      "learning_rate": 5.805977145801975e-06,
      "loss": 0.5622,
      "step": 7340
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9257678985595703,
      "learning_rate": 5.8049645525846785e-06,
      "loss": 0.4662,
      "step": 7341
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9679080247879028,
      "learning_rate": 5.8039519254731245e-06,
      "loss": 0.505,
      "step": 7342
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9533823132514954,
      "learning_rate": 5.802939264509954e-06,
      "loss": 0.5402,
      "step": 7343
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9467368125915527,
      "learning_rate": 5.801926569737802e-06,
      "loss": 0.5128,
      "step": 7344
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.991810142993927,
      "learning_rate": 5.800913841199312e-06,
      "loss": 0.5508,
      "step": 7345
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0695257186889648,
      "learning_rate": 5.799901078937127e-06,
      "loss": 0.521,
      "step": 7346
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9737617373466492,
      "learning_rate": 5.798888282993891e-06,
      "loss": 0.4757,
      "step": 7347
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0166969299316406,
      "learning_rate": 5.7978754534122465e-06,
      "loss": 0.5163,
      "step": 7348
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9834796786308289,
      "learning_rate": 5.7968625902348445e-06,
      "loss": 0.5261,
      "step": 7349
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0013189315795898,
      "learning_rate": 5.7958496935043296e-06,
      "loss": 0.5152,
      "step": 7350
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9282286763191223,
      "learning_rate": 5.794836763263353e-06,
      "loss": 0.5351,
      "step": 7351
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0409108400344849,
      "learning_rate": 5.793823799554564e-06,
      "loss": 0.5005,
      "step": 7352
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9965145587921143,
      "learning_rate": 5.792810802420618e-06,
      "loss": 0.5648,
      "step": 7353
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9893680214881897,
      "learning_rate": 5.791797771904168e-06,
      "loss": 0.5249,
      "step": 7354
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1029082536697388,
      "learning_rate": 5.790784708047866e-06,
      "loss": 0.5759,
      "step": 7355
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0168418884277344,
      "learning_rate": 5.789771610894371e-06,
      "loss": 0.5042,
      "step": 7356
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0601954460144043,
      "learning_rate": 5.7887584804863414e-06,
      "loss": 0.5,
      "step": 7357
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.008843183517456,
      "learning_rate": 5.787745316866438e-06,
      "loss": 0.5029,
      "step": 7358
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9807408452033997,
      "learning_rate": 5.786732120077318e-06,
      "loss": 0.4781,
      "step": 7359
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0779192447662354,
      "learning_rate": 5.7857188901616444e-06,
      "loss": 0.5461,
      "step": 7360
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0375462770462036,
      "learning_rate": 5.7847056271620815e-06,
      "loss": 0.5284,
      "step": 7361
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0602962970733643,
      "learning_rate": 5.783692331121296e-06,
      "loss": 0.5502,
      "step": 7362
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1337302923202515,
      "learning_rate": 5.7826790020819525e-06,
      "loss": 0.5508,
      "step": 7363
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0568877458572388,
      "learning_rate": 5.781665640086719e-06,
      "loss": 0.4774,
      "step": 7364
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1909563541412354,
      "learning_rate": 5.780652245178263e-06,
      "loss": 0.5323,
      "step": 7365
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.220807433128357,
      "learning_rate": 5.779638817399259e-06,
      "loss": 0.4864,
      "step": 7366
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.078238606452942,
      "learning_rate": 5.778625356792376e-06,
      "loss": 0.5182,
      "step": 7367
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9133523106575012,
      "learning_rate": 5.7776118634002865e-06,
      "loss": 0.4301,
      "step": 7368
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0682060718536377,
      "learning_rate": 5.776598337265668e-06,
      "loss": 0.5082,
      "step": 7369
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0951640605926514,
      "learning_rate": 5.775584778431194e-06,
      "loss": 0.5636,
      "step": 7370
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0886175632476807,
      "learning_rate": 5.774571186939543e-06,
      "loss": 0.524,
      "step": 7371
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1518522500991821,
      "learning_rate": 5.773557562833394e-06,
      "loss": 0.5627,
      "step": 7372
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1411592960357666,
      "learning_rate": 5.772543906155429e-06,
      "loss": 0.5511,
      "step": 7373
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9977938532829285,
      "learning_rate": 5.7715302169483254e-06,
      "loss": 0.5045,
      "step": 7374
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0202271938323975,
      "learning_rate": 5.770516495254769e-06,
      "loss": 0.597,
      "step": 7375
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0154244899749756,
      "learning_rate": 5.769502741117443e-06,
      "loss": 0.564,
      "step": 7376
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.016966700553894,
      "learning_rate": 5.7684889545790346e-06,
      "loss": 0.5074,
      "step": 7377
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0249959230422974,
      "learning_rate": 5.767475135682228e-06,
      "loss": 0.5406,
      "step": 7378
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.005169153213501,
      "learning_rate": 5.766461284469714e-06,
      "loss": 0.5431,
      "step": 7379
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0918866395950317,
      "learning_rate": 5.765447400984182e-06,
      "loss": 0.5594,
      "step": 7380
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0976300239562988,
      "learning_rate": 5.7644334852683236e-06,
      "loss": 0.5582,
      "step": 7381
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0252267122268677,
      "learning_rate": 5.763419537364828e-06,
      "loss": 0.5008,
      "step": 7382
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0654020309448242,
      "learning_rate": 5.762405557316393e-06,
      "loss": 0.5281,
      "step": 7383
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9876992702484131,
      "learning_rate": 5.761391545165713e-06,
      "loss": 0.514,
      "step": 7384
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0419373512268066,
      "learning_rate": 5.760377500955483e-06,
      "loss": 0.5574,
      "step": 7385
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0095206499099731,
      "learning_rate": 5.759363424728401e-06,
      "loss": 0.5356,
      "step": 7386
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.118194580078125,
      "learning_rate": 5.758349316527166e-06,
      "loss": 0.5572,
      "step": 7387
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.035229206085205,
      "learning_rate": 5.7573351763944815e-06,
      "loss": 0.5283,
      "step": 7388
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.990778923034668,
      "learning_rate": 5.756321004373047e-06,
      "loss": 0.5052,
      "step": 7389
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0303928852081299,
      "learning_rate": 5.755306800505564e-06,
      "loss": 0.5484,
      "step": 7390
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0018138885498047,
      "learning_rate": 5.754292564834741e-06,
      "loss": 0.5501,
      "step": 7391
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1125024557113647,
      "learning_rate": 5.753278297403282e-06,
      "loss": 0.5509,
      "step": 7392
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9921433329582214,
      "learning_rate": 5.752263998253893e-06,
      "loss": 0.496,
      "step": 7393
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1274008750915527,
      "learning_rate": 5.751249667429285e-06,
      "loss": 0.5595,
      "step": 7394
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0307135581970215,
      "learning_rate": 5.7502353049721674e-06,
      "loss": 0.5318,
      "step": 7395
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9848186373710632,
      "learning_rate": 5.74922091092525e-06,
      "loss": 0.5003,
      "step": 7396
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0523111820220947,
      "learning_rate": 5.748206485331247e-06,
      "loss": 0.5423,
      "step": 7397
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9812511205673218,
      "learning_rate": 5.747192028232872e-06,
      "loss": 0.5282,
      "step": 7398
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0850028991699219,
      "learning_rate": 5.746177539672841e-06,
      "loss": 0.5266,
      "step": 7399
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.083706021308899,
      "learning_rate": 5.745163019693867e-06,
      "loss": 0.5036,
      "step": 7400
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9930241703987122,
      "learning_rate": 5.744148468338671e-06,
      "loss": 0.5442,
      "step": 7401
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1262766122817993,
      "learning_rate": 5.743133885649972e-06,
      "loss": 0.5537,
      "step": 7402
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.022183895111084,
      "learning_rate": 5.742119271670491e-06,
      "loss": 0.5127,
      "step": 7403
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.06358003616333,
      "learning_rate": 5.741104626442948e-06,
      "loss": 0.5542,
      "step": 7404
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9712068438529968,
      "learning_rate": 5.740089950010068e-06,
      "loss": 0.4641,
      "step": 7405
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9932650327682495,
      "learning_rate": 5.739075242414575e-06,
      "loss": 0.5319,
      "step": 7406
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.936883270740509,
      "learning_rate": 5.738060503699194e-06,
      "loss": 0.4861,
      "step": 7407
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0825799703598022,
      "learning_rate": 5.737045733906653e-06,
      "loss": 0.5237,
      "step": 7408
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1196141242980957,
      "learning_rate": 5.7360309330796805e-06,
      "loss": 0.5853,
      "step": 7409
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0770564079284668,
      "learning_rate": 5.735016101261005e-06,
      "loss": 0.5376,
      "step": 7410
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9987449049949646,
      "learning_rate": 5.7340012384933595e-06,
      "loss": 0.5465,
      "step": 7411
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1566431522369385,
      "learning_rate": 5.732986344819475e-06,
      "loss": 0.5475,
      "step": 7412
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9625303149223328,
      "learning_rate": 5.731971420282085e-06,
      "loss": 0.4451,
      "step": 7413
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0463659763336182,
      "learning_rate": 5.730956464923926e-06,
      "loss": 0.5262,
      "step": 7414
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0219457149505615,
      "learning_rate": 5.729941478787732e-06,
      "loss": 0.5347,
      "step": 7415
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.053633451461792,
      "learning_rate": 5.728926461916242e-06,
      "loss": 0.4918,
      "step": 7416
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0385435819625854,
      "learning_rate": 5.727911414352192e-06,
      "loss": 0.5032,
      "step": 7417
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.042497158050537,
      "learning_rate": 5.726896336138328e-06,
      "loss": 0.5376,
      "step": 7418
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.910614013671875,
      "learning_rate": 5.725881227317386e-06,
      "loss": 0.4965,
      "step": 7419
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9734965562820435,
      "learning_rate": 5.724866087932113e-06,
      "loss": 0.5052,
      "step": 7420
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.014836311340332,
      "learning_rate": 5.723850918025246e-06,
      "loss": 0.5158,
      "step": 7421
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0883235931396484,
      "learning_rate": 5.722835717639539e-06,
      "loss": 0.5289,
      "step": 7422
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.045098900794983,
      "learning_rate": 5.721820486817733e-06,
      "loss": 0.5294,
      "step": 7423
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9732691645622253,
      "learning_rate": 5.720805225602579e-06,
      "loss": 0.508,
      "step": 7424
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.999125063419342,
      "learning_rate": 5.719789934036821e-06,
      "loss": 0.5315,
      "step": 7425
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0546282529830933,
      "learning_rate": 5.718774612163216e-06,
      "loss": 0.5232,
      "step": 7426
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0071479082107544,
      "learning_rate": 5.717759260024511e-06,
      "loss": 0.5603,
      "step": 7427
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0235660076141357,
      "learning_rate": 5.716743877663462e-06,
      "loss": 0.5142,
      "step": 7428
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.014546275138855,
      "learning_rate": 5.715728465122821e-06,
      "loss": 0.5312,
      "step": 7429
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0218431949615479,
      "learning_rate": 5.714713022445344e-06,
      "loss": 0.5317,
      "step": 7430
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.017401099205017,
      "learning_rate": 5.713697549673788e-06,
      "loss": 0.5115,
      "step": 7431
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0051920413970947,
      "learning_rate": 5.712682046850909e-06,
      "loss": 0.5265,
      "step": 7432
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0701223611831665,
      "learning_rate": 5.711666514019472e-06,
      "loss": 0.5939,
      "step": 7433
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9435673356056213,
      "learning_rate": 5.710650951222231e-06,
      "loss": 0.5134,
      "step": 7434
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.01740562915802,
      "learning_rate": 5.709635358501952e-06,
      "loss": 0.52,
      "step": 7435
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0538522005081177,
      "learning_rate": 5.708619735901394e-06,
      "loss": 0.5062,
      "step": 7436
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0109902620315552,
      "learning_rate": 5.707604083463327e-06,
      "loss": 0.5181,
      "step": 7437
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0090477466583252,
      "learning_rate": 5.706588401230513e-06,
      "loss": 0.5117,
      "step": 7438
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0345077514648438,
      "learning_rate": 5.70557268924572e-06,
      "loss": 0.5434,
      "step": 7439
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9655662775039673,
      "learning_rate": 5.7045569475517126e-06,
      "loss": 0.4792,
      "step": 7440
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.113183617591858,
      "learning_rate": 5.703541176191266e-06,
      "loss": 0.5593,
      "step": 7441
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9703465104103088,
      "learning_rate": 5.702525375207147e-06,
      "loss": 0.4756,
      "step": 7442
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0538132190704346,
      "learning_rate": 5.70150954464213e-06,
      "loss": 0.4799,
      "step": 7443
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0759013891220093,
      "learning_rate": 5.700493684538984e-06,
      "loss": 0.531,
      "step": 7444
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0236819982528687,
      "learning_rate": 5.699477794940487e-06,
      "loss": 0.5299,
      "step": 7445
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0337623357772827,
      "learning_rate": 5.698461875889414e-06,
      "loss": 0.5307,
      "step": 7446
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9565263986587524,
      "learning_rate": 5.6974459274285395e-06,
      "loss": 0.5615,
      "step": 7447
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9748880863189697,
      "learning_rate": 5.696429949600643e-06,
      "loss": 0.5106,
      "step": 7448
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0070738792419434,
      "learning_rate": 5.695413942448505e-06,
      "loss": 0.5267,
      "step": 7449
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0051288604736328,
      "learning_rate": 5.694397906014907e-06,
      "loss": 0.516,
      "step": 7450
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.027597427368164,
      "learning_rate": 5.693381840342626e-06,
      "loss": 0.5177,
      "step": 7451
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.013757586479187,
      "learning_rate": 5.692365745474448e-06,
      "loss": 0.5277,
      "step": 7452
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9197251796722412,
      "learning_rate": 5.691349621453158e-06,
      "loss": 0.4893,
      "step": 7453
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9335641860961914,
      "learning_rate": 5.6903334683215416e-06,
      "loss": 0.4918,
      "step": 7454
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0981416702270508,
      "learning_rate": 5.689317286122382e-06,
      "loss": 0.5927,
      "step": 7455
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0974715948104858,
      "learning_rate": 5.68830107489847e-06,
      "loss": 0.5241,
      "step": 7456
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1248267889022827,
      "learning_rate": 5.687284834692595e-06,
      "loss": 0.5348,
      "step": 7457
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0581552982330322,
      "learning_rate": 5.686268565547547e-06,
      "loss": 0.5987,
      "step": 7458
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9802426695823669,
      "learning_rate": 5.685252267506116e-06,
      "loss": 0.5243,
      "step": 7459
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1204898357391357,
      "learning_rate": 5.6842359406110945e-06,
      "loss": 0.5413,
      "step": 7460
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9641440510749817,
      "learning_rate": 5.683219584905281e-06,
      "loss": 0.5212,
      "step": 7461
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.077943205833435,
      "learning_rate": 5.682203200431465e-06,
      "loss": 0.5271,
      "step": 7462
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0182640552520752,
      "learning_rate": 5.6811867872324465e-06,
      "loss": 0.5443,
      "step": 7463
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.078507661819458,
      "learning_rate": 5.680170345351021e-06,
      "loss": 0.5489,
      "step": 7464
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0855196714401245,
      "learning_rate": 5.67915387482999e-06,
      "loss": 0.5549,
      "step": 7465
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.047361135482788,
      "learning_rate": 5.678137375712152e-06,
      "loss": 0.5683,
      "step": 7466
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9580419659614563,
      "learning_rate": 5.6771208480403065e-06,
      "loss": 0.5284,
      "step": 7467
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9908188581466675,
      "learning_rate": 5.6761042918572585e-06,
      "loss": 0.5679,
      "step": 7468
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.076659083366394,
      "learning_rate": 5.675087707205811e-06,
      "loss": 0.5352,
      "step": 7469
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9653880000114441,
      "learning_rate": 5.674071094128768e-06,
      "loss": 0.4909,
      "step": 7470
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0040349960327148,
      "learning_rate": 5.673054452668936e-06,
      "loss": 0.4924,
      "step": 7471
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9688589572906494,
      "learning_rate": 5.672037782869123e-06,
      "loss": 0.5142,
      "step": 7472
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.058455228805542,
      "learning_rate": 5.671021084772137e-06,
      "loss": 0.5664,
      "step": 7473
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.992531418800354,
      "learning_rate": 5.670004358420786e-06,
      "loss": 0.5211,
      "step": 7474
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0130715370178223,
      "learning_rate": 5.668987603857884e-06,
      "loss": 0.5056,
      "step": 7475
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9969527125358582,
      "learning_rate": 5.6679708211262415e-06,
      "loss": 0.5038,
      "step": 7476
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9759190082550049,
      "learning_rate": 5.66695401026867e-06,
      "loss": 0.527,
      "step": 7477
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0427945852279663,
      "learning_rate": 5.665937171327985e-06,
      "loss": 0.5542,
      "step": 7478
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9948601126670837,
      "learning_rate": 5.664920304347004e-06,
      "loss": 0.5351,
      "step": 7479
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0073400735855103,
      "learning_rate": 5.6639034093685416e-06,
      "loss": 0.5366,
      "step": 7480
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0551327466964722,
      "learning_rate": 5.662886486435415e-06,
      "loss": 0.5293,
      "step": 7481
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9984437823295593,
      "learning_rate": 5.6618695355904456e-06,
      "loss": 0.521,
      "step": 7482
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9802277088165283,
      "learning_rate": 5.660852556876452e-06,
      "loss": 0.5206,
      "step": 7483
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9552007913589478,
      "learning_rate": 5.659835550336257e-06,
      "loss": 0.4878,
      "step": 7484
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9671801924705505,
      "learning_rate": 5.658818516012681e-06,
      "loss": 0.5358,
      "step": 7485
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.019771695137024,
      "learning_rate": 5.65780145394855e-06,
      "loss": 0.5134,
      "step": 7486
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0270360708236694,
      "learning_rate": 5.656784364186687e-06,
      "loss": 0.5547,
      "step": 7487
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9477678537368774,
      "learning_rate": 5.655767246769921e-06,
      "loss": 0.5307,
      "step": 7488
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9901880621910095,
      "learning_rate": 5.6547501017410765e-06,
      "loss": 0.4953,
      "step": 7489
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0108751058578491,
      "learning_rate": 5.6537329291429835e-06,
      "loss": 0.5092,
      "step": 7490
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1157495975494385,
      "learning_rate": 5.652715729018471e-06,
      "loss": 0.4955,
      "step": 7491
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0335301160812378,
      "learning_rate": 5.65169850141037e-06,
      "loss": 0.5353,
      "step": 7492
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9454318881034851,
      "learning_rate": 5.650681246361511e-06,
      "loss": 0.5018,
      "step": 7493
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.953421950340271,
      "learning_rate": 5.649663963914729e-06,
      "loss": 0.5033,
      "step": 7494
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0138174295425415,
      "learning_rate": 5.6486466541128575e-06,
      "loss": 0.5348,
      "step": 7495
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.002591848373413,
      "learning_rate": 5.6476293169987314e-06,
      "loss": 0.5153,
      "step": 7496
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0464341640472412,
      "learning_rate": 5.646611952615188e-06,
      "loss": 0.5146,
      "step": 7497
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0779297351837158,
      "learning_rate": 5.645594561005064e-06,
      "loss": 0.5705,
      "step": 7498
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0479854345321655,
      "learning_rate": 5.6445771422112005e-06,
      "loss": 0.4992,
      "step": 7499
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1008824110031128,
      "learning_rate": 5.643559696276435e-06,
      "loss": 0.5279,
      "step": 7500
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9838743805885315,
      "learning_rate": 5.6425422232436085e-06,
      "loss": 0.4645,
      "step": 7501
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0333632230758667,
      "learning_rate": 5.641524723155566e-06,
      "loss": 0.5329,
      "step": 7502
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0836952924728394,
      "learning_rate": 5.6405071960551485e-06,
      "loss": 0.4969,
      "step": 7503
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0006016492843628,
      "learning_rate": 5.639489641985201e-06,
      "loss": 0.5532,
      "step": 7504
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1209015846252441,
      "learning_rate": 5.638472060988569e-06,
      "loss": 0.5947,
      "step": 7505
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.001816749572754,
      "learning_rate": 5.6374544531081e-06,
      "loss": 0.4769,
      "step": 7506
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0426093339920044,
      "learning_rate": 5.636436818386641e-06,
      "loss": 0.5528,
      "step": 7507
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9544576406478882,
      "learning_rate": 5.635419156867043e-06,
      "loss": 0.4907,
      "step": 7508
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9945775866508484,
      "learning_rate": 5.634401468592152e-06,
      "loss": 0.5223,
      "step": 7509
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0476137399673462,
      "learning_rate": 5.6333837536048255e-06,
      "loss": 0.4948,
      "step": 7510
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.964775800704956,
      "learning_rate": 5.63236601194791e-06,
      "loss": 0.4961,
      "step": 7511
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9926450848579407,
      "learning_rate": 5.631348243664263e-06,
      "loss": 0.5498,
      "step": 7512
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.049108862876892,
      "learning_rate": 5.630330448796736e-06,
      "loss": 0.5514,
      "step": 7513
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0942082405090332,
      "learning_rate": 5.629312627388188e-06,
      "loss": 0.5991,
      "step": 7514
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0420613288879395,
      "learning_rate": 5.628294779481474e-06,
      "loss": 0.5274,
      "step": 7515
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0404502153396606,
      "learning_rate": 5.6272769051194535e-06,
      "loss": 0.6015,
      "step": 7516
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0156911611557007,
      "learning_rate": 5.626259004344983e-06,
      "loss": 0.5138,
      "step": 7517
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0107665061950684,
      "learning_rate": 5.625241077200926e-06,
      "loss": 0.5352,
      "step": 7518
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0115772485733032,
      "learning_rate": 5.624223123730141e-06,
      "loss": 0.5144,
      "step": 7519
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.98555588722229,
      "learning_rate": 5.6232051439754935e-06,
      "loss": 0.524,
      "step": 7520
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0191307067871094,
      "learning_rate": 5.622187137979843e-06,
      "loss": 0.4849,
      "step": 7521
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0475789308547974,
      "learning_rate": 5.621169105786057e-06,
      "loss": 0.5079,
      "step": 7522
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0014190673828125,
      "learning_rate": 5.620151047437004e-06,
      "loss": 0.5712,
      "step": 7523
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9675139784812927,
      "learning_rate": 5.619132962975544e-06,
      "loss": 0.5339,
      "step": 7524
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.063788890838623,
      "learning_rate": 5.6181148524445506e-06,
      "loss": 0.5195,
      "step": 7525
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0479037761688232,
      "learning_rate": 5.617096715886889e-06,
      "loss": 0.532,
      "step": 7526
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.13522469997406,
      "learning_rate": 5.616078553345434e-06,
      "loss": 0.5687,
      "step": 7527
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0124516487121582,
      "learning_rate": 5.615060364863053e-06,
      "loss": 0.521,
      "step": 7528
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0758651494979858,
      "learning_rate": 5.6140421504826205e-06,
      "loss": 0.5213,
      "step": 7529
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0261516571044922,
      "learning_rate": 5.6130239102470075e-06,
      "loss": 0.5589,
      "step": 7530
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0139498710632324,
      "learning_rate": 5.612005644199092e-06,
      "loss": 0.5123,
      "step": 7531
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9417591691017151,
      "learning_rate": 5.610987352381747e-06,
      "loss": 0.5109,
      "step": 7532
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1065438985824585,
      "learning_rate": 5.60996903483785e-06,
      "loss": 0.5689,
      "step": 7533
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.055302381515503,
      "learning_rate": 5.608950691610279e-06,
      "loss": 0.526,
      "step": 7534
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0546671152114868,
      "learning_rate": 5.607932322741912e-06,
      "loss": 0.5658,
      "step": 7535
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0092540979385376,
      "learning_rate": 5.60691392827563e-06,
      "loss": 0.5359,
      "step": 7536
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9884355068206787,
      "learning_rate": 5.605895508254315e-06,
      "loss": 0.4899,
      "step": 7537
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0973706245422363,
      "learning_rate": 5.604877062720848e-06,
      "loss": 0.5866,
      "step": 7538
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.958016037940979,
      "learning_rate": 5.603858591718111e-06,
      "loss": 0.5111,
      "step": 7539
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0794581174850464,
      "learning_rate": 5.602840095288989e-06,
      "loss": 0.5572,
      "step": 7540
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0130765438079834,
      "learning_rate": 5.6018215734763685e-06,
      "loss": 0.5632,
      "step": 7541
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0051946640014648,
      "learning_rate": 5.600803026323136e-06,
      "loss": 0.4771,
      "step": 7542
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.996223509311676,
      "learning_rate": 5.599784453872177e-06,
      "loss": 0.523,
      "step": 7543
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0174285173416138,
      "learning_rate": 5.5987658561663805e-06,
      "loss": 0.5544,
      "step": 7544
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9483619332313538,
      "learning_rate": 5.597747233248637e-06,
      "loss": 0.5308,
      "step": 7545
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.006107211112976,
      "learning_rate": 5.596728585161838e-06,
      "loss": 0.5132,
      "step": 7546
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.967338502407074,
      "learning_rate": 5.595709911948873e-06,
      "loss": 0.5632,
      "step": 7547
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0335711240768433,
      "learning_rate": 5.5946912136526365e-06,
      "loss": 0.5433,
      "step": 7548
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.052869439125061,
      "learning_rate": 5.593672490316022e-06,
      "loss": 0.5595,
      "step": 7549
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0248395204544067,
      "learning_rate": 5.5926537419819234e-06,
      "loss": 0.5632,
      "step": 7550
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0395309925079346,
      "learning_rate": 5.591634968693238e-06,
      "loss": 0.5384,
      "step": 7551
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1257261037826538,
      "learning_rate": 5.590616170492862e-06,
      "loss": 0.5595,
      "step": 7552
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0620990991592407,
      "learning_rate": 5.589597347423696e-06,
      "loss": 0.5428,
      "step": 7553
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0337821245193481,
      "learning_rate": 5.588578499528633e-06,
      "loss": 0.5271,
      "step": 7554
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9734302163124084,
      "learning_rate": 5.587559626850578e-06,
      "loss": 0.5032,
      "step": 7555
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0158849954605103,
      "learning_rate": 5.586540729432431e-06,
      "loss": 0.5196,
      "step": 7556
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.018912434577942,
      "learning_rate": 5.585521807317097e-06,
      "loss": 0.589,
      "step": 7557
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.00540030002594,
      "learning_rate": 5.584502860547474e-06,
      "loss": 0.5473,
      "step": 7558
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.018280029296875,
      "learning_rate": 5.5834838891664685e-06,
      "loss": 0.5522,
      "step": 7559
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0685259103775024,
      "learning_rate": 5.582464893216987e-06,
      "loss": 0.5125,
      "step": 7560
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.96529620885849,
      "learning_rate": 5.5814458727419365e-06,
      "loss": 0.5196,
      "step": 7561
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.99607253074646,
      "learning_rate": 5.580426827784221e-06,
      "loss": 0.5406,
      "step": 7562
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0173125267028809,
      "learning_rate": 5.579407758386751e-06,
      "loss": 0.5668,
      "step": 7563
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.13144850730896,
      "learning_rate": 5.578388664592435e-06,
      "loss": 0.5369,
      "step": 7564
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9925999045372009,
      "learning_rate": 5.577369546444188e-06,
      "loss": 0.549,
      "step": 7565
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0043359994888306,
      "learning_rate": 5.576350403984915e-06,
      "loss": 0.5658,
      "step": 7566
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9701277613639832,
      "learning_rate": 5.575331237257532e-06,
      "loss": 0.5118,
      "step": 7567
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1391804218292236,
      "learning_rate": 5.574312046304954e-06,
      "loss": 0.5759,
      "step": 7568
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0357731580734253,
      "learning_rate": 5.5732928311700906e-06,
      "loss": 0.5158,
      "step": 7569
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.991754412651062,
      "learning_rate": 5.5722735918958614e-06,
      "loss": 0.5159,
      "step": 7570
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1032490730285645,
      "learning_rate": 5.571254328525183e-06,
      "loss": 0.5228,
      "step": 7571
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0968703031539917,
      "learning_rate": 5.570235041100972e-06,
      "loss": 0.5555,
      "step": 7572
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.003477692604065,
      "learning_rate": 5.569215729666146e-06,
      "loss": 0.529,
      "step": 7573
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0698657035827637,
      "learning_rate": 5.568196394263626e-06,
      "loss": 0.5357,
      "step": 7574
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9531198143959045,
      "learning_rate": 5.567177034936333e-06,
      "loss": 0.4848,
      "step": 7575
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.00896418094635,
      "learning_rate": 5.566157651727189e-06,
      "loss": 0.5829,
      "step": 7576
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9287537932395935,
      "learning_rate": 5.5651382446791134e-06,
      "loss": 0.5138,
      "step": 7577
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0205477476119995,
      "learning_rate": 5.564118813835033e-06,
      "loss": 0.517,
      "step": 7578
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9876137375831604,
      "learning_rate": 5.563099359237872e-06,
      "loss": 0.522,
      "step": 7579
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0472246408462524,
      "learning_rate": 5.5620798809305575e-06,
      "loss": 0.5319,
      "step": 7580
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.028066873550415,
      "learning_rate": 5.561060378956014e-06,
      "loss": 0.5435,
      "step": 7581
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9206072092056274,
      "learning_rate": 5.560040853357168e-06,
      "loss": 0.5235,
      "step": 7582
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1040226221084595,
      "learning_rate": 5.55902130417695e-06,
      "loss": 0.552,
      "step": 7583
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0034788846969604,
      "learning_rate": 5.558001731458293e-06,
      "loss": 0.5498,
      "step": 7584
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.090221643447876,
      "learning_rate": 5.556982135244121e-06,
      "loss": 0.5496,
      "step": 7585
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0335744619369507,
      "learning_rate": 5.5559625155773685e-06,
      "loss": 0.5127,
      "step": 7586
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9486215710639954,
      "learning_rate": 5.554942872500971e-06,
      "loss": 0.506,
      "step": 7587
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9887097477912903,
      "learning_rate": 5.5539232060578574e-06,
      "loss": 0.5326,
      "step": 7588
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9414849281311035,
      "learning_rate": 5.552903516290966e-06,
      "loss": 0.5164,
      "step": 7589
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9639526605606079,
      "learning_rate": 5.55188380324323e-06,
      "loss": 0.5163,
      "step": 7590
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0122463703155518,
      "learning_rate": 5.550864066957587e-06,
      "loss": 0.5385,
      "step": 7591
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0007060766220093,
      "learning_rate": 5.549844307476975e-06,
      "loss": 0.5596,
      "step": 7592
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0017606019973755,
      "learning_rate": 5.548824524844333e-06,
      "loss": 0.5603,
      "step": 7593
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0292510986328125,
      "learning_rate": 5.547804719102596e-06,
      "loss": 0.4978,
      "step": 7594
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.049460530281067,
      "learning_rate": 5.546784890294712e-06,
      "loss": 0.5703,
      "step": 7595
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0576456785202026,
      "learning_rate": 5.545765038463615e-06,
      "loss": 0.5687,
      "step": 7596
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.138731598854065,
      "learning_rate": 5.544745163652253e-06,
      "loss": 0.5332,
      "step": 7597
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0252635478973389,
      "learning_rate": 5.543725265903565e-06,
      "loss": 0.5096,
      "step": 7598
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0496039390563965,
      "learning_rate": 5.5427053452605004e-06,
      "loss": 0.5393,
      "step": 7599
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.97575443983078,
      "learning_rate": 5.541685401766001e-06,
      "loss": 0.5464,
      "step": 7600
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.069856882095337,
      "learning_rate": 5.540665435463013e-06,
      "loss": 0.4764,
      "step": 7601
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9672538042068481,
      "learning_rate": 5.539645446394485e-06,
      "loss": 0.5317,
      "step": 7602
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.077881932258606,
      "learning_rate": 5.538625434603363e-06,
      "loss": 0.5585,
      "step": 7603
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1011637449264526,
      "learning_rate": 5.5376054001326e-06,
      "loss": 0.5458,
      "step": 7604
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9810386896133423,
      "learning_rate": 5.53658534302514e-06,
      "loss": 0.5768,
      "step": 7605
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9890307784080505,
      "learning_rate": 5.535565263323942e-06,
      "loss": 0.4688,
      "step": 7606
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0282771587371826,
      "learning_rate": 5.534545161071951e-06,
      "loss": 0.5566,
      "step": 7607
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9736272692680359,
      "learning_rate": 5.533525036312125e-06,
      "loss": 0.5212,
      "step": 7608
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0575153827667236,
      "learning_rate": 5.532504889087413e-06,
      "loss": 0.5358,
      "step": 7609
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.133524775505066,
      "learning_rate": 5.531484719440776e-06,
      "loss": 0.5454,
      "step": 7610
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0144935846328735,
      "learning_rate": 5.530464527415164e-06,
      "loss": 0.4708,
      "step": 7611
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1847485303878784,
      "learning_rate": 5.529444313053538e-06,
      "loss": 0.5269,
      "step": 7612
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0995430946350098,
      "learning_rate": 5.528424076398851e-06,
      "loss": 0.5347,
      "step": 7613
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9704165458679199,
      "learning_rate": 5.527403817494067e-06,
      "loss": 0.5345,
      "step": 7614
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0518720149993896,
      "learning_rate": 5.526383536382142e-06,
      "loss": 0.4638,
      "step": 7615
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0203839540481567,
      "learning_rate": 5.525363233106037e-06,
      "loss": 0.5295,
      "step": 7616
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.931056022644043,
      "learning_rate": 5.524342907708714e-06,
      "loss": 0.5434,
      "step": 7617
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9502934217453003,
      "learning_rate": 5.5233225602331355e-06,
      "loss": 0.4904,
      "step": 7618
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.056097149848938,
      "learning_rate": 5.522302190722264e-06,
      "loss": 0.5118,
      "step": 7619
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9836977124214172,
      "learning_rate": 5.5212817992190644e-06,
      "loss": 0.5366,
      "step": 7620
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0775401592254639,
      "learning_rate": 5.5202613857665025e-06,
      "loss": 0.5572,
      "step": 7621
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9739736914634705,
      "learning_rate": 5.5192409504075416e-06,
      "loss": 0.5008,
      "step": 7622
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0487024784088135,
      "learning_rate": 5.518220493185153e-06,
      "loss": 0.5212,
      "step": 7623
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0653893947601318,
      "learning_rate": 5.517200014142301e-06,
      "loss": 0.5076,
      "step": 7624
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0490169525146484,
      "learning_rate": 5.516179513321955e-06,
      "loss": 0.5149,
      "step": 7625
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0116205215454102,
      "learning_rate": 5.5151589907670856e-06,
      "loss": 0.5025,
      "step": 7626
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0689759254455566,
      "learning_rate": 5.514138446520664e-06,
      "loss": 0.5473,
      "step": 7627
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9866931438446045,
      "learning_rate": 5.51311788062566e-06,
      "loss": 0.5215,
      "step": 7628
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9698858261108398,
      "learning_rate": 5.512097293125047e-06,
      "loss": 0.4779,
      "step": 7629
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0759952068328857,
      "learning_rate": 5.511076684061799e-06,
      "loss": 0.5025,
      "step": 7630
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.093837022781372,
      "learning_rate": 5.51005605347889e-06,
      "loss": 0.5604,
      "step": 7631
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0192079544067383,
      "learning_rate": 5.509035401419296e-06,
      "loss": 0.4878,
      "step": 7632
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0973546504974365,
      "learning_rate": 5.50801472792599e-06,
      "loss": 0.5226,
      "step": 7633
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9534756541252136,
      "learning_rate": 5.5069940330419525e-06,
      "loss": 0.4877,
      "step": 7634
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0228980779647827,
      "learning_rate": 5.5059733168101596e-06,
      "loss": 0.5558,
      "step": 7635
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0513986349105835,
      "learning_rate": 5.504952579273589e-06,
      "loss": 0.497,
      "step": 7636
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0697723627090454,
      "learning_rate": 5.503931820475223e-06,
      "loss": 0.5738,
      "step": 7637
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9767886996269226,
      "learning_rate": 5.502911040458042e-06,
      "loss": 0.5528,
      "step": 7638
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9490932822227478,
      "learning_rate": 5.501890239265025e-06,
      "loss": 0.5065,
      "step": 7639
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.006232500076294,
      "learning_rate": 5.500869416939156e-06,
      "loss": 0.5653,
      "step": 7640
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.022708773612976,
      "learning_rate": 5.49984857352342e-06,
      "loss": 0.5219,
      "step": 7641
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.036411166191101,
      "learning_rate": 5.4988277090607986e-06,
      "loss": 0.4857,
      "step": 7642
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.144761562347412,
      "learning_rate": 5.4978068235942775e-06,
      "loss": 0.5774,
      "step": 7643
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1377125978469849,
      "learning_rate": 5.496785917166843e-06,
      "loss": 0.537,
      "step": 7644
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0362329483032227,
      "learning_rate": 5.49576498982148e-06,
      "loss": 0.5017,
      "step": 7645
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9684587121009827,
      "learning_rate": 5.49474404160118e-06,
      "loss": 0.4949,
      "step": 7646
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0577255487442017,
      "learning_rate": 5.4937230725489285e-06,
      "loss": 0.5589,
      "step": 7647
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.99142986536026,
      "learning_rate": 5.492702082707716e-06,
      "loss": 0.5267,
      "step": 7648
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0030139684677124,
      "learning_rate": 5.491681072120534e-06,
      "loss": 0.5461,
      "step": 7649
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0612186193466187,
      "learning_rate": 5.4906600408303715e-06,
      "loss": 0.5408,
      "step": 7650
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0308854579925537,
      "learning_rate": 5.489638988880222e-06,
      "loss": 0.5,
      "step": 7651
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9938170909881592,
      "learning_rate": 5.488617916313077e-06,
      "loss": 0.4949,
      "step": 7652
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9927681684494019,
      "learning_rate": 5.487596823171932e-06,
      "loss": 0.4685,
      "step": 7653
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9878483414649963,
      "learning_rate": 5.486575709499782e-06,
      "loss": 0.5507,
      "step": 7654
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0441032648086548,
      "learning_rate": 5.48555457533962e-06,
      "loss": 0.5432,
      "step": 7655
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9428449869155884,
      "learning_rate": 5.484533420734444e-06,
      "loss": 0.494,
      "step": 7656
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.008828043937683,
      "learning_rate": 5.483512245727252e-06,
      "loss": 0.5541,
      "step": 7657
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9554852247238159,
      "learning_rate": 5.482491050361041e-06,
      "loss": 0.5062,
      "step": 7658
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9811081886291504,
      "learning_rate": 5.48146983467881e-06,
      "loss": 0.5395,
      "step": 7659
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1356273889541626,
      "learning_rate": 5.480448598723559e-06,
      "loss": 0.5582,
      "step": 7660
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0313886404037476,
      "learning_rate": 5.47942734253829e-06,
      "loss": 0.5369,
      "step": 7661
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0309735536575317,
      "learning_rate": 5.478406066166003e-06,
      "loss": 0.4946,
      "step": 7662
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0279486179351807,
      "learning_rate": 5.477384769649701e-06,
      "loss": 0.5469,
      "step": 7663
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0054008960723877,
      "learning_rate": 5.476363453032387e-06,
      "loss": 0.5179,
      "step": 7664
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8933504223823547,
      "learning_rate": 5.475342116357064e-06,
      "loss": 0.547,
      "step": 7665
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.058522343635559,
      "learning_rate": 5.474320759666739e-06,
      "loss": 0.5159,
      "step": 7666
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.122342824935913,
      "learning_rate": 5.473299383004417e-06,
      "loss": 0.5809,
      "step": 7667
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0336556434631348,
      "learning_rate": 5.472277986413104e-06,
      "loss": 0.4835,
      "step": 7668
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9952534437179565,
      "learning_rate": 5.471256569935809e-06,
      "loss": 0.5157,
      "step": 7669
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0328069925308228,
      "learning_rate": 5.470235133615538e-06,
      "loss": 0.5165,
      "step": 7670
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0147979259490967,
      "learning_rate": 5.4692136774953004e-06,
      "loss": 0.5189,
      "step": 7671
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0270156860351562,
      "learning_rate": 5.46819220161811e-06,
      "loss": 0.5101,
      "step": 7672
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9940213561058044,
      "learning_rate": 5.467170706026973e-06,
      "loss": 0.4904,
      "step": 7673
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0557513236999512,
      "learning_rate": 5.466149190764902e-06,
      "loss": 0.5072,
      "step": 7674
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0868325233459473,
      "learning_rate": 5.465127655874911e-06,
      "loss": 0.5929,
      "step": 7675
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.991359293460846,
      "learning_rate": 5.464106101400013e-06,
      "loss": 0.5201,
      "step": 7676
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9942566752433777,
      "learning_rate": 5.463084527383222e-06,
      "loss": 0.5014,
      "step": 7677
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9597623944282532,
      "learning_rate": 5.4620629338675505e-06,
      "loss": 0.501,
      "step": 7678
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0973620414733887,
      "learning_rate": 5.461041320896019e-06,
      "loss": 0.5523,
      "step": 7679
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0960867404937744,
      "learning_rate": 5.460019688511639e-06,
      "loss": 0.5053,
      "step": 7680
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9990050196647644,
      "learning_rate": 5.458998036757431e-06,
      "loss": 0.5863,
      "step": 7681
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0510493516921997,
      "learning_rate": 5.4579763656764115e-06,
      "loss": 0.5487,
      "step": 7682
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0192859172821045,
      "learning_rate": 5.456954675311602e-06,
      "loss": 0.5118,
      "step": 7683
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.112775444984436,
      "learning_rate": 5.45593296570602e-06,
      "loss": 0.5487,
      "step": 7684
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.047751784324646,
      "learning_rate": 5.454911236902687e-06,
      "loss": 0.561,
      "step": 7685
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1452687978744507,
      "learning_rate": 5.453889488944623e-06,
      "loss": 0.4948,
      "step": 7686
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0465176105499268,
      "learning_rate": 5.452867721874854e-06,
      "loss": 0.5825,
      "step": 7687
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.027136206626892,
      "learning_rate": 5.4518459357364e-06,
      "loss": 0.5623,
      "step": 7688
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9339892864227295,
      "learning_rate": 5.4508241305722856e-06,
      "loss": 0.5346,
      "step": 7689
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0112874507904053,
      "learning_rate": 5.449802306425532e-06,
      "loss": 0.496,
      "step": 7690
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0279783010482788,
      "learning_rate": 5.448780463339172e-06,
      "loss": 0.5555,
      "step": 7691
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0531165599822998,
      "learning_rate": 5.447758601356226e-06,
      "loss": 0.5527,
      "step": 7692
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.04063880443573,
      "learning_rate": 5.446736720519725e-06,
      "loss": 0.5314,
      "step": 7693
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0666983127593994,
      "learning_rate": 5.445714820872693e-06,
      "loss": 0.5058,
      "step": 7694
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9285444617271423,
      "learning_rate": 5.4446929024581606e-06,
      "loss": 0.4899,
      "step": 7695
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0237271785736084,
      "learning_rate": 5.4436709653191575e-06,
      "loss": 0.5197,
      "step": 7696
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9723763465881348,
      "learning_rate": 5.442649009498713e-06,
      "loss": 0.5023,
      "step": 7697
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9989533424377441,
      "learning_rate": 5.441627035039859e-06,
      "loss": 0.5363,
      "step": 7698
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0461578369140625,
      "learning_rate": 5.440605041985626e-06,
      "loss": 0.4827,
      "step": 7699
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.984526515007019,
      "learning_rate": 5.439583030379049e-06,
      "loss": 0.4742,
      "step": 7700
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0510518550872803,
      "learning_rate": 5.438561000263157e-06,
      "loss": 0.4966,
      "step": 7701
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0577447414398193,
      "learning_rate": 5.4375389516809895e-06,
      "loss": 0.5713,
      "step": 7702
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0560086965560913,
      "learning_rate": 5.436516884675579e-06,
      "loss": 0.5584,
      "step": 7703
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.076753854751587,
      "learning_rate": 5.43549479928996e-06,
      "loss": 0.5296,
      "step": 7704
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9767786860466003,
      "learning_rate": 5.434472695567169e-06,
      "loss": 0.4962,
      "step": 7705
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9865174889564514,
      "learning_rate": 5.433450573550246e-06,
      "loss": 0.5034,
      "step": 7706
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0956785678863525,
      "learning_rate": 5.432428433282226e-06,
      "loss": 0.5249,
      "step": 7707
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0960456132888794,
      "learning_rate": 5.43140627480615e-06,
      "loss": 0.5717,
      "step": 7708
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9936690926551819,
      "learning_rate": 5.4303840981650565e-06,
      "loss": 0.4751,
      "step": 7709
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0348507165908813,
      "learning_rate": 5.429361903401985e-06,
      "loss": 0.5432,
      "step": 7710
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0245975255966187,
      "learning_rate": 5.4283396905599785e-06,
      "loss": 0.5356,
      "step": 7711
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0867342948913574,
      "learning_rate": 5.427317459682076e-06,
      "loss": 0.5326,
      "step": 7712
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9968048334121704,
      "learning_rate": 5.426295210811323e-06,
      "loss": 0.5311,
      "step": 7713
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9450915455818176,
      "learning_rate": 5.425272943990761e-06,
      "loss": 0.503,
      "step": 7714
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1199814081192017,
      "learning_rate": 5.4242506592634354e-06,
      "loss": 0.5206,
      "step": 7715
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0570131540298462,
      "learning_rate": 5.423228356672391e-06,
      "loss": 0.4955,
      "step": 7716
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9853617548942566,
      "learning_rate": 5.422206036260671e-06,
      "loss": 0.5114,
      "step": 7717
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9816334247589111,
      "learning_rate": 5.421183698071325e-06,
      "loss": 0.5552,
      "step": 7718
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.049434781074524,
      "learning_rate": 5.420161342147399e-06,
      "loss": 0.5132,
      "step": 7719
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0932750701904297,
      "learning_rate": 5.4191389685319395e-06,
      "loss": 0.5444,
      "step": 7720
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9652199149131775,
      "learning_rate": 5.4181165772679955e-06,
      "loss": 0.5003,
      "step": 7721
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9985790252685547,
      "learning_rate": 5.417094168398618e-06,
      "loss": 0.5291,
      "step": 7722
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0097248554229736,
      "learning_rate": 5.416071741966856e-06,
      "loss": 0.5368,
      "step": 7723
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0083731412887573,
      "learning_rate": 5.41504929801576e-06,
      "loss": 0.5365,
      "step": 7724
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9989120364189148,
      "learning_rate": 5.414026836588382e-06,
      "loss": 0.5189,
      "step": 7725
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9898789525032043,
      "learning_rate": 5.413004357727775e-06,
      "loss": 0.5192,
      "step": 7726
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0092414617538452,
      "learning_rate": 5.411981861476991e-06,
      "loss": 0.4877,
      "step": 7727
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.991629958152771,
      "learning_rate": 5.4109593478790825e-06,
      "loss": 0.4838,
      "step": 7728
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0284823179244995,
      "learning_rate": 5.409936816977106e-06,
      "loss": 0.5843,
      "step": 7729
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9995983839035034,
      "learning_rate": 5.408914268814117e-06,
      "loss": 0.5284,
      "step": 7730
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0161820650100708,
      "learning_rate": 5.4078917034331705e-06,
      "loss": 0.4845,
      "step": 7731
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0265183448791504,
      "learning_rate": 5.4068691208773225e-06,
      "loss": 0.4619,
      "step": 7732
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.013788104057312,
      "learning_rate": 5.405846521189632e-06,
      "loss": 0.5413,
      "step": 7733
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.008409857749939,
      "learning_rate": 5.404823904413157e-06,
      "loss": 0.5253,
      "step": 7734
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.005803108215332,
      "learning_rate": 5.403801270590955e-06,
      "loss": 0.5186,
      "step": 7735
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.036096453666687,
      "learning_rate": 5.402778619766086e-06,
      "loss": 0.5444,
      "step": 7736
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.055092453956604,
      "learning_rate": 5.40175595198161e-06,
      "loss": 0.5386,
      "step": 7737
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0218758583068848,
      "learning_rate": 5.400733267280589e-06,
      "loss": 0.5209,
      "step": 7738
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9987779259681702,
      "learning_rate": 5.399710565706084e-06,
      "loss": 0.5508,
      "step": 7739
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9801649451255798,
      "learning_rate": 5.3986878473011585e-06,
      "loss": 0.5269,
      "step": 7740
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9721705317497253,
      "learning_rate": 5.397665112108874e-06,
      "loss": 0.4668,
      "step": 7741
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9974575638771057,
      "learning_rate": 5.3966423601722955e-06,
      "loss": 0.5104,
      "step": 7742
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.067785382270813,
      "learning_rate": 5.3956195915344855e-06,
      "loss": 0.5503,
      "step": 7743
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.03829026222229,
      "learning_rate": 5.394596806238511e-06,
      "loss": 0.552,
      "step": 7744
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.060089349746704,
      "learning_rate": 5.39357400432744e-06,
      "loss": 0.4677,
      "step": 7745
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0661898851394653,
      "learning_rate": 5.392551185844334e-06,
      "loss": 0.5309,
      "step": 7746
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0674138069152832,
      "learning_rate": 5.391528350832265e-06,
      "loss": 0.5272,
      "step": 7747
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.078151822090149,
      "learning_rate": 5.3905054993342985e-06,
      "loss": 0.5373,
      "step": 7748
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0482679605484009,
      "learning_rate": 5.389482631393504e-06,
      "loss": 0.552,
      "step": 7749
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.006722092628479,
      "learning_rate": 5.388459747052951e-06,
      "loss": 0.5065,
      "step": 7750
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.036423921585083,
      "learning_rate": 5.387436846355709e-06,
      "loss": 0.5223,
      "step": 7751
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0063308477401733,
      "learning_rate": 5.386413929344849e-06,
      "loss": 0.5187,
      "step": 7752
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0499755144119263,
      "learning_rate": 5.3853909960634446e-06,
      "loss": 0.474,
      "step": 7753
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.002184271812439,
      "learning_rate": 5.3843680465545635e-06,
      "loss": 0.5435,
      "step": 7754
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0605528354644775,
      "learning_rate": 5.3833450808612816e-06,
      "loss": 0.5474,
      "step": 7755
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9395731091499329,
      "learning_rate": 5.382322099026673e-06,
      "loss": 0.5307,
      "step": 7756
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9526917934417725,
      "learning_rate": 5.38129910109381e-06,
      "loss": 0.5048,
      "step": 7757
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0066355466842651,
      "learning_rate": 5.380276087105769e-06,
      "loss": 0.5045,
      "step": 7758
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9914907217025757,
      "learning_rate": 5.379253057105623e-06,
      "loss": 0.5049,
      "step": 7759
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0168482065200806,
      "learning_rate": 5.378230011136453e-06,
      "loss": 0.5707,
      "step": 7760
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1721969842910767,
      "learning_rate": 5.37720694924133e-06,
      "loss": 0.4713,
      "step": 7761
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9489725828170776,
      "learning_rate": 5.376183871463336e-06,
      "loss": 0.4936,
      "step": 7762
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0816004276275635,
      "learning_rate": 5.375160777845548e-06,
      "loss": 0.5248,
      "step": 7763
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0063161849975586,
      "learning_rate": 5.3741376684310455e-06,
      "loss": 0.5209,
      "step": 7764
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9822127223014832,
      "learning_rate": 5.3731145432629065e-06,
      "loss": 0.529,
      "step": 7765
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0101993083953857,
      "learning_rate": 5.3720914023842105e-06,
      "loss": 0.5136,
      "step": 7766
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0614701509475708,
      "learning_rate": 5.371068245838042e-06,
      "loss": 0.5653,
      "step": 7767
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9922329187393188,
      "learning_rate": 5.37004507366748e-06,
      "loss": 0.5353,
      "step": 7768
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0334720611572266,
      "learning_rate": 5.369021885915607e-06,
      "loss": 0.5286,
      "step": 7769
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0917425155639648,
      "learning_rate": 5.367998682625506e-06,
      "loss": 0.5455,
      "step": 7770
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9876338243484497,
      "learning_rate": 5.366975463840262e-06,
      "loss": 0.4676,
      "step": 7771
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0127947330474854,
      "learning_rate": 5.365952229602956e-06,
      "loss": 0.4777,
      "step": 7772
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0870912075042725,
      "learning_rate": 5.3649289799566766e-06,
      "loss": 0.5259,
      "step": 7773
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9871841669082642,
      "learning_rate": 5.363905714944505e-06,
      "loss": 0.5157,
      "step": 7774
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1193543672561646,
      "learning_rate": 5.362882434609531e-06,
      "loss": 0.5634,
      "step": 7775
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0339303016662598,
      "learning_rate": 5.36185913899484e-06,
      "loss": 0.5139,
      "step": 7776
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0583903789520264,
      "learning_rate": 5.36083582814352e-06,
      "loss": 0.5469,
      "step": 7777
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0183539390563965,
      "learning_rate": 5.359812502098657e-06,
      "loss": 0.5168,
      "step": 7778
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0020984411239624,
      "learning_rate": 5.358789160903343e-06,
      "loss": 0.5192,
      "step": 7779
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.016532063484192,
      "learning_rate": 5.357765804600664e-06,
      "loss": 0.5113,
      "step": 7780
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0904549360275269,
      "learning_rate": 5.3567424332337125e-06,
      "loss": 0.5276,
      "step": 7781
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.070608377456665,
      "learning_rate": 5.355719046845577e-06,
      "loss": 0.5316,
      "step": 7782
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0256794691085815,
      "learning_rate": 5.354695645479352e-06,
      "loss": 0.5148,
      "step": 7783
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0035301446914673,
      "learning_rate": 5.353672229178125e-06,
      "loss": 0.5005,
      "step": 7784
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0000191926956177,
      "learning_rate": 5.352648797984993e-06,
      "loss": 0.5198,
      "step": 7785
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0152604579925537,
      "learning_rate": 5.351625351943044e-06,
      "loss": 0.5477,
      "step": 7786
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9971880316734314,
      "learning_rate": 5.350601891095377e-06,
      "loss": 0.52,
      "step": 7787
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9151439070701599,
      "learning_rate": 5.349578415485085e-06,
      "loss": 0.4491,
      "step": 7788
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0387840270996094,
      "learning_rate": 5.34855492515526e-06,
      "loss": 0.5188,
      "step": 7789
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0091477632522583,
      "learning_rate": 5.347531420148999e-06,
      "loss": 0.5585,
      "step": 7790
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.001717209815979,
      "learning_rate": 5.3465079005094e-06,
      "loss": 0.4615,
      "step": 7791
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0388224124908447,
      "learning_rate": 5.34548436627956e-06,
      "loss": 0.5465,
      "step": 7792
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0266194343566895,
      "learning_rate": 5.344460817502573e-06,
      "loss": 0.5628,
      "step": 7793
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9543954730033875,
      "learning_rate": 5.34343725422154e-06,
      "loss": 0.5228,
      "step": 7794
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9868411421775818,
      "learning_rate": 5.342413676479559e-06,
      "loss": 0.5379,
      "step": 7795
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0662660598754883,
      "learning_rate": 5.34139008431973e-06,
      "loss": 0.5012,
      "step": 7796
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9981074333190918,
      "learning_rate": 5.34036647778515e-06,
      "loss": 0.5613,
      "step": 7797
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9298437833786011,
      "learning_rate": 5.3393428569189235e-06,
      "loss": 0.489,
      "step": 7798
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.973460853099823,
      "learning_rate": 5.338319221764149e-06,
      "loss": 0.509,
      "step": 7799
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9743896126747131,
      "learning_rate": 5.33729557236393e-06,
      "loss": 0.5396,
      "step": 7800
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9878454804420471,
      "learning_rate": 5.336271908761367e-06,
      "loss": 0.5333,
      "step": 7801
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.89174884557724,
      "learning_rate": 5.335248230999565e-06,
      "loss": 0.4602,
      "step": 7802
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1391502618789673,
      "learning_rate": 5.334224539121625e-06,
      "loss": 0.5299,
      "step": 7803
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9654562473297119,
      "learning_rate": 5.333200833170652e-06,
      "loss": 0.489,
      "step": 7804
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0083357095718384,
      "learning_rate": 5.332177113189751e-06,
      "loss": 0.5592,
      "step": 7805
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0468686819076538,
      "learning_rate": 5.331153379222028e-06,
      "loss": 0.55,
      "step": 7806
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0573049783706665,
      "learning_rate": 5.330129631310589e-06,
      "loss": 0.5602,
      "step": 7807
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0374951362609863,
      "learning_rate": 5.3291058694985385e-06,
      "loss": 0.5454,
      "step": 7808
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.983371376991272,
      "learning_rate": 5.328082093828984e-06,
      "loss": 0.5491,
      "step": 7809
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0157434940338135,
      "learning_rate": 5.327058304345035e-06,
      "loss": 0.5071,
      "step": 7810
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0032756328582764,
      "learning_rate": 5.3260345010898e-06,
      "loss": 0.5273,
      "step": 7811
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0313307046890259,
      "learning_rate": 5.325010684106384e-06,
      "loss": 0.514,
      "step": 7812
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.046082854270935,
      "learning_rate": 5.323986853437899e-06,
      "loss": 0.4815,
      "step": 7813
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1349530220031738,
      "learning_rate": 5.322963009127454e-06,
      "loss": 0.5488,
      "step": 7814
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0096465349197388,
      "learning_rate": 5.321939151218163e-06,
      "loss": 0.5026,
      "step": 7815
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0394556522369385,
      "learning_rate": 5.320915279753132e-06,
      "loss": 0.5339,
      "step": 7816
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9634128212928772,
      "learning_rate": 5.319891394775475e-06,
      "loss": 0.4668,
      "step": 7817
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2029508352279663,
      "learning_rate": 5.3188674963283064e-06,
      "loss": 0.5302,
      "step": 7818
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0494053363800049,
      "learning_rate": 5.317843584454734e-06,
      "loss": 0.4982,
      "step": 7819
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0254759788513184,
      "learning_rate": 5.316819659197875e-06,
      "loss": 0.5218,
      "step": 7820
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0763325691223145,
      "learning_rate": 5.315795720600842e-06,
      "loss": 0.5358,
      "step": 7821
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9637579917907715,
      "learning_rate": 5.314771768706751e-06,
      "loss": 0.5269,
      "step": 7822
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.058788776397705,
      "learning_rate": 5.313747803558714e-06,
      "loss": 0.5326,
      "step": 7823
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9709265828132629,
      "learning_rate": 5.312723825199849e-06,
      "loss": 0.5564,
      "step": 7824
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.089979648590088,
      "learning_rate": 5.311699833673273e-06,
      "loss": 0.5756,
      "step": 7825
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0571144819259644,
      "learning_rate": 5.310675829022101e-06,
      "loss": 0.499,
      "step": 7826
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0553808212280273,
      "learning_rate": 5.309651811289449e-06,
      "loss": 0.5218,
      "step": 7827
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0515599250793457,
      "learning_rate": 5.308627780518437e-06,
      "loss": 0.5052,
      "step": 7828
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0557936429977417,
      "learning_rate": 5.307603736752183e-06,
      "loss": 0.4942,
      "step": 7829
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9929682612419128,
      "learning_rate": 5.306579680033807e-06,
      "loss": 0.5244,
      "step": 7830
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0507532358169556,
      "learning_rate": 5.305555610406425e-06,
      "loss": 0.5383,
      "step": 7831
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.044619083404541,
      "learning_rate": 5.30453152791316e-06,
      "loss": 0.6029,
      "step": 7832
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9375192523002625,
      "learning_rate": 5.303507432597134e-06,
      "loss": 0.5224,
      "step": 7833
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9598952531814575,
      "learning_rate": 5.302483324501463e-06,
      "loss": 0.5442,
      "step": 7834
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9932873845100403,
      "learning_rate": 5.3014592036692715e-06,
      "loss": 0.5299,
      "step": 7835
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.100043773651123,
      "learning_rate": 5.300435070143683e-06,
      "loss": 0.5309,
      "step": 7836
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0243855714797974,
      "learning_rate": 5.2994109239678185e-06,
      "loss": 0.5571,
      "step": 7837
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9657698273658752,
      "learning_rate": 5.298386765184801e-06,
      "loss": 0.5374,
      "step": 7838
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9283653497695923,
      "learning_rate": 5.297362593837755e-06,
      "loss": 0.4895,
      "step": 7839
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0488038063049316,
      "learning_rate": 5.296338409969805e-06,
      "loss": 0.4973,
      "step": 7840
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9657244682312012,
      "learning_rate": 5.295314213624076e-06,
      "loss": 0.5156,
      "step": 7841
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9581754207611084,
      "learning_rate": 5.2942900048436914e-06,
      "loss": 0.4893,
      "step": 7842
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.078657865524292,
      "learning_rate": 5.293265783671778e-06,
      "loss": 0.5152,
      "step": 7843
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9514079689979553,
      "learning_rate": 5.292241550151465e-06,
      "loss": 0.5408,
      "step": 7844
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1053167581558228,
      "learning_rate": 5.291217304325875e-06,
      "loss": 0.5604,
      "step": 7845
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0533970594406128,
      "learning_rate": 5.290193046238139e-06,
      "loss": 0.5402,
      "step": 7846
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0341068506240845,
      "learning_rate": 5.289168775931381e-06,
      "loss": 0.5277,
      "step": 7847
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9752747416496277,
      "learning_rate": 5.288144493448733e-06,
      "loss": 0.5015,
      "step": 7848
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.034314513206482,
      "learning_rate": 5.287120198833324e-06,
      "loss": 0.5076,
      "step": 7849
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0576122999191284,
      "learning_rate": 5.286095892128282e-06,
      "loss": 0.5438,
      "step": 7850
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.103843331336975,
      "learning_rate": 5.285071573376735e-06,
      "loss": 0.5414,
      "step": 7851
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0938315391540527,
      "learning_rate": 5.2840472426218185e-06,
      "loss": 0.5247,
      "step": 7852
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0951156616210938,
      "learning_rate": 5.283022899906659e-06,
      "loss": 0.5565,
      "step": 7853
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.049530267715454,
      "learning_rate": 5.28199854527439e-06,
      "loss": 0.5209,
      "step": 7854
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0040522813796997,
      "learning_rate": 5.280974178768144e-06,
      "loss": 0.4782,
      "step": 7855
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3000999689102173,
      "learning_rate": 5.279949800431052e-06,
      "loss": 0.5432,
      "step": 7856
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0034246444702148,
      "learning_rate": 5.278925410306248e-06,
      "loss": 0.5224,
      "step": 7857
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9741222858428955,
      "learning_rate": 5.277901008436865e-06,
      "loss": 0.5138,
      "step": 7858
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0248442888259888,
      "learning_rate": 5.276876594866037e-06,
      "loss": 0.5259,
      "step": 7859
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0938154458999634,
      "learning_rate": 5.2758521696369e-06,
      "loss": 0.5056,
      "step": 7860
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0054714679718018,
      "learning_rate": 5.274827732792587e-06,
      "loss": 0.4986,
      "step": 7861
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0986206531524658,
      "learning_rate": 5.273803284376234e-06,
      "loss": 0.5365,
      "step": 7862
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1783478260040283,
      "learning_rate": 5.272778824430977e-06,
      "loss": 0.5317,
      "step": 7863
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0371789932250977,
      "learning_rate": 5.271754352999953e-06,
      "loss": 0.5394,
      "step": 7864
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9669798612594604,
      "learning_rate": 5.2707298701263e-06,
      "loss": 0.5093,
      "step": 7865
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0162286758422852,
      "learning_rate": 5.269705375853151e-06,
      "loss": 0.5385,
      "step": 7866
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0095473527908325,
      "learning_rate": 5.26868087022365e-06,
      "loss": 0.5413,
      "step": 7867
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0174813270568848,
      "learning_rate": 5.26765635328093e-06,
      "loss": 0.5052,
      "step": 7868
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1494885683059692,
      "learning_rate": 5.266631825068134e-06,
      "loss": 0.5739,
      "step": 7869
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.041465163230896,
      "learning_rate": 5.265607285628397e-06,
      "loss": 0.5501,
      "step": 7870
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9956937432289124,
      "learning_rate": 5.264582735004863e-06,
      "loss": 0.5237,
      "step": 7871
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9891913533210754,
      "learning_rate": 5.26355817324067e-06,
      "loss": 0.5737,
      "step": 7872
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.93519127368927,
      "learning_rate": 5.26253360037896e-06,
      "loss": 0.547,
      "step": 7873
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0138989686965942,
      "learning_rate": 5.2615090164628705e-06,
      "loss": 0.5397,
      "step": 7874
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.075402021408081,
      "learning_rate": 5.2604844215355484e-06,
      "loss": 0.4961,
      "step": 7875
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0560067892074585,
      "learning_rate": 5.259459815640133e-06,
      "loss": 0.5353,
      "step": 7876
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.066972255706787,
      "learning_rate": 5.258435198819768e-06,
      "loss": 0.5427,
      "step": 7877
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0539740324020386,
      "learning_rate": 5.257410571117594e-06,
      "loss": 0.5777,
      "step": 7878
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9533175230026245,
      "learning_rate": 5.256385932576759e-06,
      "loss": 0.5012,
      "step": 7879
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1580781936645508,
      "learning_rate": 5.255361283240402e-06,
      "loss": 0.4792,
      "step": 7880
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0315744876861572,
      "learning_rate": 5.254336623151672e-06,
      "loss": 0.5604,
      "step": 7881
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9291620850563049,
      "learning_rate": 5.253311952353708e-06,
      "loss": 0.5167,
      "step": 7882
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1018723249435425,
      "learning_rate": 5.252287270889661e-06,
      "loss": 0.5823,
      "step": 7883
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9712012410163879,
      "learning_rate": 5.251262578802675e-06,
      "loss": 0.5012,
      "step": 7884
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0360691547393799,
      "learning_rate": 5.250237876135895e-06,
      "loss": 0.5469,
      "step": 7885
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.002305507659912,
      "learning_rate": 5.2492131629324695e-06,
      "loss": 0.4992,
      "step": 7886
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0135037899017334,
      "learning_rate": 5.248188439235544e-06,
      "loss": 0.5125,
      "step": 7887
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.004456877708435,
      "learning_rate": 5.247163705088267e-06,
      "loss": 0.509,
      "step": 7888
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0056803226470947,
      "learning_rate": 5.246138960533786e-06,
      "loss": 0.5611,
      "step": 7889
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0572956800460815,
      "learning_rate": 5.245114205615249e-06,
      "loss": 0.4943,
      "step": 7890
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.105042576789856,
      "learning_rate": 5.244089440375807e-06,
      "loss": 0.5174,
      "step": 7891
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9719240665435791,
      "learning_rate": 5.243064664858607e-06,
      "loss": 0.4899,
      "step": 7892
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0321283340454102,
      "learning_rate": 5.242039879106799e-06,
      "loss": 0.5553,
      "step": 7893
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0477603673934937,
      "learning_rate": 5.241015083163534e-06,
      "loss": 0.6115,
      "step": 7894
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9469429850578308,
      "learning_rate": 5.239990277071962e-06,
      "loss": 0.551,
      "step": 7895
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0114669799804688,
      "learning_rate": 5.238965460875236e-06,
      "loss": 0.5004,
      "step": 7896
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9960516691207886,
      "learning_rate": 5.237940634616504e-06,
      "loss": 0.5269,
      "step": 7897
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.985478937625885,
      "learning_rate": 5.2369157983389205e-06,
      "loss": 0.5471,
      "step": 7898
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9954918026924133,
      "learning_rate": 5.235890952085637e-06,
      "loss": 0.503,
      "step": 7899
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.015634536743164,
      "learning_rate": 5.234866095899806e-06,
      "loss": 0.5127,
      "step": 7900
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0375806093215942,
      "learning_rate": 5.23384122982458e-06,
      "loss": 0.5291,
      "step": 7901
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1058557033538818,
      "learning_rate": 5.232816353903113e-06,
      "loss": 0.5343,
      "step": 7902
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9224012494087219,
      "learning_rate": 5.231791468178561e-06,
      "loss": 0.4541,
      "step": 7903
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0257867574691772,
      "learning_rate": 5.230766572694075e-06,
      "loss": 0.5085,
      "step": 7904
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.092187523841858,
      "learning_rate": 5.229741667492811e-06,
      "loss": 0.5596,
      "step": 7905
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0415695905685425,
      "learning_rate": 5.228716752617926e-06,
      "loss": 0.524,
      "step": 7906
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0576292276382446,
      "learning_rate": 5.2276918281125744e-06,
      "loss": 0.5467,
      "step": 7907
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9662138223648071,
      "learning_rate": 5.22666689401991e-06,
      "loss": 0.5185,
      "step": 7908
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0126597881317139,
      "learning_rate": 5.225641950383094e-06,
      "loss": 0.5587,
      "step": 7909
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9800564646720886,
      "learning_rate": 5.2246169972452775e-06,
      "loss": 0.5487,
      "step": 7910
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0493261814117432,
      "learning_rate": 5.223592034649624e-06,
      "loss": 0.4969,
      "step": 7911
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0116685628890991,
      "learning_rate": 5.2225670626392845e-06,
      "loss": 0.4894,
      "step": 7912
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0772850513458252,
      "learning_rate": 5.221542081257421e-06,
      "loss": 0.5611,
      "step": 7913
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0064325332641602,
      "learning_rate": 5.220517090547194e-06,
      "loss": 0.499,
      "step": 7914
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0418468713760376,
      "learning_rate": 5.219492090551757e-06,
      "loss": 0.5412,
      "step": 7915
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0826489925384521,
      "learning_rate": 5.21846708131427e-06,
      "loss": 0.5556,
      "step": 7916
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1069979667663574,
      "learning_rate": 5.217442062877897e-06,
      "loss": 0.5309,
      "step": 7917
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9450262188911438,
      "learning_rate": 5.216417035285795e-06,
      "loss": 0.5483,
      "step": 7918
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9797537922859192,
      "learning_rate": 5.215391998581123e-06,
      "loss": 0.525,
      "step": 7919
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9857979416847229,
      "learning_rate": 5.214366952807043e-06,
      "loss": 0.5083,
      "step": 7920
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9631537795066833,
      "learning_rate": 5.213341898006718e-06,
      "loss": 0.4687,
      "step": 7921
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0123926401138306,
      "learning_rate": 5.212316834223307e-06,
      "loss": 0.5513,
      "step": 7922
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9671790599822998,
      "learning_rate": 5.211291761499973e-06,
      "loss": 0.4902,
      "step": 7923
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0273030996322632,
      "learning_rate": 5.210266679879877e-06,
      "loss": 0.4917,
      "step": 7924
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.070820689201355,
      "learning_rate": 5.209241589406183e-06,
      "loss": 0.5306,
      "step": 7925
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9976992607116699,
      "learning_rate": 5.208216490122055e-06,
      "loss": 0.5001,
      "step": 7926
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0579338073730469,
      "learning_rate": 5.207191382070653e-06,
      "loss": 0.518,
      "step": 7927
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0654932260513306,
      "learning_rate": 5.206166265295143e-06,
      "loss": 0.5334,
      "step": 7928
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.046481728553772,
      "learning_rate": 5.205141139838691e-06,
      "loss": 0.4989,
      "step": 7929
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9963456988334656,
      "learning_rate": 5.204116005744456e-06,
      "loss": 0.5171,
      "step": 7930
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0503910779953003,
      "learning_rate": 5.2030908630556075e-06,
      "loss": 0.5327,
      "step": 7931
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0344712734222412,
      "learning_rate": 5.202065711815309e-06,
      "loss": 0.4963,
      "step": 7932
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9059045314788818,
      "learning_rate": 5.201040552066727e-06,
      "loss": 0.4543,
      "step": 7933
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0495176315307617,
      "learning_rate": 5.200015383853026e-06,
      "loss": 0.5298,
      "step": 7934
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9795430898666382,
      "learning_rate": 5.1989902072173735e-06,
      "loss": 0.5515,
      "step": 7935
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0523420572280884,
      "learning_rate": 5.197965022202935e-06,
      "loss": 0.5265,
      "step": 7936
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.996130645275116,
      "learning_rate": 5.196939828852879e-06,
      "loss": 0.5299,
      "step": 7937
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9796398878097534,
      "learning_rate": 5.195914627210372e-06,
      "loss": 0.5059,
      "step": 7938
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9897952675819397,
      "learning_rate": 5.19488941731858e-06,
      "loss": 0.541,
      "step": 7939
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9556361436843872,
      "learning_rate": 5.193864199220674e-06,
      "loss": 0.4893,
      "step": 7940
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.047682523727417,
      "learning_rate": 5.192838972959821e-06,
      "loss": 0.5366,
      "step": 7941
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0672297477722168,
      "learning_rate": 5.19181373857919e-06,
      "loss": 0.5293,
      "step": 7942
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9972975850105286,
      "learning_rate": 5.190788496121948e-06,
      "loss": 0.522,
      "step": 7943
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9845316410064697,
      "learning_rate": 5.189763245631268e-06,
      "loss": 0.4624,
      "step": 7944
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0090800523757935,
      "learning_rate": 5.188737987150316e-06,
      "loss": 0.548,
      "step": 7945
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9909870028495789,
      "learning_rate": 5.1877127207222666e-06,
      "loss": 0.502,
      "step": 7946
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0190807580947876,
      "learning_rate": 5.186687446390284e-06,
      "loss": 0.5031,
      "step": 7947
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0893406867980957,
      "learning_rate": 5.185662164197546e-06,
      "loss": 0.5275,
      "step": 7948
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0079681873321533,
      "learning_rate": 5.184636874187218e-06,
      "loss": 0.5121,
      "step": 7949
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0926753282546997,
      "learning_rate": 5.183611576402474e-06,
      "loss": 0.5142,
      "step": 7950
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.982645571231842,
      "learning_rate": 5.182586270886485e-06,
      "loss": 0.5325,
      "step": 7951
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.001665711402893,
      "learning_rate": 5.181560957682423e-06,
      "loss": 0.5317,
      "step": 7952
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9564476609230042,
      "learning_rate": 5.180535636833462e-06,
      "loss": 0.4977,
      "step": 7953
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9797195792198181,
      "learning_rate": 5.179510308382773e-06,
      "loss": 0.5138,
      "step": 7954
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0832334756851196,
      "learning_rate": 5.178484972373528e-06,
      "loss": 0.555,
      "step": 7955
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.011929988861084,
      "learning_rate": 5.177459628848903e-06,
      "loss": 0.5478,
      "step": 7956
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.05528724193573,
      "learning_rate": 5.17643427785207e-06,
      "loss": 0.5586,
      "step": 7957
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9753866791725159,
      "learning_rate": 5.175408919426204e-06,
      "loss": 0.5098,
      "step": 7958
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.102591633796692,
      "learning_rate": 5.174383553614478e-06,
      "loss": 0.485,
      "step": 7959
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9618451595306396,
      "learning_rate": 5.1733581804600674e-06,
      "loss": 0.528,
      "step": 7960
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.053686499595642,
      "learning_rate": 5.172332800006147e-06,
      "loss": 0.5683,
      "step": 7961
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.02312433719635,
      "learning_rate": 5.171307412295892e-06,
      "loss": 0.5234,
      "step": 7962
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9336697459220886,
      "learning_rate": 5.1702820173724766e-06,
      "loss": 0.4938,
      "step": 7963
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1132444143295288,
      "learning_rate": 5.169256615279078e-06,
      "loss": 0.5379,
      "step": 7964
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.061083436012268,
      "learning_rate": 5.168231206058874e-06,
      "loss": 0.497,
      "step": 7965
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9309837818145752,
      "learning_rate": 5.167205789755037e-06,
      "loss": 0.483,
      "step": 7966
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0379348993301392,
      "learning_rate": 5.1661803664107465e-06,
      "loss": 0.5018,
      "step": 7967
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9954214096069336,
      "learning_rate": 5.16515493606918e-06,
      "loss": 0.5241,
      "step": 7968
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9716700315475464,
      "learning_rate": 5.164129498773513e-06,
      "loss": 0.5286,
      "step": 7969
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0255597829818726,
      "learning_rate": 5.163104054566922e-06,
      "loss": 0.5416,
      "step": 7970
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0864416360855103,
      "learning_rate": 5.16207860349259e-06,
      "loss": 0.5285,
      "step": 7971
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9192246794700623,
      "learning_rate": 5.16105314559369e-06,
      "loss": 0.4966,
      "step": 7972
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0215632915496826,
      "learning_rate": 5.160027680913402e-06,
      "loss": 0.5171,
      "step": 7973
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9981522560119629,
      "learning_rate": 5.159002209494905e-06,
      "loss": 0.5831,
      "step": 7974
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0376025438308716,
      "learning_rate": 5.157976731381379e-06,
      "loss": 0.5321,
      "step": 7975
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1233857870101929,
      "learning_rate": 5.1569512466160025e-06,
      "loss": 0.5601,
      "step": 7976
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0849857330322266,
      "learning_rate": 5.155925755241954e-06,
      "loss": 0.5672,
      "step": 7977
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9773584008216858,
      "learning_rate": 5.1549002573024144e-06,
      "loss": 0.4966,
      "step": 7978
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0027586221694946,
      "learning_rate": 5.153874752840564e-06,
      "loss": 0.5014,
      "step": 7979
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0651005506515503,
      "learning_rate": 5.152849241899585e-06,
      "loss": 0.5096,
      "step": 7980
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1254360675811768,
      "learning_rate": 5.151823724522653e-06,
      "loss": 0.5215,
      "step": 7981
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9797729849815369,
      "learning_rate": 5.150798200752953e-06,
      "loss": 0.5305,
      "step": 7982
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.045276165008545,
      "learning_rate": 5.149772670633666e-06,
      "loss": 0.5164,
      "step": 7983
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.019736409187317,
      "learning_rate": 5.148747134207974e-06,
      "loss": 0.4887,
      "step": 7984
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.034010410308838,
      "learning_rate": 5.147721591519056e-06,
      "loss": 0.5115,
      "step": 7985
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0028609037399292,
      "learning_rate": 5.146696042610095e-06,
      "loss": 0.4889,
      "step": 7986
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.123826503753662,
      "learning_rate": 5.145670487524276e-06,
      "loss": 0.5456,
      "step": 7987
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.024096965789795,
      "learning_rate": 5.144644926304778e-06,
      "loss": 0.5716,
      "step": 7988
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.121075987815857,
      "learning_rate": 5.1436193589947855e-06,
      "loss": 0.516,
      "step": 7989
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1426475048065186,
      "learning_rate": 5.1425937856374816e-06,
      "loss": 0.5672,
      "step": 7990
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0405372381210327,
      "learning_rate": 5.141568206276051e-06,
      "loss": 0.5529,
      "step": 7991
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0194613933563232,
      "learning_rate": 5.140542620953675e-06,
      "loss": 0.4764,
      "step": 7992
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1488856077194214,
      "learning_rate": 5.139517029713537e-06,
      "loss": 0.5922,
      "step": 7993
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.020204782485962,
      "learning_rate": 5.138491432598822e-06,
      "loss": 0.5008,
      "step": 7994
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0580018758773804,
      "learning_rate": 5.137465829652716e-06,
      "loss": 0.5173,
      "step": 7995
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0638741254806519,
      "learning_rate": 5.136440220918401e-06,
      "loss": 0.5124,
      "step": 7996
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.088953971862793,
      "learning_rate": 5.135414606439063e-06,
      "loss": 0.5547,
      "step": 7997
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0534359216690063,
      "learning_rate": 5.134388986257887e-06,
      "loss": 0.5194,
      "step": 7998
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9983291625976562,
      "learning_rate": 5.133363360418059e-06,
      "loss": 0.5332,
      "step": 7999
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.041108250617981,
      "learning_rate": 5.132337728962763e-06,
      "loss": 0.5423,
      "step": 8000
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0383859872817993,
      "learning_rate": 5.131312091935186e-06,
      "loss": 0.5261,
      "step": 8001
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0309644937515259,
      "learning_rate": 5.130286449378513e-06,
      "loss": 0.5788,
      "step": 8002
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0815945863723755,
      "learning_rate": 5.129260801335932e-06,
      "loss": 0.5365,
      "step": 8003
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.018567442893982,
      "learning_rate": 5.128235147850629e-06,
      "loss": 0.5332,
      "step": 8004
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9869685769081116,
      "learning_rate": 5.127209488965787e-06,
      "loss": 0.5147,
      "step": 8005
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.973550021648407,
      "learning_rate": 5.1261838247246e-06,
      "loss": 0.4759,
      "step": 8006
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0792498588562012,
      "learning_rate": 5.125158155170248e-06,
      "loss": 0.514,
      "step": 8007
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0632809400558472,
      "learning_rate": 5.124132480345922e-06,
      "loss": 0.4947,
      "step": 8008
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0588916540145874,
      "learning_rate": 5.123106800294809e-06,
      "loss": 0.5306,
      "step": 8009
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0726609230041504,
      "learning_rate": 5.122081115060098e-06,
      "loss": 0.5241,
      "step": 8010
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0798453092575073,
      "learning_rate": 5.121055424684975e-06,
      "loss": 0.4875,
      "step": 8011
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0017881393432617,
      "learning_rate": 5.12002972921263e-06,
      "loss": 0.5088,
      "step": 8012
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0158437490463257,
      "learning_rate": 5.119004028686249e-06,
      "loss": 0.5124,
      "step": 8013
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9737855195999146,
      "learning_rate": 5.117978323149025e-06,
      "loss": 0.5514,
      "step": 8014
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.107811450958252,
      "learning_rate": 5.116952612644141e-06,
      "loss": 0.4671,
      "step": 8015
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1360200643539429,
      "learning_rate": 5.1159268972147915e-06,
      "loss": 0.5371,
      "step": 8016
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0516135692596436,
      "learning_rate": 5.114901176904164e-06,
      "loss": 0.5106,
      "step": 8017
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0169031620025635,
      "learning_rate": 5.113875451755447e-06,
      "loss": 0.5439,
      "step": 8018
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9697759747505188,
      "learning_rate": 5.11284972181183e-06,
      "loss": 0.4929,
      "step": 8019
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0336737632751465,
      "learning_rate": 5.111823987116504e-06,
      "loss": 0.514,
      "step": 8020
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.02176034450531,
      "learning_rate": 5.110798247712661e-06,
      "loss": 0.5333,
      "step": 8021
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.036531925201416,
      "learning_rate": 5.109772503643486e-06,
      "loss": 0.5656,
      "step": 8022
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9949049949645996,
      "learning_rate": 5.108746754952177e-06,
      "loss": 0.4844,
      "step": 8023
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0902539491653442,
      "learning_rate": 5.107721001681915e-06,
      "loss": 0.5508,
      "step": 8024
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0021989345550537,
      "learning_rate": 5.1066952438759e-06,
      "loss": 0.4943,
      "step": 8025
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0186973810195923,
      "learning_rate": 5.105669481577319e-06,
      "loss": 0.541,
      "step": 8026
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0388914346694946,
      "learning_rate": 5.104643714829362e-06,
      "loss": 0.5638,
      "step": 8027
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9377147555351257,
      "learning_rate": 5.103617943675224e-06,
      "loss": 0.5213,
      "step": 8028
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0112721920013428,
      "learning_rate": 5.102592168158095e-06,
      "loss": 0.531,
      "step": 8029
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0095853805541992,
      "learning_rate": 5.101566388321165e-06,
      "loss": 0.5159,
      "step": 8030
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9937832951545715,
      "learning_rate": 5.100540604207629e-06,
      "loss": 0.51,
      "step": 8031
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.125806450843811,
      "learning_rate": 5.099514815860678e-06,
      "loss": 0.5244,
      "step": 8032
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9696172475814819,
      "learning_rate": 5.098489023323504e-06,
      "loss": 0.4601,
      "step": 8033
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0181834697723389,
      "learning_rate": 5.0974632266393e-06,
      "loss": 0.5248,
      "step": 8034
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0194844007492065,
      "learning_rate": 5.0964374258512585e-06,
      "loss": 0.5156,
      "step": 8035
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1436933279037476,
      "learning_rate": 5.0954116210025725e-06,
      "loss": 0.5529,
      "step": 8036
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0950855016708374,
      "learning_rate": 5.094385812136435e-06,
      "loss": 0.5164,
      "step": 8037
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.067351222038269,
      "learning_rate": 5.09335999929604e-06,
      "loss": 0.5117,
      "step": 8038
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0079240798950195,
      "learning_rate": 5.092334182524578e-06,
      "loss": 0.5118,
      "step": 8039
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1458498239517212,
      "learning_rate": 5.091308361865247e-06,
      "loss": 0.5854,
      "step": 8040
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9869314432144165,
      "learning_rate": 5.090282537361237e-06,
      "loss": 0.4858,
      "step": 8041
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0945738554000854,
      "learning_rate": 5.089256709055745e-06,
      "loss": 0.5682,
      "step": 8042
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0837138891220093,
      "learning_rate": 5.088230876991962e-06,
      "loss": 0.522,
      "step": 8043
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0137262344360352,
      "learning_rate": 5.087205041213085e-06,
      "loss": 0.5351,
      "step": 8044
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1037157773971558,
      "learning_rate": 5.086179201762306e-06,
      "loss": 0.5433,
      "step": 8045
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0440733432769775,
      "learning_rate": 5.085153358682822e-06,
      "loss": 0.5122,
      "step": 8046
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0042705535888672,
      "learning_rate": 5.084127512017823e-06,
      "loss": 0.5224,
      "step": 8047
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9671124219894409,
      "learning_rate": 5.083101661810511e-06,
      "loss": 0.5237,
      "step": 8048
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.04704749584198,
      "learning_rate": 5.082075808104075e-06,
      "loss": 0.5165,
      "step": 8049
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0171769857406616,
      "learning_rate": 5.081049950941713e-06,
      "loss": 0.5362,
      "step": 8050
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0862504243850708,
      "learning_rate": 5.080024090366618e-06,
      "loss": 0.528,
      "step": 8051
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0223110914230347,
      "learning_rate": 5.078998226421989e-06,
      "loss": 0.5447,
      "step": 8052
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0615509748458862,
      "learning_rate": 5.07797235915102e-06,
      "loss": 0.5355,
      "step": 8053
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9888608455657959,
      "learning_rate": 5.076946488596905e-06,
      "loss": 0.5384,
      "step": 8054
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1075698137283325,
      "learning_rate": 5.07592061480284e-06,
      "loss": 0.5596,
      "step": 8055
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0511996746063232,
      "learning_rate": 5.074894737812023e-06,
      "loss": 0.5158,
      "step": 8056
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0970938205718994,
      "learning_rate": 5.07386885766765e-06,
      "loss": 0.5757,
      "step": 8057
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9284539818763733,
      "learning_rate": 5.072842974412916e-06,
      "loss": 0.4619,
      "step": 8058
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9632412791252136,
      "learning_rate": 5.071817088091017e-06,
      "loss": 0.4778,
      "step": 8059
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0104366540908813,
      "learning_rate": 5.0707911987451496e-06,
      "loss": 0.5728,
      "step": 8060
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1155264377593994,
      "learning_rate": 5.0697653064185125e-06,
      "loss": 0.5494,
      "step": 8061
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0364484786987305,
      "learning_rate": 5.068739411154301e-06,
      "loss": 0.5255,
      "step": 8062
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0144158601760864,
      "learning_rate": 5.0677135129957115e-06,
      "loss": 0.5346,
      "step": 8063
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9359563589096069,
      "learning_rate": 5.066687611985941e-06,
      "loss": 0.5088,
      "step": 8064
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0331517457962036,
      "learning_rate": 5.065661708168188e-06,
      "loss": 0.4937,
      "step": 8065
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1041889190673828,
      "learning_rate": 5.064635801585649e-06,
      "loss": 0.4994,
      "step": 8066
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.930534839630127,
      "learning_rate": 5.06360989228152e-06,
      "loss": 0.4854,
      "step": 8067
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0794141292572021,
      "learning_rate": 5.062583980299002e-06,
      "loss": 0.5193,
      "step": 8068
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9313056468963623,
      "learning_rate": 5.061558065681288e-06,
      "loss": 0.5059,
      "step": 8069
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0706911087036133,
      "learning_rate": 5.060532148471578e-06,
      "loss": 0.5567,
      "step": 8070
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0160199403762817,
      "learning_rate": 5.059506228713071e-06,
      "loss": 0.4841,
      "step": 8071
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9667502641677856,
      "learning_rate": 5.058480306448965e-06,
      "loss": 0.5376,
      "step": 8072
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.044235110282898,
      "learning_rate": 5.057454381722455e-06,
      "loss": 0.5092,
      "step": 8073
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0147597789764404,
      "learning_rate": 5.056428454576741e-06,
      "loss": 0.5273,
      "step": 8074
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9804866909980774,
      "learning_rate": 5.0554025250550195e-06,
      "loss": 0.5348,
      "step": 8075
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0162832736968994,
      "learning_rate": 5.054376593200493e-06,
      "loss": 0.4554,
      "step": 8076
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.035589575767517,
      "learning_rate": 5.053350659056356e-06,
      "loss": 0.5056,
      "step": 8077
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0157426595687866,
      "learning_rate": 5.052324722665809e-06,
      "loss": 0.5324,
      "step": 8078
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0081170797348022,
      "learning_rate": 5.0512987840720495e-06,
      "loss": 0.5206,
      "step": 8079
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0904947519302368,
      "learning_rate": 5.0502728433182765e-06,
      "loss": 0.5142,
      "step": 8080
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0651867389678955,
      "learning_rate": 5.049246900447689e-06,
      "loss": 0.5411,
      "step": 8081
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9559816718101501,
      "learning_rate": 5.048220955503487e-06,
      "loss": 0.5081,
      "step": 8082
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8969957232475281,
      "learning_rate": 5.047195008528868e-06,
      "loss": 0.4831,
      "step": 8083
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9508232474327087,
      "learning_rate": 5.04616905956703e-06,
      "loss": 0.49,
      "step": 8084
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9438045620918274,
      "learning_rate": 5.045143108661174e-06,
      "loss": 0.5268,
      "step": 8085
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9639183878898621,
      "learning_rate": 5.044117155854499e-06,
      "loss": 0.5587,
      "step": 8086
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8572230935096741,
      "learning_rate": 5.043091201190204e-06,
      "loss": 0.5037,
      "step": 8087
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.997568666934967,
      "learning_rate": 5.042065244711488e-06,
      "loss": 0.4654,
      "step": 8088
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0512139797210693,
      "learning_rate": 5.041039286461552e-06,
      "loss": 0.5253,
      "step": 8089
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0104326009750366,
      "learning_rate": 5.040013326483593e-06,
      "loss": 0.5375,
      "step": 8090
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1923975944519043,
      "learning_rate": 5.038987364820813e-06,
      "loss": 0.5782,
      "step": 8091
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0443609952926636,
      "learning_rate": 5.037961401516411e-06,
      "loss": 0.4763,
      "step": 8092
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9891601800918579,
      "learning_rate": 5.036935436613586e-06,
      "loss": 0.5175,
      "step": 8093
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0420844554901123,
      "learning_rate": 5.0359094701555375e-06,
      "loss": 0.512,
      "step": 8094
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.922382652759552,
      "learning_rate": 5.034883502185467e-06,
      "loss": 0.5143,
      "step": 8095
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.006429672241211,
      "learning_rate": 5.033857532746573e-06,
      "loss": 0.5191,
      "step": 8096
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9326863884925842,
      "learning_rate": 5.032831561882057e-06,
      "loss": 0.4857,
      "step": 8097
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9927276372909546,
      "learning_rate": 5.0318055896351185e-06,
      "loss": 0.5147,
      "step": 8098
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.885697066783905,
      "learning_rate": 5.030779616048955e-06,
      "loss": 0.4553,
      "step": 8099
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1350202560424805,
      "learning_rate": 5.02975364116677e-06,
      "loss": 0.5308,
      "step": 8100
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0157482624053955,
      "learning_rate": 5.0287276650317626e-06,
      "loss": 0.5095,
      "step": 8101
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0490323305130005,
      "learning_rate": 5.027701687687135e-06,
      "loss": 0.5334,
      "step": 8102
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9174306392669678,
      "learning_rate": 5.026675709176084e-06,
      "loss": 0.456,
      "step": 8103
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9931322336196899,
      "learning_rate": 5.0256497295418115e-06,
      "loss": 0.5302,
      "step": 8104
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.018126130104065,
      "learning_rate": 5.0246237488275185e-06,
      "loss": 0.4929,
      "step": 8105
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.996679961681366,
      "learning_rate": 5.0235977670764055e-06,
      "loss": 0.4864,
      "step": 8106
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0220783948898315,
      "learning_rate": 5.022571784331672e-06,
      "loss": 0.5348,
      "step": 8107
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9700262546539307,
      "learning_rate": 5.021545800636519e-06,
      "loss": 0.511,
      "step": 8108
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.045074701309204,
      "learning_rate": 5.020519816034148e-06,
      "loss": 0.5139,
      "step": 8109
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0336368083953857,
      "learning_rate": 5.019493830567758e-06,
      "loss": 0.5456,
      "step": 8110
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0149891376495361,
      "learning_rate": 5.018467844280553e-06,
      "loss": 0.5,
      "step": 8111
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.019803524017334,
      "learning_rate": 5.0174418572157276e-06,
      "loss": 0.4974,
      "step": 8112
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9631032943725586,
      "learning_rate": 5.0164158694164884e-06,
      "loss": 0.4931,
      "step": 8113
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.086776614189148,
      "learning_rate": 5.015389880926035e-06,
      "loss": 0.5513,
      "step": 8114
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9517618417739868,
      "learning_rate": 5.014363891787567e-06,
      "loss": 0.5024,
      "step": 8115
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0779685974121094,
      "learning_rate": 5.013337902044283e-06,
      "loss": 0.5292,
      "step": 8116
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1412652730941772,
      "learning_rate": 5.0123119117393894e-06,
      "loss": 0.5368,
      "step": 8117
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0637733936309814,
      "learning_rate": 5.011285920916082e-06,
      "loss": 0.5402,
      "step": 8118
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0142605304718018,
      "learning_rate": 5.010259929617565e-06,
      "loss": 0.5356,
      "step": 8119
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.202275276184082,
      "learning_rate": 5.009233937887036e-06,
      "loss": 0.5115,
      "step": 8120
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0249601602554321,
      "learning_rate": 5.0082079457677e-06,
      "loss": 0.49,
      "step": 8121
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1110284328460693,
      "learning_rate": 5.007181953302755e-06,
      "loss": 0.5404,
      "step": 8122
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0863715410232544,
      "learning_rate": 5.006155960535405e-06,
      "loss": 0.5385,
      "step": 8123
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.015622854232788,
      "learning_rate": 5.005129967508845e-06,
      "loss": 0.5406,
      "step": 8124
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0340197086334229,
      "learning_rate": 5.004103974266284e-06,
      "loss": 0.5186,
      "step": 8125
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0327247381210327,
      "learning_rate": 5.0030779808509155e-06,
      "loss": 0.5257,
      "step": 8126
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9580963253974915,
      "learning_rate": 5.002051987305947e-06,
      "loss": 0.5182,
      "step": 8127
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9721316695213318,
      "learning_rate": 5.0010259936745735e-06,
      "loss": 0.4748,
      "step": 8128
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0096826553344727,
      "learning_rate": 5e-06,
      "loss": 0.548,
      "step": 8129
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9705838561058044,
      "learning_rate": 4.998974006325428e-06,
      "loss": 0.5073,
      "step": 8130
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.019299864768982,
      "learning_rate": 4.997948012694056e-06,
      "loss": 0.5197,
      "step": 8131
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1114896535873413,
      "learning_rate": 4.9969220191490845e-06,
      "loss": 0.5462,
      "step": 8132
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0755159854888916,
      "learning_rate": 4.995896025733719e-06,
      "loss": 0.5358,
      "step": 8133
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1081016063690186,
      "learning_rate": 4.994870032491156e-06,
      "loss": 0.5108,
      "step": 8134
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0805017948150635,
      "learning_rate": 4.993844039464598e-06,
      "loss": 0.6051,
      "step": 8135
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0222865343093872,
      "learning_rate": 4.992818046697245e-06,
      "loss": 0.5212,
      "step": 8136
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.035110592842102,
      "learning_rate": 4.991792054232301e-06,
      "loss": 0.4815,
      "step": 8137
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9688609838485718,
      "learning_rate": 4.990766062112966e-06,
      "loss": 0.5028,
      "step": 8138
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0254881381988525,
      "learning_rate": 4.989740070382438e-06,
      "loss": 0.5143,
      "step": 8139
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9833453893661499,
      "learning_rate": 4.988714079083918e-06,
      "loss": 0.5275,
      "step": 8140
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.152333378791809,
      "learning_rate": 4.987688088260613e-06,
      "loss": 0.5085,
      "step": 8141
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0296305418014526,
      "learning_rate": 4.986662097955718e-06,
      "loss": 0.518,
      "step": 8142
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9811147451400757,
      "learning_rate": 4.985636108212435e-06,
      "loss": 0.5338,
      "step": 8143
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9948538541793823,
      "learning_rate": 4.984610119073965e-06,
      "loss": 0.4752,
      "step": 8144
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0104701519012451,
      "learning_rate": 4.9835841305835115e-06,
      "loss": 0.5064,
      "step": 8145
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9781093597412109,
      "learning_rate": 4.982558142784273e-06,
      "loss": 0.5434,
      "step": 8146
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0168230533599854,
      "learning_rate": 4.98153215571945e-06,
      "loss": 0.5495,
      "step": 8147
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0309807062149048,
      "learning_rate": 4.980506169432243e-06,
      "loss": 0.525,
      "step": 8148
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.058572769165039,
      "learning_rate": 4.979480183965852e-06,
      "loss": 0.5429,
      "step": 8149
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9933605194091797,
      "learning_rate": 4.9784541993634824e-06,
      "loss": 0.5254,
      "step": 8150
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0911980867385864,
      "learning_rate": 4.977428215668329e-06,
      "loss": 0.5847,
      "step": 8151
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0300828218460083,
      "learning_rate": 4.976402232923597e-06,
      "loss": 0.5364,
      "step": 8152
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0467183589935303,
      "learning_rate": 4.9753762511724815e-06,
      "loss": 0.4921,
      "step": 8153
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.990119218826294,
      "learning_rate": 4.974350270458189e-06,
      "loss": 0.5429,
      "step": 8154
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0183179378509521,
      "learning_rate": 4.9733242908239175e-06,
      "loss": 0.5546,
      "step": 8155
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0725822448730469,
      "learning_rate": 4.972298312312867e-06,
      "loss": 0.539,
      "step": 8156
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1482937335968018,
      "learning_rate": 4.9712723349682365e-06,
      "loss": 0.5243,
      "step": 8157
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0681301355361938,
      "learning_rate": 4.970246358833231e-06,
      "loss": 0.4775,
      "step": 8158
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0565322637557983,
      "learning_rate": 4.969220383951046e-06,
      "loss": 0.4802,
      "step": 8159
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9825900197029114,
      "learning_rate": 4.968194410364884e-06,
      "loss": 0.5165,
      "step": 8160
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9784873723983765,
      "learning_rate": 4.967168438117945e-06,
      "loss": 0.5198,
      "step": 8161
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9857720732688904,
      "learning_rate": 4.966142467253428e-06,
      "loss": 0.4751,
      "step": 8162
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.069770336151123,
      "learning_rate": 4.965116497814534e-06,
      "loss": 0.5385,
      "step": 8163
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.05392587184906,
      "learning_rate": 4.964090529844464e-06,
      "loss": 0.5194,
      "step": 8164
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9713079333305359,
      "learning_rate": 4.963064563386416e-06,
      "loss": 0.5253,
      "step": 8165
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9919078350067139,
      "learning_rate": 4.96203859848359e-06,
      "loss": 0.5185,
      "step": 8166
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9249306917190552,
      "learning_rate": 4.961012635179188e-06,
      "loss": 0.4828,
      "step": 8167
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0226560831069946,
      "learning_rate": 4.959986673516408e-06,
      "loss": 0.513,
      "step": 8168
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9974196553230286,
      "learning_rate": 4.95896071353845e-06,
      "loss": 0.5031,
      "step": 8169
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.070699691772461,
      "learning_rate": 4.9579347552885125e-06,
      "loss": 0.5519,
      "step": 8170
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.057245135307312,
      "learning_rate": 4.956908798809797e-06,
      "loss": 0.5376,
      "step": 8171
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1630752086639404,
      "learning_rate": 4.955882844145503e-06,
      "loss": 0.5544,
      "step": 8172
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0155560970306396,
      "learning_rate": 4.954856891338827e-06,
      "loss": 0.5533,
      "step": 8173
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.047834873199463,
      "learning_rate": 4.95383094043297e-06,
      "loss": 0.5008,
      "step": 8174
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9876782298088074,
      "learning_rate": 4.952804991471134e-06,
      "loss": 0.4999,
      "step": 8175
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9726967215538025,
      "learning_rate": 4.951779044496515e-06,
      "loss": 0.4991,
      "step": 8176
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.105119228363037,
      "learning_rate": 4.9507530995523115e-06,
      "loss": 0.516,
      "step": 8177
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9466031789779663,
      "learning_rate": 4.949727156681726e-06,
      "loss": 0.5342,
      "step": 8178
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0259852409362793,
      "learning_rate": 4.948701215927951e-06,
      "loss": 0.4601,
      "step": 8179
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.952790379524231,
      "learning_rate": 4.947675277334193e-06,
      "loss": 0.5096,
      "step": 8180
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0431325435638428,
      "learning_rate": 4.946649340943645e-06,
      "loss": 0.5657,
      "step": 8181
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0588798522949219,
      "learning_rate": 4.9456234067995094e-06,
      "loss": 0.5562,
      "step": 8182
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.000832200050354,
      "learning_rate": 4.9445974749449805e-06,
      "loss": 0.4955,
      "step": 8183
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9583312273025513,
      "learning_rate": 4.9435715454232615e-06,
      "loss": 0.5197,
      "step": 8184
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0641454458236694,
      "learning_rate": 4.942545618277547e-06,
      "loss": 0.5462,
      "step": 8185
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0573210716247559,
      "learning_rate": 4.9415196935510375e-06,
      "loss": 0.5063,
      "step": 8186
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0898346900939941,
      "learning_rate": 4.940493771286929e-06,
      "loss": 0.5157,
      "step": 8187
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0173532962799072,
      "learning_rate": 4.939467851528423e-06,
      "loss": 0.5257,
      "step": 8188
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9410024881362915,
      "learning_rate": 4.938441934318713e-06,
      "loss": 0.4716,
      "step": 8189
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9916033148765564,
      "learning_rate": 4.937416019701e-06,
      "loss": 0.5225,
      "step": 8190
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0031166076660156,
      "learning_rate": 4.93639010771848e-06,
      "loss": 0.4652,
      "step": 8191
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0323004722595215,
      "learning_rate": 4.9353641984143526e-06,
      "loss": 0.528,
      "step": 8192
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9885910153388977,
      "learning_rate": 4.934338291831813e-06,
      "loss": 0.5162,
      "step": 8193
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0446704626083374,
      "learning_rate": 4.93331238801406e-06,
      "loss": 0.5019,
      "step": 8194
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1711422204971313,
      "learning_rate": 4.932286487004291e-06,
      "loss": 0.596,
      "step": 8195
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0227354764938354,
      "learning_rate": 4.931260588845701e-06,
      "loss": 0.517,
      "step": 8196
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9722004532814026,
      "learning_rate": 4.930234693581489e-06,
      "loss": 0.531,
      "step": 8197
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0352272987365723,
      "learning_rate": 4.929208801254851e-06,
      "loss": 0.5281,
      "step": 8198
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0903185606002808,
      "learning_rate": 4.928182911908987e-06,
      "loss": 0.581,
      "step": 8199
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.969157874584198,
      "learning_rate": 4.927157025587086e-06,
      "loss": 0.4981,
      "step": 8200
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.968953549861908,
      "learning_rate": 4.926131142332351e-06,
      "loss": 0.5285,
      "step": 8201
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0791070461273193,
      "learning_rate": 4.925105262187978e-06,
      "loss": 0.5594,
      "step": 8202
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0036054849624634,
      "learning_rate": 4.924079385197162e-06,
      "loss": 0.5002,
      "step": 8203
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.017087697982788,
      "learning_rate": 4.923053511403096e-06,
      "loss": 0.5861,
      "step": 8204
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9648655652999878,
      "learning_rate": 4.922027640848981e-06,
      "loss": 0.5136,
      "step": 8205
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1127065420150757,
      "learning_rate": 4.921001773578012e-06,
      "loss": 0.5753,
      "step": 8206
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1435563564300537,
      "learning_rate": 4.9199759096333825e-06,
      "loss": 0.5837,
      "step": 8207
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0285307168960571,
      "learning_rate": 4.918950049058289e-06,
      "loss": 0.5234,
      "step": 8208
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9845409989356995,
      "learning_rate": 4.9179241918959255e-06,
      "loss": 0.5261,
      "step": 8209
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9043030142784119,
      "learning_rate": 4.916898338189491e-06,
      "loss": 0.4721,
      "step": 8210
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9924689531326294,
      "learning_rate": 4.9158724879821775e-06,
      "loss": 0.5078,
      "step": 8211
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9461621046066284,
      "learning_rate": 4.914846641317181e-06,
      "loss": 0.4674,
      "step": 8212
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0354009866714478,
      "learning_rate": 4.913820798237695e-06,
      "loss": 0.5153,
      "step": 8213
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0802775621414185,
      "learning_rate": 4.912794958786917e-06,
      "loss": 0.5283,
      "step": 8214
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0849580764770508,
      "learning_rate": 4.91176912300804e-06,
      "loss": 0.5152,
      "step": 8215
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.101136326789856,
      "learning_rate": 4.9107432909442575e-06,
      "loss": 0.5241,
      "step": 8216
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0249545574188232,
      "learning_rate": 4.909717462638763e-06,
      "loss": 0.5575,
      "step": 8217
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.073858380317688,
      "learning_rate": 4.908691638134754e-06,
      "loss": 0.5598,
      "step": 8218
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.037700891494751,
      "learning_rate": 4.907665817475424e-06,
      "loss": 0.5018,
      "step": 8219
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0014963150024414,
      "learning_rate": 4.906640000703963e-06,
      "loss": 0.5335,
      "step": 8220
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0454741716384888,
      "learning_rate": 4.905614187863565e-06,
      "loss": 0.5591,
      "step": 8221
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1414268016815186,
      "learning_rate": 4.904588378997428e-06,
      "loss": 0.5646,
      "step": 8222
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.002533197402954,
      "learning_rate": 4.903562574148744e-06,
      "loss": 0.5633,
      "step": 8223
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0724406242370605,
      "learning_rate": 4.902536773360702e-06,
      "loss": 0.4934,
      "step": 8224
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9824497699737549,
      "learning_rate": 4.9015109766764985e-06,
      "loss": 0.5355,
      "step": 8225
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.024309515953064,
      "learning_rate": 4.900485184139323e-06,
      "loss": 0.5124,
      "step": 8226
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.008635401725769,
      "learning_rate": 4.899459395792373e-06,
      "loss": 0.533,
      "step": 8227
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9910851120948792,
      "learning_rate": 4.8984336116788355e-06,
      "loss": 0.4683,
      "step": 8228
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9705127477645874,
      "learning_rate": 4.897407831841908e-06,
      "loss": 0.4987,
      "step": 8229
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0432102680206299,
      "learning_rate": 4.8963820563247765e-06,
      "loss": 0.5263,
      "step": 8230
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0487322807312012,
      "learning_rate": 4.8953562851706385e-06,
      "loss": 0.4836,
      "step": 8231
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0196106433868408,
      "learning_rate": 4.894330518422683e-06,
      "loss": 0.5178,
      "step": 8232
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0171598196029663,
      "learning_rate": 4.893304756124102e-06,
      "loss": 0.5162,
      "step": 8233
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.055601716041565,
      "learning_rate": 4.8922789983180854e-06,
      "loss": 0.5432,
      "step": 8234
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.013988733291626,
      "learning_rate": 4.891253245047826e-06,
      "loss": 0.4949,
      "step": 8235
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9848381876945496,
      "learning_rate": 4.890227496356515e-06,
      "loss": 0.5142,
      "step": 8236
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9790887832641602,
      "learning_rate": 4.889201752287342e-06,
      "loss": 0.5156,
      "step": 8237
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.032889723777771,
      "learning_rate": 4.888176012883496e-06,
      "loss": 0.537,
      "step": 8238
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9301667213439941,
      "learning_rate": 4.88715027818817e-06,
      "loss": 0.5155,
      "step": 8239
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0242931842803955,
      "learning_rate": 4.886124548244555e-06,
      "loss": 0.5752,
      "step": 8240
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0368810892105103,
      "learning_rate": 4.885098823095838e-06,
      "loss": 0.5318,
      "step": 8241
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0037161111831665,
      "learning_rate": 4.884073102785209e-06,
      "loss": 0.5325,
      "step": 8242
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0490317344665527,
      "learning_rate": 4.883047387355858e-06,
      "loss": 0.5131,
      "step": 8243
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0288115739822388,
      "learning_rate": 4.882021676850977e-06,
      "loss": 0.5095,
      "step": 8244
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0677239894866943,
      "learning_rate": 4.880995971313752e-06,
      "loss": 0.4767,
      "step": 8245
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0304758548736572,
      "learning_rate": 4.879970270787372e-06,
      "loss": 0.567,
      "step": 8246
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1207501888275146,
      "learning_rate": 4.878944575315025e-06,
      "loss": 0.5218,
      "step": 8247
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9568483829498291,
      "learning_rate": 4.877918884939903e-06,
      "loss": 0.5296,
      "step": 8248
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0585397481918335,
      "learning_rate": 4.8768931997051925e-06,
      "loss": 0.5318,
      "step": 8249
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9968030452728271,
      "learning_rate": 4.8758675196540795e-06,
      "loss": 0.5101,
      "step": 8250
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0286331176757812,
      "learning_rate": 4.874841844829753e-06,
      "loss": 0.5026,
      "step": 8251
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0161633491516113,
      "learning_rate": 4.873816175275402e-06,
      "loss": 0.5088,
      "step": 8252
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0071805715560913,
      "learning_rate": 4.8727905110342135e-06,
      "loss": 0.5457,
      "step": 8253
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9614889025688171,
      "learning_rate": 4.871764852149373e-06,
      "loss": 0.5366,
      "step": 8254
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.009331226348877,
      "learning_rate": 4.87073919866407e-06,
      "loss": 0.5317,
      "step": 8255
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9906996488571167,
      "learning_rate": 4.869713550621487e-06,
      "loss": 0.5224,
      "step": 8256
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9802281856536865,
      "learning_rate": 4.868687908064815e-06,
      "loss": 0.5189,
      "step": 8257
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9882171750068665,
      "learning_rate": 4.867662271037238e-06,
      "loss": 0.4723,
      "step": 8258
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1115453243255615,
      "learning_rate": 4.866636639581943e-06,
      "loss": 0.5251,
      "step": 8259
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9748610258102417,
      "learning_rate": 4.865611013742114e-06,
      "loss": 0.5112,
      "step": 8260
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0606300830841064,
      "learning_rate": 4.864585393560939e-06,
      "loss": 0.5332,
      "step": 8261
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.117438793182373,
      "learning_rate": 4.863559779081601e-06,
      "loss": 0.5163,
      "step": 8262
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9546915888786316,
      "learning_rate": 4.862534170347287e-06,
      "loss": 0.5315,
      "step": 8263
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9329437613487244,
      "learning_rate": 4.861508567401179e-06,
      "loss": 0.4849,
      "step": 8264
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0045757293701172,
      "learning_rate": 4.860482970286465e-06,
      "loss": 0.4874,
      "step": 8265
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9793338775634766,
      "learning_rate": 4.859457379046327e-06,
      "loss": 0.5172,
      "step": 8266
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.05069899559021,
      "learning_rate": 4.858431793723952e-06,
      "loss": 0.4986,
      "step": 8267
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9594676494598389,
      "learning_rate": 4.857406214362518e-06,
      "loss": 0.4993,
      "step": 8268
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0364305973052979,
      "learning_rate": 4.856380641005215e-06,
      "loss": 0.5177,
      "step": 8269
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0088708400726318,
      "learning_rate": 4.855355073695223e-06,
      "loss": 0.587,
      "step": 8270
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0791456699371338,
      "learning_rate": 4.8543295124757265e-06,
      "loss": 0.5208,
      "step": 8271
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.078839659690857,
      "learning_rate": 4.8533039573899075e-06,
      "loss": 0.537,
      "step": 8272
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.02333402633667,
      "learning_rate": 4.852278408480946e-06,
      "loss": 0.5355,
      "step": 8273
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1506671905517578,
      "learning_rate": 4.8512528657920275e-06,
      "loss": 0.5161,
      "step": 8274
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0825648307800293,
      "learning_rate": 4.850227329366335e-06,
      "loss": 0.5492,
      "step": 8275
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0078729391098022,
      "learning_rate": 4.849201799247049e-06,
      "loss": 0.4773,
      "step": 8276
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9716581702232361,
      "learning_rate": 4.848176275477348e-06,
      "loss": 0.4887,
      "step": 8277
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9962989091873169,
      "learning_rate": 4.847150758100418e-06,
      "loss": 0.5134,
      "step": 8278
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0206736326217651,
      "learning_rate": 4.846125247159437e-06,
      "loss": 0.513,
      "step": 8279
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9762862920761108,
      "learning_rate": 4.845099742697588e-06,
      "loss": 0.4922,
      "step": 8280
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9894128441810608,
      "learning_rate": 4.844074244758047e-06,
      "loss": 0.4491,
      "step": 8281
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0172109603881836,
      "learning_rate": 4.843048753383998e-06,
      "loss": 0.5184,
      "step": 8282
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0501362085342407,
      "learning_rate": 4.8420232686186226e-06,
      "loss": 0.5013,
      "step": 8283
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0115443468093872,
      "learning_rate": 4.840997790505097e-06,
      "loss": 0.4965,
      "step": 8284
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0318505764007568,
      "learning_rate": 4.8399723190866e-06,
      "loss": 0.4768,
      "step": 8285
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9210131764411926,
      "learning_rate": 4.838946854406311e-06,
      "loss": 0.5059,
      "step": 8286
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0285263061523438,
      "learning_rate": 4.8379213965074125e-06,
      "loss": 0.5317,
      "step": 8287
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0240325927734375,
      "learning_rate": 4.83689594543308e-06,
      "loss": 0.5145,
      "step": 8288
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0739846229553223,
      "learning_rate": 4.835870501226489e-06,
      "loss": 0.5382,
      "step": 8289
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0172021389007568,
      "learning_rate": 4.834845063930821e-06,
      "loss": 0.5106,
      "step": 8290
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.904426634311676,
      "learning_rate": 4.833819633589254e-06,
      "loss": 0.5247,
      "step": 8291
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9170308113098145,
      "learning_rate": 4.832794210244965e-06,
      "loss": 0.4646,
      "step": 8292
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.143886923789978,
      "learning_rate": 4.831768793941129e-06,
      "loss": 0.569,
      "step": 8293
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.993350625038147,
      "learning_rate": 4.830743384720922e-06,
      "loss": 0.514,
      "step": 8294
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.063253402709961,
      "learning_rate": 4.829717982627525e-06,
      "loss": 0.5275,
      "step": 8295
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0138853788375854,
      "learning_rate": 4.82869258770411e-06,
      "loss": 0.5617,
      "step": 8296
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8883842825889587,
      "learning_rate": 4.827667199993855e-06,
      "loss": 0.4943,
      "step": 8297
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.048701524734497,
      "learning_rate": 4.826641819539933e-06,
      "loss": 0.493,
      "step": 8298
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.979813277721405,
      "learning_rate": 4.825616446385523e-06,
      "loss": 0.5189,
      "step": 8299
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9900126457214355,
      "learning_rate": 4.824591080573797e-06,
      "loss": 0.5252,
      "step": 8300
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0179758071899414,
      "learning_rate": 4.823565722147932e-06,
      "loss": 0.544,
      "step": 8301
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0025790929794312,
      "learning_rate": 4.8225403711511e-06,
      "loss": 0.4974,
      "step": 8302
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0100517272949219,
      "learning_rate": 4.821515027626473e-06,
      "loss": 0.5133,
      "step": 8303
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9746113419532776,
      "learning_rate": 4.8204896916172285e-06,
      "loss": 0.4728,
      "step": 8304
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9883295893669128,
      "learning_rate": 4.819464363166539e-06,
      "loss": 0.473,
      "step": 8305
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.025826096534729,
      "learning_rate": 4.818439042317578e-06,
      "loss": 0.5227,
      "step": 8306
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9911962747573853,
      "learning_rate": 4.817413729113516e-06,
      "loss": 0.5128,
      "step": 8307
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.089560627937317,
      "learning_rate": 4.816388423597527e-06,
      "loss": 0.5084,
      "step": 8308
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0557607412338257,
      "learning_rate": 4.815363125812784e-06,
      "loss": 0.5609,
      "step": 8309
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0592869520187378,
      "learning_rate": 4.814337835802457e-06,
      "loss": 0.5411,
      "step": 8310
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0457826852798462,
      "learning_rate": 4.813312553609716e-06,
      "loss": 0.4667,
      "step": 8311
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9835514426231384,
      "learning_rate": 4.812287279277735e-06,
      "loss": 0.5102,
      "step": 8312
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9978135228157043,
      "learning_rate": 4.811262012849685e-06,
      "loss": 0.5234,
      "step": 8313
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9925533533096313,
      "learning_rate": 4.810236754368735e-06,
      "loss": 0.5346,
      "step": 8314
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0429518222808838,
      "learning_rate": 4.8092115038780525e-06,
      "loss": 0.5445,
      "step": 8315
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.012973427772522,
      "learning_rate": 4.808186261420811e-06,
      "loss": 0.5187,
      "step": 8316
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9603223204612732,
      "learning_rate": 4.80716102704018e-06,
      "loss": 0.5143,
      "step": 8317
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.066994547843933,
      "learning_rate": 4.806135800779328e-06,
      "loss": 0.5151,
      "step": 8318
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9137266874313354,
      "learning_rate": 4.805110582681421e-06,
      "loss": 0.4991,
      "step": 8319
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0653810501098633,
      "learning_rate": 4.804085372789629e-06,
      "loss": 0.5267,
      "step": 8320
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0767114162445068,
      "learning_rate": 4.803060171147122e-06,
      "loss": 0.5417,
      "step": 8321
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0749000310897827,
      "learning_rate": 4.802034977797066e-06,
      "loss": 0.5645,
      "step": 8322
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0301631689071655,
      "learning_rate": 4.801009792782627e-06,
      "loss": 0.5231,
      "step": 8323
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8945549726486206,
      "learning_rate": 4.799984616146974e-06,
      "loss": 0.495,
      "step": 8324
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0703390836715698,
      "learning_rate": 4.798959447933274e-06,
      "loss": 0.5903,
      "step": 8325
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0251332521438599,
      "learning_rate": 4.797934288184692e-06,
      "loss": 0.51,
      "step": 8326
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1210784912109375,
      "learning_rate": 4.796909136944394e-06,
      "loss": 0.5686,
      "step": 8327
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.000189185142517,
      "learning_rate": 4.795883994255544e-06,
      "loss": 0.4993,
      "step": 8328
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0635324716567993,
      "learning_rate": 4.794858860161311e-06,
      "loss": 0.6022,
      "step": 8329
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.010852575302124,
      "learning_rate": 4.793833734704858e-06,
      "loss": 0.5292,
      "step": 8330
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0470019578933716,
      "learning_rate": 4.792808617929348e-06,
      "loss": 0.5249,
      "step": 8331
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9671838283538818,
      "learning_rate": 4.791783509877948e-06,
      "loss": 0.5258,
      "step": 8332
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0945289134979248,
      "learning_rate": 4.790758410593818e-06,
      "loss": 0.5389,
      "step": 8333
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.033327341079712,
      "learning_rate": 4.789733320120124e-06,
      "loss": 0.5051,
      "step": 8334
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0071804523468018,
      "learning_rate": 4.788708238500029e-06,
      "loss": 0.5091,
      "step": 8335
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0532846450805664,
      "learning_rate": 4.787683165776695e-06,
      "loss": 0.5137,
      "step": 8336
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0724596977233887,
      "learning_rate": 4.786658101993283e-06,
      "loss": 0.5375,
      "step": 8337
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.977753221988678,
      "learning_rate": 4.785633047192959e-06,
      "loss": 0.4983,
      "step": 8338
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0433781147003174,
      "learning_rate": 4.7846080014188786e-06,
      "loss": 0.5464,
      "step": 8339
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9462462663650513,
      "learning_rate": 4.783582964714209e-06,
      "loss": 0.5299,
      "step": 8340
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0582375526428223,
      "learning_rate": 4.782557937122104e-06,
      "loss": 0.5019,
      "step": 8341
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.099599838256836,
      "learning_rate": 4.781532918685731e-06,
      "loss": 0.4984,
      "step": 8342
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1724430322647095,
      "learning_rate": 4.780507909448246e-06,
      "loss": 0.5364,
      "step": 8343
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1686534881591797,
      "learning_rate": 4.77948290945281e-06,
      "loss": 0.507,
      "step": 8344
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1176148653030396,
      "learning_rate": 4.778457918742579e-06,
      "loss": 0.5317,
      "step": 8345
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0374079942703247,
      "learning_rate": 4.777432937360716e-06,
      "loss": 0.5507,
      "step": 8346
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.043051838874817,
      "learning_rate": 4.776407965350378e-06,
      "loss": 0.5252,
      "step": 8347
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9879152774810791,
      "learning_rate": 4.775383002754723e-06,
      "loss": 0.5514,
      "step": 8348
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9836734533309937,
      "learning_rate": 4.7743580496169095e-06,
      "loss": 0.5587,
      "step": 8349
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.044381856918335,
      "learning_rate": 4.773333105980091e-06,
      "loss": 0.5032,
      "step": 8350
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9817684292793274,
      "learning_rate": 4.772308171887427e-06,
      "loss": 0.4559,
      "step": 8351
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1289509534835815,
      "learning_rate": 4.771283247382076e-06,
      "loss": 0.5294,
      "step": 8352
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9817908406257629,
      "learning_rate": 4.770258332507191e-06,
      "loss": 0.4814,
      "step": 8353
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.103776454925537,
      "learning_rate": 4.7692334273059265e-06,
      "loss": 0.5398,
      "step": 8354
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0110007524490356,
      "learning_rate": 4.768208531821441e-06,
      "loss": 0.4905,
      "step": 8355
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.983505368232727,
      "learning_rate": 4.767183646096889e-06,
      "loss": 0.5312,
      "step": 8356
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0207186937332153,
      "learning_rate": 4.766158770175422e-06,
      "loss": 0.5324,
      "step": 8357
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0847700834274292,
      "learning_rate": 4.765133904100196e-06,
      "loss": 0.4881,
      "step": 8358
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1356765031814575,
      "learning_rate": 4.764109047914365e-06,
      "loss": 0.5199,
      "step": 8359
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.091707706451416,
      "learning_rate": 4.763084201661081e-06,
      "loss": 0.5283,
      "step": 8360
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0814317464828491,
      "learning_rate": 4.762059365383497e-06,
      "loss": 0.5562,
      "step": 8361
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9616392850875854,
      "learning_rate": 4.761034539124765e-06,
      "loss": 0.4963,
      "step": 8362
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1386058330535889,
      "learning_rate": 4.760009722928038e-06,
      "loss": 0.5513,
      "step": 8363
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9828891158103943,
      "learning_rate": 4.7589849168364675e-06,
      "loss": 0.5039,
      "step": 8364
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0613303184509277,
      "learning_rate": 4.7579601208932015e-06,
      "loss": 0.5441,
      "step": 8365
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0633760690689087,
      "learning_rate": 4.756935335141395e-06,
      "loss": 0.5289,
      "step": 8366
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.004191279411316,
      "learning_rate": 4.755910559624194e-06,
      "loss": 0.5025,
      "step": 8367
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.016272783279419,
      "learning_rate": 4.754885794384752e-06,
      "loss": 0.5273,
      "step": 8368
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0552794933319092,
      "learning_rate": 4.7538610394662156e-06,
      "loss": 0.47,
      "step": 8369
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9942074418067932,
      "learning_rate": 4.7528362949117355e-06,
      "loss": 0.4947,
      "step": 8370
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.042699933052063,
      "learning_rate": 4.751811560764457e-06,
      "loss": 0.4983,
      "step": 8371
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0090547800064087,
      "learning_rate": 4.750786837067532e-06,
      "loss": 0.5255,
      "step": 8372
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2265372276306152,
      "learning_rate": 4.7497621238641055e-06,
      "loss": 0.5495,
      "step": 8373
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0453957319259644,
      "learning_rate": 4.7487374211973266e-06,
      "loss": 0.5459,
      "step": 8374
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9630346298217773,
      "learning_rate": 4.747712729110339e-06,
      "loss": 0.4726,
      "step": 8375
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0352476835250854,
      "learning_rate": 4.746688047646293e-06,
      "loss": 0.497,
      "step": 8376
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0061421394348145,
      "learning_rate": 4.745663376848331e-06,
      "loss": 0.4905,
      "step": 8377
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0827962160110474,
      "learning_rate": 4.744638716759599e-06,
      "loss": 0.5524,
      "step": 8378
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9328921437263489,
      "learning_rate": 4.743614067423245e-06,
      "loss": 0.5545,
      "step": 8379
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.996181309223175,
      "learning_rate": 4.742589428882406e-06,
      "loss": 0.4888,
      "step": 8380
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9578323364257812,
      "learning_rate": 4.7415648011802335e-06,
      "loss": 0.5488,
      "step": 8381
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0353081226348877,
      "learning_rate": 4.7405401843598686e-06,
      "loss": 0.5308,
      "step": 8382
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9680172204971313,
      "learning_rate": 4.739515578464454e-06,
      "loss": 0.4932,
      "step": 8383
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0518368482589722,
      "learning_rate": 4.73849098353713e-06,
      "loss": 0.5648,
      "step": 8384
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9996711611747742,
      "learning_rate": 4.737466399621043e-06,
      "loss": 0.5298,
      "step": 8385
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0345032215118408,
      "learning_rate": 4.736441826759332e-06,
      "loss": 0.4943,
      "step": 8386
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.020096778869629,
      "learning_rate": 4.73541726499514e-06,
      "loss": 0.5162,
      "step": 8387
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0417054891586304,
      "learning_rate": 4.734392714371603e-06,
      "loss": 0.5666,
      "step": 8388
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0406177043914795,
      "learning_rate": 4.733368174931867e-06,
      "loss": 0.5087,
      "step": 8389
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0037623643875122,
      "learning_rate": 4.7323436467190705e-06,
      "loss": 0.488,
      "step": 8390
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1015545129776,
      "learning_rate": 4.7313191297763524e-06,
      "loss": 0.5243,
      "step": 8391
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9793073534965515,
      "learning_rate": 4.730294624146849e-06,
      "loss": 0.5298,
      "step": 8392
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1413781642913818,
      "learning_rate": 4.729270129873701e-06,
      "loss": 0.5088,
      "step": 8393
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.040779709815979,
      "learning_rate": 4.728245647000047e-06,
      "loss": 0.512,
      "step": 8394
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9371123313903809,
      "learning_rate": 4.7272211755690245e-06,
      "loss": 0.5446,
      "step": 8395
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.961765706539154,
      "learning_rate": 4.7261967156237676e-06,
      "loss": 0.502,
      "step": 8396
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.990039050579071,
      "learning_rate": 4.725172267207413e-06,
      "loss": 0.4656,
      "step": 8397
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9950096607208252,
      "learning_rate": 4.724147830363101e-06,
      "loss": 0.5172,
      "step": 8398
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9665403366088867,
      "learning_rate": 4.723123405133965e-06,
      "loss": 0.5397,
      "step": 8399
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0354504585266113,
      "learning_rate": 4.722098991563137e-06,
      "loss": 0.5202,
      "step": 8400
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0049880743026733,
      "learning_rate": 4.721074589693753e-06,
      "loss": 0.5133,
      "step": 8401
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0299439430236816,
      "learning_rate": 4.72005019956895e-06,
      "loss": 0.5117,
      "step": 8402
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0537679195404053,
      "learning_rate": 4.719025821231859e-06,
      "loss": 0.5176,
      "step": 8403
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0134973526000977,
      "learning_rate": 4.718001454725612e-06,
      "loss": 0.4598,
      "step": 8404
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0414093732833862,
      "learning_rate": 4.716977100093342e-06,
      "loss": 0.5476,
      "step": 8405
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0167829990386963,
      "learning_rate": 4.715952757378183e-06,
      "loss": 0.51,
      "step": 8406
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0440165996551514,
      "learning_rate": 4.714928426623266e-06,
      "loss": 0.5288,
      "step": 8407
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0283722877502441,
      "learning_rate": 4.71390410787172e-06,
      "loss": 0.5408,
      "step": 8408
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.045358657836914,
      "learning_rate": 4.712879801166676e-06,
      "loss": 0.5227,
      "step": 8409
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.03081476688385,
      "learning_rate": 4.711855506551267e-06,
      "loss": 0.5446,
      "step": 8410
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0450913906097412,
      "learning_rate": 4.71083122406862e-06,
      "loss": 0.5193,
      "step": 8411
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0645527839660645,
      "learning_rate": 4.709806953761863e-06,
      "loss": 0.5577,
      "step": 8412
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9882965683937073,
      "learning_rate": 4.7087826956741266e-06,
      "loss": 0.532,
      "step": 8413
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0072165727615356,
      "learning_rate": 4.707758449848536e-06,
      "loss": 0.4521,
      "step": 8414
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1032558679580688,
      "learning_rate": 4.7067342163282225e-06,
      "loss": 0.5005,
      "step": 8415
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9673135876655579,
      "learning_rate": 4.70570999515631e-06,
      "loss": 0.5064,
      "step": 8416
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0546165704727173,
      "learning_rate": 4.704685786375927e-06,
      "loss": 0.5241,
      "step": 8417
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.070737361907959,
      "learning_rate": 4.703661590030196e-06,
      "loss": 0.519,
      "step": 8418
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0162523984909058,
      "learning_rate": 4.702637406162247e-06,
      "loss": 0.5676,
      "step": 8419
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9657307863235474,
      "learning_rate": 4.7016132348152e-06,
      "loss": 0.4602,
      "step": 8420
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0089730024337769,
      "learning_rate": 4.700589076032184e-06,
      "loss": 0.5214,
      "step": 8421
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0143710374832153,
      "learning_rate": 4.699564929856318e-06,
      "loss": 0.5651,
      "step": 8422
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9930520057678223,
      "learning_rate": 4.698540796330729e-06,
      "loss": 0.5328,
      "step": 8423
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.998888373374939,
      "learning_rate": 4.697516675498538e-06,
      "loss": 0.5252,
      "step": 8424
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0171403884887695,
      "learning_rate": 4.69649256740287e-06,
      "loss": 0.5095,
      "step": 8425
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1262316703796387,
      "learning_rate": 4.695468472086841e-06,
      "loss": 0.5916,
      "step": 8426
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0274018049240112,
      "learning_rate": 4.694444389593576e-06,
      "loss": 0.5097,
      "step": 8427
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0473886728286743,
      "learning_rate": 4.693420319966195e-06,
      "loss": 0.5476,
      "step": 8428
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9588348269462585,
      "learning_rate": 4.692396263247818e-06,
      "loss": 0.5464,
      "step": 8429
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.990731418132782,
      "learning_rate": 4.691372219481564e-06,
      "loss": 0.5093,
      "step": 8430
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9880147576332092,
      "learning_rate": 4.690348188710552e-06,
      "loss": 0.5624,
      "step": 8431
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1084983348846436,
      "learning_rate": 4.689324170977901e-06,
      "loss": 0.5151,
      "step": 8432
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9567615985870361,
      "learning_rate": 4.688300166326729e-06,
      "loss": 0.5076,
      "step": 8433
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.052372932434082,
      "learning_rate": 4.6872761748001515e-06,
      "loss": 0.519,
      "step": 8434
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0967961549758911,
      "learning_rate": 4.6862521964412865e-06,
      "loss": 0.5539,
      "step": 8435
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1107224225997925,
      "learning_rate": 4.6852282312932505e-06,
      "loss": 0.5518,
      "step": 8436
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0529141426086426,
      "learning_rate": 4.684204279399159e-06,
      "loss": 0.5703,
      "step": 8437
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0516138076782227,
      "learning_rate": 4.683180340802126e-06,
      "loss": 0.5452,
      "step": 8438
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0087449550628662,
      "learning_rate": 4.682156415545266e-06,
      "loss": 0.4957,
      "step": 8439
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.019419550895691,
      "learning_rate": 4.681132503671696e-06,
      "loss": 0.511,
      "step": 8440
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1684471368789673,
      "learning_rate": 4.680108605224526e-06,
      "loss": 0.5277,
      "step": 8441
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0348973274230957,
      "learning_rate": 4.679084720246869e-06,
      "loss": 0.5267,
      "step": 8442
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9964972138404846,
      "learning_rate": 4.67806084878184e-06,
      "loss": 0.5104,
      "step": 8443
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9671248197555542,
      "learning_rate": 4.677036990872546e-06,
      "loss": 0.4965,
      "step": 8444
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0187405347824097,
      "learning_rate": 4.676013146562103e-06,
      "loss": 0.5092,
      "step": 8445
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2161133289337158,
      "learning_rate": 4.674989315893618e-06,
      "loss": 0.555,
      "step": 8446
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9527446627616882,
      "learning_rate": 4.6739654989102034e-06,
      "loss": 0.5235,
      "step": 8447
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0290900468826294,
      "learning_rate": 4.672941695654965e-06,
      "loss": 0.5317,
      "step": 8448
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9144114851951599,
      "learning_rate": 4.6719179061710164e-06,
      "loss": 0.4646,
      "step": 8449
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0445489883422852,
      "learning_rate": 4.670894130501462e-06,
      "loss": 0.4481,
      "step": 8450
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0031083822250366,
      "learning_rate": 4.669870368689414e-06,
      "loss": 0.5274,
      "step": 8451
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1422038078308105,
      "learning_rate": 4.668846620777972e-06,
      "loss": 0.5328,
      "step": 8452
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.969397783279419,
      "learning_rate": 4.6678228868102495e-06,
      "loss": 0.54,
      "step": 8453
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9985876679420471,
      "learning_rate": 4.666799166829349e-06,
      "loss": 0.531,
      "step": 8454
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0527873039245605,
      "learning_rate": 4.665775460878377e-06,
      "loss": 0.5221,
      "step": 8455
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9826925992965698,
      "learning_rate": 4.664751769000436e-06,
      "loss": 0.4628,
      "step": 8456
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0141879320144653,
      "learning_rate": 4.663728091238634e-06,
      "loss": 0.5114,
      "step": 8457
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9625977277755737,
      "learning_rate": 4.662704427636071e-06,
      "loss": 0.4867,
      "step": 8458
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0554953813552856,
      "learning_rate": 4.661680778235852e-06,
      "loss": 0.5338,
      "step": 8459
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0381543636322021,
      "learning_rate": 4.660657143081079e-06,
      "loss": 0.5401,
      "step": 8460
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0382134914398193,
      "learning_rate": 4.65963352221485e-06,
      "loss": 0.5198,
      "step": 8461
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9821277856826782,
      "learning_rate": 4.658609915680272e-06,
      "loss": 0.4602,
      "step": 8462
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9753869771957397,
      "learning_rate": 4.657586323520443e-06,
      "loss": 0.4994,
      "step": 8463
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.028468132019043,
      "learning_rate": 4.6565627457784625e-06,
      "loss": 0.5271,
      "step": 8464
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9985352158546448,
      "learning_rate": 4.655539182497428e-06,
      "loss": 0.5155,
      "step": 8465
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9296327233314514,
      "learning_rate": 4.654515633720442e-06,
      "loss": 0.4305,
      "step": 8466
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9554487466812134,
      "learning_rate": 4.653492099490601e-06,
      "loss": 0.4575,
      "step": 8467
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.037149429321289,
      "learning_rate": 4.6524685798510025e-06,
      "loss": 0.5524,
      "step": 8468
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9794292449951172,
      "learning_rate": 4.651445074844742e-06,
      "loss": 0.5105,
      "step": 8469
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0618610382080078,
      "learning_rate": 4.650421584514917e-06,
      "loss": 0.5193,
      "step": 8470
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9309068918228149,
      "learning_rate": 4.649398108904624e-06,
      "loss": 0.4767,
      "step": 8471
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0112494230270386,
      "learning_rate": 4.648374648056957e-06,
      "loss": 0.5232,
      "step": 8472
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0387423038482666,
      "learning_rate": 4.64735120201501e-06,
      "loss": 0.5237,
      "step": 8473
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0283302068710327,
      "learning_rate": 4.646327770821875e-06,
      "loss": 0.5583,
      "step": 8474
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.013685941696167,
      "learning_rate": 4.64530435452065e-06,
      "loss": 0.5209,
      "step": 8475
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.009364128112793,
      "learning_rate": 4.644280953154424e-06,
      "loss": 0.5592,
      "step": 8476
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0646246671676636,
      "learning_rate": 4.643257566766289e-06,
      "loss": 0.5199,
      "step": 8477
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9765976071357727,
      "learning_rate": 4.642234195399336e-06,
      "loss": 0.5128,
      "step": 8478
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9888172149658203,
      "learning_rate": 4.641210839096659e-06,
      "loss": 0.5092,
      "step": 8479
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0132274627685547,
      "learning_rate": 4.6401874979013455e-06,
      "loss": 0.5179,
      "step": 8480
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0327469110488892,
      "learning_rate": 4.639164171856483e-06,
      "loss": 0.5518,
      "step": 8481
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0490937232971191,
      "learning_rate": 4.6381408610051605e-06,
      "loss": 0.5321,
      "step": 8482
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0915193557739258,
      "learning_rate": 4.63711756539047e-06,
      "loss": 0.5386,
      "step": 8483
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9249290823936462,
      "learning_rate": 4.636094285055497e-06,
      "loss": 0.4814,
      "step": 8484
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9355958700180054,
      "learning_rate": 4.635071020043326e-06,
      "loss": 0.4708,
      "step": 8485
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.037896752357483,
      "learning_rate": 4.634047770397044e-06,
      "loss": 0.5087,
      "step": 8486
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0038915872573853,
      "learning_rate": 4.633024536159739e-06,
      "loss": 0.4977,
      "step": 8487
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0237152576446533,
      "learning_rate": 4.632001317374495e-06,
      "loss": 0.5282,
      "step": 8488
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.994515597820282,
      "learning_rate": 4.630978114084394e-06,
      "loss": 0.4939,
      "step": 8489
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0709589719772339,
      "learning_rate": 4.629954926332522e-06,
      "loss": 0.5543,
      "step": 8490
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0311912298202515,
      "learning_rate": 4.628931754161959e-06,
      "loss": 0.5189,
      "step": 8491
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9902054071426392,
      "learning_rate": 4.62790859761579e-06,
      "loss": 0.5191,
      "step": 8492
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0637824535369873,
      "learning_rate": 4.626885456737095e-06,
      "loss": 0.5301,
      "step": 8493
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0846922397613525,
      "learning_rate": 4.625862331568957e-06,
      "loss": 0.5169,
      "step": 8494
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1294918060302734,
      "learning_rate": 4.624839222154453e-06,
      "loss": 0.5251,
      "step": 8495
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0639559030532837,
      "learning_rate": 4.623816128536665e-06,
      "loss": 0.5209,
      "step": 8496
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0102713108062744,
      "learning_rate": 4.6227930507586705e-06,
      "loss": 0.4967,
      "step": 8497
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1638081073760986,
      "learning_rate": 4.62176998886355e-06,
      "loss": 0.5388,
      "step": 8498
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9671724438667297,
      "learning_rate": 4.620746942894377e-06,
      "loss": 0.5346,
      "step": 8499
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0830614566802979,
      "learning_rate": 4.619723912894232e-06,
      "loss": 0.521,
      "step": 8500
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0154008865356445,
      "learning_rate": 4.618700898906191e-06,
      "loss": 0.4904,
      "step": 8501
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.967204213142395,
      "learning_rate": 4.6176779009733295e-06,
      "loss": 0.477,
      "step": 8502
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9524750709533691,
      "learning_rate": 4.616654919138719e-06,
      "loss": 0.4654,
      "step": 8503
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.08669114112854,
      "learning_rate": 4.6156319534454365e-06,
      "loss": 0.5724,
      "step": 8504
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.047778844833374,
      "learning_rate": 4.614609003936558e-06,
      "loss": 0.5093,
      "step": 8505
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9351442456245422,
      "learning_rate": 4.613586070655152e-06,
      "loss": 0.5105,
      "step": 8506
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0905542373657227,
      "learning_rate": 4.612563153644292e-06,
      "loss": 0.5206,
      "step": 8507
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9523585438728333,
      "learning_rate": 4.6115402529470495e-06,
      "loss": 0.5325,
      "step": 8508
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0420037508010864,
      "learning_rate": 4.610517368606497e-06,
      "loss": 0.5033,
      "step": 8509
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0466331243515015,
      "learning_rate": 4.609494500665703e-06,
      "loss": 0.5121,
      "step": 8510
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1029798984527588,
      "learning_rate": 4.608471649167737e-06,
      "loss": 0.5173,
      "step": 8511
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0008111000061035,
      "learning_rate": 4.6074488141556656e-06,
      "loss": 0.5627,
      "step": 8512
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9665666818618774,
      "learning_rate": 4.606425995672562e-06,
      "loss": 0.5089,
      "step": 8513
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9968435764312744,
      "learning_rate": 4.605403193761489e-06,
      "loss": 0.5333,
      "step": 8514
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9083412885665894,
      "learning_rate": 4.604380408465516e-06,
      "loss": 0.4867,
      "step": 8515
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.011513352394104,
      "learning_rate": 4.603357639827705e-06,
      "loss": 0.545,
      "step": 8516
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0549994707107544,
      "learning_rate": 4.602334887891127e-06,
      "loss": 0.5561,
      "step": 8517
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.101017951965332,
      "learning_rate": 4.601312152698843e-06,
      "loss": 0.5535,
      "step": 8518
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0652989149093628,
      "learning_rate": 4.600289434293917e-06,
      "loss": 0.5307,
      "step": 8519
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9657049179077148,
      "learning_rate": 4.599266732719413e-06,
      "loss": 0.5296,
      "step": 8520
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9676050543785095,
      "learning_rate": 4.598244048018391e-06,
      "loss": 0.5069,
      "step": 8521
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0080175399780273,
      "learning_rate": 4.5972213802339165e-06,
      "loss": 0.5375,
      "step": 8522
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0194203853607178,
      "learning_rate": 4.596198729409047e-06,
      "loss": 0.5634,
      "step": 8523
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2580246925354004,
      "learning_rate": 4.5951760955868455e-06,
      "loss": 0.535,
      "step": 8524
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0555814504623413,
      "learning_rate": 4.594153478810368e-06,
      "loss": 0.5286,
      "step": 8525
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0560455322265625,
      "learning_rate": 4.593130879122678e-06,
      "loss": 0.4989,
      "step": 8526
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.077176809310913,
      "learning_rate": 4.59210829656683e-06,
      "loss": 0.5223,
      "step": 8527
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9897431135177612,
      "learning_rate": 4.591085731185885e-06,
      "loss": 0.487,
      "step": 8528
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.020951271057129,
      "learning_rate": 4.590063183022894e-06,
      "loss": 0.5112,
      "step": 8529
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9736301302909851,
      "learning_rate": 4.589040652120919e-06,
      "loss": 0.521,
      "step": 8530
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1638489961624146,
      "learning_rate": 4.588018138523011e-06,
      "loss": 0.5338,
      "step": 8531
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0394287109375,
      "learning_rate": 4.5869956422722274e-06,
      "loss": 0.5025,
      "step": 8532
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0351699590682983,
      "learning_rate": 4.585973163411618e-06,
      "loss": 0.5207,
      "step": 8533
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9668200612068176,
      "learning_rate": 4.584950701984241e-06,
      "loss": 0.4933,
      "step": 8534
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9660054445266724,
      "learning_rate": 4.583928258033145e-06,
      "loss": 0.5076,
      "step": 8535
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0498343706130981,
      "learning_rate": 4.5829058316013835e-06,
      "loss": 0.4908,
      "step": 8536
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.191595196723938,
      "learning_rate": 4.581883422732007e-06,
      "loss": 0.5428,
      "step": 8537
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.024022102355957,
      "learning_rate": 4.580861031468062e-06,
      "loss": 0.4978,
      "step": 8538
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9271550178527832,
      "learning_rate": 4.579838657852603e-06,
      "loss": 0.5192,
      "step": 8539
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9976838827133179,
      "learning_rate": 4.578816301928677e-06,
      "loss": 0.5176,
      "step": 8540
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.048948049545288,
      "learning_rate": 4.577793963739331e-06,
      "loss": 0.5165,
      "step": 8541
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9421103000640869,
      "learning_rate": 4.576771643327611e-06,
      "loss": 0.4898,
      "step": 8542
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9716400504112244,
      "learning_rate": 4.575749340736565e-06,
      "loss": 0.5073,
      "step": 8543
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0311450958251953,
      "learning_rate": 4.57472705600924e-06,
      "loss": 0.4947,
      "step": 8544
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0700623989105225,
      "learning_rate": 4.573704789188679e-06,
      "loss": 0.5679,
      "step": 8545
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9896937608718872,
      "learning_rate": 4.5726825403179245e-06,
      "loss": 0.5628,
      "step": 8546
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.929032027721405,
      "learning_rate": 4.571660309440022e-06,
      "loss": 0.4865,
      "step": 8547
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.082908034324646,
      "learning_rate": 4.570638096598016e-06,
      "loss": 0.5504,
      "step": 8548
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0592923164367676,
      "learning_rate": 4.569615901834946e-06,
      "loss": 0.5115,
      "step": 8549
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1382687091827393,
      "learning_rate": 4.568593725193852e-06,
      "loss": 0.5198,
      "step": 8550
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9509519338607788,
      "learning_rate": 4.567571566717774e-06,
      "loss": 0.5091,
      "step": 8551
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9528672099113464,
      "learning_rate": 4.566549426449755e-06,
      "loss": 0.477,
      "step": 8552
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0432177782058716,
      "learning_rate": 4.565527304432833e-06,
      "loss": 0.5301,
      "step": 8553
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9746152758598328,
      "learning_rate": 4.564505200710042e-06,
      "loss": 0.4672,
      "step": 8554
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9885959029197693,
      "learning_rate": 4.5634831153244215e-06,
      "loss": 0.5086,
      "step": 8555
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.022283911705017,
      "learning_rate": 4.562461048319011e-06,
      "loss": 0.4954,
      "step": 8556
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1526809930801392,
      "learning_rate": 4.561438999736844e-06,
      "loss": 0.5885,
      "step": 8557
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0040638446807861,
      "learning_rate": 4.5604169696209535e-06,
      "loss": 0.5376,
      "step": 8558
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9949778318405151,
      "learning_rate": 4.559394958014375e-06,
      "loss": 0.5441,
      "step": 8559
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0233004093170166,
      "learning_rate": 4.558372964960142e-06,
      "loss": 0.5206,
      "step": 8560
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9786447882652283,
      "learning_rate": 4.557350990501288e-06,
      "loss": 0.4969,
      "step": 8561
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.03334641456604,
      "learning_rate": 4.556329034680845e-06,
      "loss": 0.5064,
      "step": 8562
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0475358963012695,
      "learning_rate": 4.55530709754184e-06,
      "loss": 0.536,
      "step": 8563
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0991499423980713,
      "learning_rate": 4.5542851791273085e-06,
      "loss": 0.5446,
      "step": 8564
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9369568228721619,
      "learning_rate": 4.5532632794802766e-06,
      "loss": 0.4684,
      "step": 8565
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9981191754341125,
      "learning_rate": 4.5522413986437745e-06,
      "loss": 0.5202,
      "step": 8566
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1952184438705444,
      "learning_rate": 4.55121953666083e-06,
      "loss": 0.503,
      "step": 8567
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0004924535751343,
      "learning_rate": 4.550197693574468e-06,
      "loss": 0.5004,
      "step": 8568
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.096377968788147,
      "learning_rate": 4.549175869427717e-06,
      "loss": 0.5176,
      "step": 8569
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9786026477813721,
      "learning_rate": 4.548154064263603e-06,
      "loss": 0.4952,
      "step": 8570
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1313389539718628,
      "learning_rate": 4.547132278125149e-06,
      "loss": 0.5221,
      "step": 8571
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0534061193466187,
      "learning_rate": 4.546110511055377e-06,
      "loss": 0.4931,
      "step": 8572
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0496286153793335,
      "learning_rate": 4.545088763097314e-06,
      "loss": 0.5005,
      "step": 8573
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0372889041900635,
      "learning_rate": 4.544067034293982e-06,
      "loss": 0.5127,
      "step": 8574
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0367099046707153,
      "learning_rate": 4.543045324688401e-06,
      "loss": 0.4881,
      "step": 8575
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.008887529373169,
      "learning_rate": 4.542023634323589e-06,
      "loss": 0.5153,
      "step": 8576
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0606989860534668,
      "learning_rate": 4.54100196324257e-06,
      "loss": 0.5216,
      "step": 8577
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0111998319625854,
      "learning_rate": 4.539980311488363e-06,
      "loss": 0.4682,
      "step": 8578
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9441483616828918,
      "learning_rate": 4.538958679103984e-06,
      "loss": 0.4991,
      "step": 8579
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9652086496353149,
      "learning_rate": 4.5379370661324495e-06,
      "loss": 0.4923,
      "step": 8580
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0206347703933716,
      "learning_rate": 4.536915472616779e-06,
      "loss": 0.5068,
      "step": 8581
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0516248941421509,
      "learning_rate": 4.535893898599988e-06,
      "loss": 0.4905,
      "step": 8582
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0783042907714844,
      "learning_rate": 4.53487234412509e-06,
      "loss": 0.5042,
      "step": 8583
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0359718799591064,
      "learning_rate": 4.533850809235099e-06,
      "loss": 0.5514,
      "step": 8584
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.06660795211792,
      "learning_rate": 4.532829293973028e-06,
      "loss": 0.5476,
      "step": 8585
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0299817323684692,
      "learning_rate": 4.531807798381892e-06,
      "loss": 0.5157,
      "step": 8586
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9901638627052307,
      "learning_rate": 4.5307863225047e-06,
      "loss": 0.4882,
      "step": 8587
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1186354160308838,
      "learning_rate": 4.529764866384464e-06,
      "loss": 0.5465,
      "step": 8588
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9637424349784851,
      "learning_rate": 4.528743430064192e-06,
      "loss": 0.4752,
      "step": 8589
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.017280101776123,
      "learning_rate": 4.527722013586897e-06,
      "loss": 0.5137,
      "step": 8590
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.065182089805603,
      "learning_rate": 4.5267006169955855e-06,
      "loss": 0.5726,
      "step": 8591
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0572706460952759,
      "learning_rate": 4.525679240333262e-06,
      "loss": 0.5353,
      "step": 8592
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0447081327438354,
      "learning_rate": 4.524657883642936e-06,
      "loss": 0.5573,
      "step": 8593
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9903708100318909,
      "learning_rate": 4.5236365469676144e-06,
      "loss": 0.4939,
      "step": 8594
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9923166632652283,
      "learning_rate": 4.522615230350302e-06,
      "loss": 0.5644,
      "step": 8595
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9883496761322021,
      "learning_rate": 4.521593933833998e-06,
      "loss": 0.4849,
      "step": 8596
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0408037900924683,
      "learning_rate": 4.520572657461712e-06,
      "loss": 0.488,
      "step": 8597
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.059937596321106,
      "learning_rate": 4.519551401276441e-06,
      "loss": 0.5129,
      "step": 8598
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9665573239326477,
      "learning_rate": 4.518530165321192e-06,
      "loss": 0.5119,
      "step": 8599
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0387489795684814,
      "learning_rate": 4.517508949638961e-06,
      "loss": 0.525,
      "step": 8600
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9976455569267273,
      "learning_rate": 4.516487754272751e-06,
      "loss": 0.5312,
      "step": 8601
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.044742226600647,
      "learning_rate": 4.515466579265557e-06,
      "loss": 0.5311,
      "step": 8602
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0177574157714844,
      "learning_rate": 4.5144454246603816e-06,
      "loss": 0.5094,
      "step": 8603
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0687826871871948,
      "learning_rate": 4.51342429050022e-06,
      "loss": 0.5238,
      "step": 8604
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0799012184143066,
      "learning_rate": 4.51240317682807e-06,
      "loss": 0.5797,
      "step": 8605
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9996159672737122,
      "learning_rate": 4.5113820836869234e-06,
      "loss": 0.5035,
      "step": 8606
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1689174175262451,
      "learning_rate": 4.51036101111978e-06,
      "loss": 0.5899,
      "step": 8607
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.946363091468811,
      "learning_rate": 4.509339959169629e-06,
      "loss": 0.4894,
      "step": 8608
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1414538621902466,
      "learning_rate": 4.508318927879468e-06,
      "loss": 0.6091,
      "step": 8609
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0159684419631958,
      "learning_rate": 4.507297917292284e-06,
      "loss": 0.4701,
      "step": 8610
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9742702841758728,
      "learning_rate": 4.506276927451072e-06,
      "loss": 0.4889,
      "step": 8611
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0368071794509888,
      "learning_rate": 4.505255958398821e-06,
      "loss": 0.5436,
      "step": 8612
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.145423173904419,
      "learning_rate": 4.504235010178521e-06,
      "loss": 0.518,
      "step": 8613
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.131622314453125,
      "learning_rate": 4.503214082833161e-06,
      "loss": 0.5244,
      "step": 8614
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0988490581512451,
      "learning_rate": 4.502193176405724e-06,
      "loss": 0.5126,
      "step": 8615
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0306994915008545,
      "learning_rate": 4.501172290939203e-06,
      "loss": 0.4994,
      "step": 8616
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0991085767745972,
      "learning_rate": 4.5001514264765826e-06,
      "loss": 0.5116,
      "step": 8617
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0253667831420898,
      "learning_rate": 4.499130583060845e-06,
      "loss": 0.4981,
      "step": 8618
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9436437487602234,
      "learning_rate": 4.4981097607349764e-06,
      "loss": 0.5227,
      "step": 8619
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0218784809112549,
      "learning_rate": 4.49708895954196e-06,
      "loss": 0.4794,
      "step": 8620
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0793664455413818,
      "learning_rate": 4.496068179524778e-06,
      "loss": 0.5312,
      "step": 8621
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9964872598648071,
      "learning_rate": 4.495047420726412e-06,
      "loss": 0.5461,
      "step": 8622
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.032078742980957,
      "learning_rate": 4.494026683189843e-06,
      "loss": 0.5549,
      "step": 8623
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1988216638565063,
      "learning_rate": 4.493005966958049e-06,
      "loss": 0.5169,
      "step": 8624
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9762097001075745,
      "learning_rate": 4.4919852720740115e-06,
      "loss": 0.4873,
      "step": 8625
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0649491548538208,
      "learning_rate": 4.490964598580706e-06,
      "loss": 0.5566,
      "step": 8626
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.026347279548645,
      "learning_rate": 4.489943946521111e-06,
      "loss": 0.512,
      "step": 8627
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.117616057395935,
      "learning_rate": 4.4889233159382e-06,
      "loss": 0.5275,
      "step": 8628
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0276168584823608,
      "learning_rate": 4.487902706874954e-06,
      "loss": 0.5575,
      "step": 8629
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2161848545074463,
      "learning_rate": 4.486882119374341e-06,
      "loss": 0.5479,
      "step": 8630
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1287305355072021,
      "learning_rate": 4.485861553479338e-06,
      "loss": 0.5908,
      "step": 8631
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1614528894424438,
      "learning_rate": 4.484841009232914e-06,
      "loss": 0.5418,
      "step": 8632
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0197196006774902,
      "learning_rate": 4.483820486678047e-06,
      "loss": 0.5131,
      "step": 8633
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.054356575012207,
      "learning_rate": 4.482799985857701e-06,
      "loss": 0.5642,
      "step": 8634
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.007958173751831,
      "learning_rate": 4.48177950681485e-06,
      "loss": 0.5021,
      "step": 8635
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0405101776123047,
      "learning_rate": 4.480759049592458e-06,
      "loss": 0.5466,
      "step": 8636
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0778744220733643,
      "learning_rate": 4.4797386142335e-06,
      "loss": 0.5221,
      "step": 8637
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1062657833099365,
      "learning_rate": 4.478718200780936e-06,
      "loss": 0.5462,
      "step": 8638
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9767703413963318,
      "learning_rate": 4.477697809277738e-06,
      "loss": 0.4649,
      "step": 8639
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.03668212890625,
      "learning_rate": 4.476677439766865e-06,
      "loss": 0.5115,
      "step": 8640
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1296067237854004,
      "learning_rate": 4.475657092291287e-06,
      "loss": 0.5405,
      "step": 8641
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1303768157958984,
      "learning_rate": 4.4746367668939646e-06,
      "loss": 0.5346,
      "step": 8642
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9903587698936462,
      "learning_rate": 4.4736164636178605e-06,
      "loss": 0.4757,
      "step": 8643
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0310946702957153,
      "learning_rate": 4.472596182505936e-06,
      "loss": 0.5133,
      "step": 8644
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1160991191864014,
      "learning_rate": 4.47157592360115e-06,
      "loss": 0.5269,
      "step": 8645
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.086572289466858,
      "learning_rate": 4.470555686946464e-06,
      "loss": 0.4952,
      "step": 8646
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0365424156188965,
      "learning_rate": 4.469535472584837e-06,
      "loss": 0.5276,
      "step": 8647
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0874872207641602,
      "learning_rate": 4.468515280559227e-06,
      "loss": 0.5314,
      "step": 8648
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0258994102478027,
      "learning_rate": 4.467495110912587e-06,
      "loss": 0.551,
      "step": 8649
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9841521382331848,
      "learning_rate": 4.466474963687876e-06,
      "loss": 0.5573,
      "step": 8650
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9820963144302368,
      "learning_rate": 4.46545483892805e-06,
      "loss": 0.5394,
      "step": 8651
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1955668926239014,
      "learning_rate": 4.464434736676061e-06,
      "loss": 0.5036,
      "step": 8652
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0836960077285767,
      "learning_rate": 4.46341465697486e-06,
      "loss": 0.5425,
      "step": 8653
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0003873109817505,
      "learning_rate": 4.462394599867402e-06,
      "loss": 0.4825,
      "step": 8654
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.045983910560608,
      "learning_rate": 4.461374565396638e-06,
      "loss": 0.4935,
      "step": 8655
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0779129266738892,
      "learning_rate": 4.460354553605518e-06,
      "loss": 0.5392,
      "step": 8656
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1067196130752563,
      "learning_rate": 4.459334564536988e-06,
      "loss": 0.5776,
      "step": 8657
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0516122579574585,
      "learning_rate": 4.458314598234e-06,
      "loss": 0.515,
      "step": 8658
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.964877188205719,
      "learning_rate": 4.4572946547395e-06,
      "loss": 0.4937,
      "step": 8659
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.990656316280365,
      "learning_rate": 4.456274734096436e-06,
      "loss": 0.5204,
      "step": 8660
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.071433424949646,
      "learning_rate": 4.455254836347749e-06,
      "loss": 0.5563,
      "step": 8661
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9665083289146423,
      "learning_rate": 4.454234961536384e-06,
      "loss": 0.525,
      "step": 8662
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9684353470802307,
      "learning_rate": 4.45321510970529e-06,
      "loss": 0.5188,
      "step": 8663
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.012878656387329,
      "learning_rate": 4.452195280897405e-06,
      "loss": 0.5086,
      "step": 8664
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.006637454032898,
      "learning_rate": 4.451175475155669e-06,
      "loss": 0.4902,
      "step": 8665
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0858285427093506,
      "learning_rate": 4.450155692523025e-06,
      "loss": 0.5124,
      "step": 8666
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0405857563018799,
      "learning_rate": 4.449135933042414e-06,
      "loss": 0.5258,
      "step": 8667
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1059043407440186,
      "learning_rate": 4.448116196756771e-06,
      "loss": 0.5433,
      "step": 8668
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9712663292884827,
      "learning_rate": 4.447096483709035e-06,
      "loss": 0.4664,
      "step": 8669
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0021990537643433,
      "learning_rate": 4.4460767939421425e-06,
      "loss": 0.525,
      "step": 8670
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0234445333480835,
      "learning_rate": 4.44505712749903e-06,
      "loss": 0.4958,
      "step": 8671
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0537341833114624,
      "learning_rate": 4.444037484422632e-06,
      "loss": 0.5475,
      "step": 8672
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.962066650390625,
      "learning_rate": 4.44301786475588e-06,
      "loss": 0.4978,
      "step": 8673
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0533349514007568,
      "learning_rate": 4.441998268541708e-06,
      "loss": 0.502,
      "step": 8674
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0353072881698608,
      "learning_rate": 4.440978695823049e-06,
      "loss": 0.5081,
      "step": 8675
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0396310091018677,
      "learning_rate": 4.439959146642833e-06,
      "loss": 0.5245,
      "step": 8676
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0958211421966553,
      "learning_rate": 4.4389396210439886e-06,
      "loss": 0.5164,
      "step": 8677
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0075626373291016,
      "learning_rate": 4.437920119069445e-06,
      "loss": 0.5302,
      "step": 8678
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1665135622024536,
      "learning_rate": 4.436900640762128e-06,
      "loss": 0.5332,
      "step": 8679
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0166640281677246,
      "learning_rate": 4.435881186164968e-06,
      "loss": 0.5204,
      "step": 8680
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0299077033996582,
      "learning_rate": 4.434861755320888e-06,
      "loss": 0.5005,
      "step": 8681
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9902662038803101,
      "learning_rate": 4.433842348272815e-06,
      "loss": 0.5081,
      "step": 8682
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0685404539108276,
      "learning_rate": 4.4328229650636676e-06,
      "loss": 0.5507,
      "step": 8683
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0103635787963867,
      "learning_rate": 4.431803605736376e-06,
      "loss": 0.4898,
      "step": 8684
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0012294054031372,
      "learning_rate": 4.430784270333855e-06,
      "loss": 0.5491,
      "step": 8685
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9294708371162415,
      "learning_rate": 4.429764958899031e-06,
      "loss": 0.4647,
      "step": 8686
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9999809861183167,
      "learning_rate": 4.428745671474818e-06,
      "loss": 0.5089,
      "step": 8687
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1061701774597168,
      "learning_rate": 4.427726408104139e-06,
      "loss": 0.5441,
      "step": 8688
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0385311841964722,
      "learning_rate": 4.42670716882991e-06,
      "loss": 0.5386,
      "step": 8689
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0141210556030273,
      "learning_rate": 4.4256879536950495e-06,
      "loss": 0.511,
      "step": 8690
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.968830406665802,
      "learning_rate": 4.4246687627424686e-06,
      "loss": 0.4983,
      "step": 8691
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0696613788604736,
      "learning_rate": 4.423649596015086e-06,
      "loss": 0.5288,
      "step": 8692
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9941192269325256,
      "learning_rate": 4.422630453555814e-06,
      "loss": 0.4934,
      "step": 8693
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0725018978118896,
      "learning_rate": 4.4216113354075654e-06,
      "loss": 0.5326,
      "step": 8694
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0443642139434814,
      "learning_rate": 4.420592241613251e-06,
      "loss": 0.5394,
      "step": 8695
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0298329591751099,
      "learning_rate": 4.4195731722157805e-06,
      "loss": 0.5062,
      "step": 8696
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0180301666259766,
      "learning_rate": 4.418554127258066e-06,
      "loss": 0.5403,
      "step": 8697
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0847681760787964,
      "learning_rate": 4.417535106783015e-06,
      "loss": 0.5775,
      "step": 8698
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9627909660339355,
      "learning_rate": 4.416516110833533e-06,
      "loss": 0.4586,
      "step": 8699
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.01035737991333,
      "learning_rate": 4.415497139452528e-06,
      "loss": 0.5501,
      "step": 8700
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.139830231666565,
      "learning_rate": 4.414478192682905e-06,
      "loss": 0.5466,
      "step": 8701
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0512480735778809,
      "learning_rate": 4.41345927056757e-06,
      "loss": 0.4649,
      "step": 8702
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0854779481887817,
      "learning_rate": 4.4124403731494235e-06,
      "loss": 0.5747,
      "step": 8703
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9707875847816467,
      "learning_rate": 4.4114215004713665e-06,
      "loss": 0.5162,
      "step": 8704
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0341905355453491,
      "learning_rate": 4.410402652576307e-06,
      "loss": 0.4722,
      "step": 8705
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0561015605926514,
      "learning_rate": 4.409383829507139e-06,
      "loss": 0.4789,
      "step": 8706
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0236762762069702,
      "learning_rate": 4.408365031306763e-06,
      "loss": 0.4962,
      "step": 8707
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0971812009811401,
      "learning_rate": 4.407346258018078e-06,
      "loss": 0.528,
      "step": 8708
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.047594428062439,
      "learning_rate": 4.4063275096839785e-06,
      "loss": 0.4914,
      "step": 8709
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.137151837348938,
      "learning_rate": 4.405308786347365e-06,
      "loss": 0.5227,
      "step": 8710
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0921539068222046,
      "learning_rate": 4.404290088051128e-06,
      "loss": 0.5255,
      "step": 8711
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9931677579879761,
      "learning_rate": 4.403271414838164e-06,
      "loss": 0.5069,
      "step": 8712
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0669801235198975,
      "learning_rate": 4.402252766751363e-06,
      "loss": 0.5159,
      "step": 8713
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0193895101547241,
      "learning_rate": 4.401234143833621e-06,
      "loss": 0.5175,
      "step": 8714
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9947845935821533,
      "learning_rate": 4.400215546127825e-06,
      "loss": 0.51,
      "step": 8715
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9431485533714294,
      "learning_rate": 4.399196973676867e-06,
      "loss": 0.4892,
      "step": 8716
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.080181360244751,
      "learning_rate": 4.398178426523632e-06,
      "loss": 0.5802,
      "step": 8717
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0866167545318604,
      "learning_rate": 4.3971599047110116e-06,
      "loss": 0.514,
      "step": 8718
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0306841135025024,
      "learning_rate": 4.3961414082818904e-06,
      "loss": 0.5033,
      "step": 8719
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0695841312408447,
      "learning_rate": 4.395122937279154e-06,
      "loss": 0.5483,
      "step": 8720
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0567008256912231,
      "learning_rate": 4.394104491745686e-06,
      "loss": 0.5153,
      "step": 8721
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.904281735420227,
      "learning_rate": 4.393086071724371e-06,
      "loss": 0.4623,
      "step": 8722
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.081354022026062,
      "learning_rate": 4.392067677258089e-06,
      "loss": 0.5676,
      "step": 8723
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0314452648162842,
      "learning_rate": 4.391049308389723e-06,
      "loss": 0.4855,
      "step": 8724
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0365545749664307,
      "learning_rate": 4.390030965162153e-06,
      "loss": 0.5059,
      "step": 8725
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9358240962028503,
      "learning_rate": 4.389012647618255e-06,
      "loss": 0.4984,
      "step": 8726
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0348268747329712,
      "learning_rate": 4.387994355800909e-06,
      "loss": 0.5419,
      "step": 8727
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9691881537437439,
      "learning_rate": 4.386976089752994e-06,
      "loss": 0.4885,
      "step": 8728
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.110189437866211,
      "learning_rate": 4.385957849517383e-06,
      "loss": 0.5961,
      "step": 8729
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9775313138961792,
      "learning_rate": 4.384939635136948e-06,
      "loss": 0.514,
      "step": 8730
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0049653053283691,
      "learning_rate": 4.383921446654567e-06,
      "loss": 0.4882,
      "step": 8731
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.04422926902771,
      "learning_rate": 4.3829032841131116e-06,
      "loss": 0.53,
      "step": 8732
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1362097263336182,
      "learning_rate": 4.381885147555453e-06,
      "loss": 0.5841,
      "step": 8733
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1133431196212769,
      "learning_rate": 4.380867037024457e-06,
      "loss": 0.4776,
      "step": 8734
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.994042158126831,
      "learning_rate": 4.379848952562999e-06,
      "loss": 0.5495,
      "step": 8735
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9828497767448425,
      "learning_rate": 4.3788308942139435e-06,
      "loss": 0.5037,
      "step": 8736
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9757148623466492,
      "learning_rate": 4.3778128620201595e-06,
      "loss": 0.4795,
      "step": 8737
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0285186767578125,
      "learning_rate": 4.376794856024509e-06,
      "loss": 0.5662,
      "step": 8738
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.112226128578186,
      "learning_rate": 4.37577687626986e-06,
      "loss": 0.525,
      "step": 8739
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0480462312698364,
      "learning_rate": 4.374758922799076e-06,
      "loss": 0.5211,
      "step": 8740
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0427954196929932,
      "learning_rate": 4.373740995655019e-06,
      "loss": 0.5334,
      "step": 8741
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0622020959854126,
      "learning_rate": 4.372723094880549e-06,
      "loss": 0.5801,
      "step": 8742
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0892544984817505,
      "learning_rate": 4.371705220518526e-06,
      "loss": 0.5184,
      "step": 8743
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0577740669250488,
      "learning_rate": 4.3706873726118135e-06,
      "loss": 0.5136,
      "step": 8744
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.012571096420288,
      "learning_rate": 4.369669551203266e-06,
      "loss": 0.5262,
      "step": 8745
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0907323360443115,
      "learning_rate": 4.368651756335739e-06,
      "loss": 0.5274,
      "step": 8746
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0260123014450073,
      "learning_rate": 4.36763398805209e-06,
      "loss": 0.507,
      "step": 8747
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0460829734802246,
      "learning_rate": 4.366616246395177e-06,
      "loss": 0.5179,
      "step": 8748
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.047634482383728,
      "learning_rate": 4.365598531407849e-06,
      "loss": 0.542,
      "step": 8749
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1334526538848877,
      "learning_rate": 4.364580843132959e-06,
      "loss": 0.5015,
      "step": 8750
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9913628101348877,
      "learning_rate": 4.363563181613359e-06,
      "loss": 0.5075,
      "step": 8751
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.002979040145874,
      "learning_rate": 4.362545546891901e-06,
      "loss": 0.5112,
      "step": 8752
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1702262163162231,
      "learning_rate": 4.361527939011433e-06,
      "loss": 0.4818,
      "step": 8753
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0306715965270996,
      "learning_rate": 4.360510358014801e-06,
      "loss": 0.5553,
      "step": 8754
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1320462226867676,
      "learning_rate": 4.359492803944854e-06,
      "loss": 0.5412,
      "step": 8755
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.000876545906067,
      "learning_rate": 4.358475276844435e-06,
      "loss": 0.5132,
      "step": 8756
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0229440927505493,
      "learning_rate": 4.357457776756392e-06,
      "loss": 0.5563,
      "step": 8757
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0038920640945435,
      "learning_rate": 4.3564403037235666e-06,
      "loss": 0.4872,
      "step": 8758
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9732649922370911,
      "learning_rate": 4.355422857788802e-06,
      "loss": 0.4896,
      "step": 8759
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9638651609420776,
      "learning_rate": 4.3544054389949366e-06,
      "loss": 0.5102,
      "step": 8760
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0643752813339233,
      "learning_rate": 4.353388047384813e-06,
      "loss": 0.527,
      "step": 8761
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9771501421928406,
      "learning_rate": 4.35237068300127e-06,
      "loss": 0.5216,
      "step": 8762
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0805439949035645,
      "learning_rate": 4.351353345887145e-06,
      "loss": 0.5344,
      "step": 8763
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0408697128295898,
      "learning_rate": 4.350336036085272e-06,
      "loss": 0.5461,
      "step": 8764
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0978062152862549,
      "learning_rate": 4.349318753638491e-06,
      "loss": 0.5936,
      "step": 8765
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0012162923812866,
      "learning_rate": 4.348301498589632e-06,
      "loss": 0.4985,
      "step": 8766
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0011204481124878,
      "learning_rate": 4.347284270981531e-06,
      "loss": 0.5391,
      "step": 8767
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.055498480796814,
      "learning_rate": 4.346267070857017e-06,
      "loss": 0.5287,
      "step": 8768
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9990624189376831,
      "learning_rate": 4.3452498982589234e-06,
      "loss": 0.5037,
      "step": 8769
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0288701057434082,
      "learning_rate": 4.34423275323008e-06,
      "loss": 0.5123,
      "step": 8770
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0007120370864868,
      "learning_rate": 4.343215635813314e-06,
      "loss": 0.4961,
      "step": 8771
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9640586376190186,
      "learning_rate": 4.3421985460514515e-06,
      "loss": 0.5214,
      "step": 8772
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0430772304534912,
      "learning_rate": 4.341181483987319e-06,
      "loss": 0.4957,
      "step": 8773
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9861213564872742,
      "learning_rate": 4.340164449663745e-06,
      "loss": 0.4997,
      "step": 8774
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0308138132095337,
      "learning_rate": 4.33914744312355e-06,
      "loss": 0.4976,
      "step": 8775
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0044180154800415,
      "learning_rate": 4.338130464409556e-06,
      "loss": 0.4805,
      "step": 8776
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0167678594589233,
      "learning_rate": 4.3371135135645845e-06,
      "loss": 0.5643,
      "step": 8777
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0586427450180054,
      "learning_rate": 4.33609659063146e-06,
      "loss": 0.5242,
      "step": 8778
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9682163596153259,
      "learning_rate": 4.335079695652998e-06,
      "loss": 0.5393,
      "step": 8779
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1850287914276123,
      "learning_rate": 4.334062828672016e-06,
      "loss": 0.551,
      "step": 8780
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0070314407348633,
      "learning_rate": 4.3330459897313305e-06,
      "loss": 0.5023,
      "step": 8781
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0112802982330322,
      "learning_rate": 4.33202917887376e-06,
      "loss": 0.5113,
      "step": 8782
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9870576858520508,
      "learning_rate": 4.331012396142117e-06,
      "loss": 0.4985,
      "step": 8783
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0064136981964111,
      "learning_rate": 4.3299956415792145e-06,
      "loss": 0.5616,
      "step": 8784
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9818264842033386,
      "learning_rate": 4.328978915227866e-06,
      "loss": 0.5323,
      "step": 8785
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.033555269241333,
      "learning_rate": 4.327962217130878e-06,
      "loss": 0.5553,
      "step": 8786
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0492753982543945,
      "learning_rate": 4.326945547331065e-06,
      "loss": 0.4836,
      "step": 8787
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0379021167755127,
      "learning_rate": 4.325928905871233e-06,
      "loss": 0.5469,
      "step": 8788
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0607675313949585,
      "learning_rate": 4.324912292794192e-06,
      "loss": 0.5407,
      "step": 8789
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0625678300857544,
      "learning_rate": 4.323895708142742e-06,
      "loss": 0.4681,
      "step": 8790
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9727704524993896,
      "learning_rate": 4.322879151959695e-06,
      "loss": 0.5138,
      "step": 8791
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9812003374099731,
      "learning_rate": 4.321862624287851e-06,
      "loss": 0.5129,
      "step": 8792
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.045271873474121,
      "learning_rate": 4.320846125170012e-06,
      "loss": 0.5434,
      "step": 8793
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9572071433067322,
      "learning_rate": 4.31982965464898e-06,
      "loss": 0.501,
      "step": 8794
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0174649953842163,
      "learning_rate": 4.318813212767555e-06,
      "loss": 0.5048,
      "step": 8795
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.009214162826538,
      "learning_rate": 4.317796799568536e-06,
      "loss": 0.5297,
      "step": 8796
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0225062370300293,
      "learning_rate": 4.316780415094722e-06,
      "loss": 0.5042,
      "step": 8797
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0147311687469482,
      "learning_rate": 4.315764059388905e-06,
      "loss": 0.5224,
      "step": 8798
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0039790868759155,
      "learning_rate": 4.314747732493886e-06,
      "loss": 0.502,
      "step": 8799
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.007888913154602,
      "learning_rate": 4.313731434452455e-06,
      "loss": 0.5,
      "step": 8800
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0893665552139282,
      "learning_rate": 4.312715165307407e-06,
      "loss": 0.5831,
      "step": 8801
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.087306022644043,
      "learning_rate": 4.311698925101532e-06,
      "loss": 0.5141,
      "step": 8802
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0267753601074219,
      "learning_rate": 4.310682713877619e-06,
      "loss": 0.49,
      "step": 8803
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0899007320404053,
      "learning_rate": 4.30966653167846e-06,
      "loss": 0.5678,
      "step": 8804
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9892306923866272,
      "learning_rate": 4.308650378546843e-06,
      "loss": 0.5193,
      "step": 8805
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.994459867477417,
      "learning_rate": 4.3076342545255535e-06,
      "loss": 0.5005,
      "step": 8806
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0411771535873413,
      "learning_rate": 4.306618159657375e-06,
      "loss": 0.5296,
      "step": 8807
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0516984462738037,
      "learning_rate": 4.305602093985095e-06,
      "loss": 0.5262,
      "step": 8808
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0397151708602905,
      "learning_rate": 4.3045860575514955e-06,
      "loss": 0.5153,
      "step": 8809
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0043975114822388,
      "learning_rate": 4.303570050399358e-06,
      "loss": 0.53,
      "step": 8810
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9643057584762573,
      "learning_rate": 4.302554072571461e-06,
      "loss": 0.4874,
      "step": 8811
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0050950050354004,
      "learning_rate": 4.301538124110588e-06,
      "loss": 0.5355,
      "step": 8812
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0209351778030396,
      "learning_rate": 4.300522205059515e-06,
      "loss": 0.5249,
      "step": 8813
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0641778707504272,
      "learning_rate": 4.299506315461018e-06,
      "loss": 0.5145,
      "step": 8814
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0395965576171875,
      "learning_rate": 4.2984904553578725e-06,
      "loss": 0.5339,
      "step": 8815
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9726712107658386,
      "learning_rate": 4.297474624792853e-06,
      "loss": 0.5004,
      "step": 8816
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0722999572753906,
      "learning_rate": 4.296458823808735e-06,
      "loss": 0.5294,
      "step": 8817
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1132960319519043,
      "learning_rate": 4.295443052448288e-06,
      "loss": 0.5647,
      "step": 8818
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.040102481842041,
      "learning_rate": 4.294427310754283e-06,
      "loss": 0.4886,
      "step": 8819
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.065936803817749,
      "learning_rate": 4.293411598769487e-06,
      "loss": 0.5641,
      "step": 8820
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9980764389038086,
      "learning_rate": 4.292395916536674e-06,
      "loss": 0.5349,
      "step": 8821
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1201890707015991,
      "learning_rate": 4.291380264098607e-06,
      "loss": 0.5481,
      "step": 8822
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1285412311553955,
      "learning_rate": 4.290364641498051e-06,
      "loss": 0.4842,
      "step": 8823
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0192679166793823,
      "learning_rate": 4.28934904877777e-06,
      "loss": 0.5005,
      "step": 8824
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0406442880630493,
      "learning_rate": 4.288333485980531e-06,
      "loss": 0.5226,
      "step": 8825
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.028008222579956,
      "learning_rate": 4.287317953149092e-06,
      "loss": 0.5267,
      "step": 8826
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9059807062149048,
      "learning_rate": 4.2863024503262146e-06,
      "loss": 0.4318,
      "step": 8827
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1068507432937622,
      "learning_rate": 4.285286977554657e-06,
      "loss": 0.5515,
      "step": 8828
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0708606243133545,
      "learning_rate": 4.284271534877181e-06,
      "loss": 0.5372,
      "step": 8829
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1273646354675293,
      "learning_rate": 4.283256122336539e-06,
      "loss": 0.5746,
      "step": 8830
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0696847438812256,
      "learning_rate": 4.28224073997549e-06,
      "loss": 0.5372,
      "step": 8831
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0077567100524902,
      "learning_rate": 4.281225387836786e-06,
      "loss": 0.4877,
      "step": 8832
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9975242018699646,
      "learning_rate": 4.280210065963179e-06,
      "loss": 0.4981,
      "step": 8833
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0427783727645874,
      "learning_rate": 4.279194774397422e-06,
      "loss": 0.4912,
      "step": 8834
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.032755732536316,
      "learning_rate": 4.278179513182268e-06,
      "loss": 0.5291,
      "step": 8835
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.051928162574768,
      "learning_rate": 4.2771642823604635e-06,
      "loss": 0.5167,
      "step": 8836
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0293514728546143,
      "learning_rate": 4.276149081974754e-06,
      "loss": 0.5173,
      "step": 8837
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9974006414413452,
      "learning_rate": 4.275133912067889e-06,
      "loss": 0.459,
      "step": 8838
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.059767723083496,
      "learning_rate": 4.274118772682615e-06,
      "loss": 0.4941,
      "step": 8839
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0430185794830322,
      "learning_rate": 4.273103663861675e-06,
      "loss": 0.5184,
      "step": 8840
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0201420783996582,
      "learning_rate": 4.272088585647808e-06,
      "loss": 0.4704,
      "step": 8841
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.027017593383789,
      "learning_rate": 4.27107353808376e-06,
      "loss": 0.5296,
      "step": 8842
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0478624105453491,
      "learning_rate": 4.2700585212122705e-06,
      "loss": 0.5028,
      "step": 8843
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0227540731430054,
      "learning_rate": 4.269043535076077e-06,
      "loss": 0.4999,
      "step": 8844
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.003603219985962,
      "learning_rate": 4.2680285797179155e-06,
      "loss": 0.5115,
      "step": 8845
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0649179220199585,
      "learning_rate": 4.267013655180526e-06,
      "loss": 0.5591,
      "step": 8846
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0227795839309692,
      "learning_rate": 4.265998761506641e-06,
      "loss": 0.5373,
      "step": 8847
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9785618185997009,
      "learning_rate": 4.264983898738996e-06,
      "loss": 0.4689,
      "step": 8848
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.12110435962677,
      "learning_rate": 4.263969066920321e-06,
      "loss": 0.5042,
      "step": 8849
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0460258722305298,
      "learning_rate": 4.262954266093347e-06,
      "loss": 0.5292,
      "step": 8850
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9611269235610962,
      "learning_rate": 4.261939496300807e-06,
      "loss": 0.4752,
      "step": 8851
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9734213948249817,
      "learning_rate": 4.260924757585427e-06,
      "loss": 0.5075,
      "step": 8852
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.006095290184021,
      "learning_rate": 4.259910049989933e-06,
      "loss": 0.4719,
      "step": 8853
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9881126284599304,
      "learning_rate": 4.258895373557051e-06,
      "loss": 0.4989,
      "step": 8854
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9929634928703308,
      "learning_rate": 4.25788072832951e-06,
      "loss": 0.5272,
      "step": 8855
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.017573356628418,
      "learning_rate": 4.256866114350029e-06,
      "loss": 0.4803,
      "step": 8856
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0489202737808228,
      "learning_rate": 4.25585153166133e-06,
      "loss": 0.597,
      "step": 8857
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0458786487579346,
      "learning_rate": 4.254836980306134e-06,
      "loss": 0.5275,
      "step": 8858
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.017008662223816,
      "learning_rate": 4.253822460327162e-06,
      "loss": 0.5081,
      "step": 8859
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1114718914031982,
      "learning_rate": 4.25280797176713e-06,
      "loss": 0.5291,
      "step": 8860
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1073929071426392,
      "learning_rate": 4.251793514668754e-06,
      "loss": 0.5235,
      "step": 8861
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9434214234352112,
      "learning_rate": 4.250779089074752e-06,
      "loss": 0.4956,
      "step": 8862
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.001192331314087,
      "learning_rate": 4.249764695027833e-06,
      "loss": 0.5378,
      "step": 8863
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0642035007476807,
      "learning_rate": 4.248750332570716e-06,
      "loss": 0.5893,
      "step": 8864
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0944091081619263,
      "learning_rate": 4.247736001746108e-06,
      "loss": 0.5033,
      "step": 8865
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0930684804916382,
      "learning_rate": 4.246721702596721e-06,
      "loss": 0.5263,
      "step": 8866
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0492407083511353,
      "learning_rate": 4.24570743516526e-06,
      "loss": 0.5593,
      "step": 8867
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0513110160827637,
      "learning_rate": 4.2446931994944375e-06,
      "loss": 0.5257,
      "step": 8868
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0380091667175293,
      "learning_rate": 4.243678995626955e-06,
      "loss": 0.5087,
      "step": 8869
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9944601058959961,
      "learning_rate": 4.242664823605521e-06,
      "loss": 0.4907,
      "step": 8870
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.045111894607544,
      "learning_rate": 4.241650683472834e-06,
      "loss": 0.4721,
      "step": 8871
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9547770023345947,
      "learning_rate": 4.240636575271601e-06,
      "loss": 0.4366,
      "step": 8872
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9458922147750854,
      "learning_rate": 4.239622499044519e-06,
      "loss": 0.4814,
      "step": 8873
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0829119682312012,
      "learning_rate": 4.23860845483429e-06,
      "loss": 0.5168,
      "step": 8874
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.032349705696106,
      "learning_rate": 4.237594442683607e-06,
      "loss": 0.5417,
      "step": 8875
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9982881546020508,
      "learning_rate": 4.236580462635173e-06,
      "loss": 0.5178,
      "step": 8876
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0686936378479004,
      "learning_rate": 4.235566514731678e-06,
      "loss": 0.5355,
      "step": 8877
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0420191287994385,
      "learning_rate": 4.23455259901582e-06,
      "loss": 0.5431,
      "step": 8878
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9994642734527588,
      "learning_rate": 4.2335387155302885e-06,
      "loss": 0.5152,
      "step": 8879
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0132908821105957,
      "learning_rate": 4.232524864317773e-06,
      "loss": 0.4807,
      "step": 8880
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9887734055519104,
      "learning_rate": 4.231511045420967e-06,
      "loss": 0.5242,
      "step": 8881
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.95689857006073,
      "learning_rate": 4.230497258882559e-06,
      "loss": 0.4774,
      "step": 8882
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.124097466468811,
      "learning_rate": 4.229483504745233e-06,
      "loss": 0.5405,
      "step": 8883
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9804500937461853,
      "learning_rate": 4.228469783051676e-06,
      "loss": 0.5318,
      "step": 8884
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0371235609054565,
      "learning_rate": 4.227456093844573e-06,
      "loss": 0.5059,
      "step": 8885
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.142540693283081,
      "learning_rate": 4.226442437166607e-06,
      "loss": 0.4845,
      "step": 8886
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0167458057403564,
      "learning_rate": 4.225428813060459e-06,
      "loss": 0.5167,
      "step": 8887
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0186748504638672,
      "learning_rate": 4.224415221568807e-06,
      "loss": 0.5142,
      "step": 8888
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0175448656082153,
      "learning_rate": 4.223401662734333e-06,
      "loss": 0.4768,
      "step": 8889
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9420079588890076,
      "learning_rate": 4.222388136599715e-06,
      "loss": 0.5242,
      "step": 8890
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0170434713363647,
      "learning_rate": 4.221374643207626e-06,
      "loss": 0.5128,
      "step": 8891
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9713902473449707,
      "learning_rate": 4.220361182600742e-06,
      "loss": 0.5407,
      "step": 8892
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0836924314498901,
      "learning_rate": 4.219347754821737e-06,
      "loss": 0.4982,
      "step": 8893
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0609179735183716,
      "learning_rate": 4.218334359913283e-06,
      "loss": 0.4988,
      "step": 8894
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9709429144859314,
      "learning_rate": 4.217320997918048e-06,
      "loss": 0.5252,
      "step": 8895
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0228431224822998,
      "learning_rate": 4.216307668878706e-06,
      "loss": 0.5595,
      "step": 8896
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0736753940582275,
      "learning_rate": 4.2152943728379185e-06,
      "loss": 0.5469,
      "step": 8897
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9818423986434937,
      "learning_rate": 4.214281109838357e-06,
      "loss": 0.4778,
      "step": 8898
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0688577890396118,
      "learning_rate": 4.213267879922685e-06,
      "loss": 0.5599,
      "step": 8899
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9578925967216492,
      "learning_rate": 4.212254683133565e-06,
      "loss": 0.4579,
      "step": 8900
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0140105485916138,
      "learning_rate": 4.2112415195136585e-06,
      "loss": 0.504,
      "step": 8901
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.115033507347107,
      "learning_rate": 4.21022838910563e-06,
      "loss": 0.498,
      "step": 8902
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0247352123260498,
      "learning_rate": 4.209215291952135e-06,
      "loss": 0.5112,
      "step": 8903
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0588626861572266,
      "learning_rate": 4.208202228095835e-06,
      "loss": 0.5315,
      "step": 8904
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0949006080627441,
      "learning_rate": 4.207189197579382e-06,
      "loss": 0.487,
      "step": 8905
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9179094433784485,
      "learning_rate": 4.2061762004454365e-06,
      "loss": 0.4714,
      "step": 8906
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9899871349334717,
      "learning_rate": 4.2051632367366485e-06,
      "loss": 0.5705,
      "step": 8907
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0226097106933594,
      "learning_rate": 4.204150306495672e-06,
      "loss": 0.5763,
      "step": 8908
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9980384111404419,
      "learning_rate": 4.203137409765159e-06,
      "loss": 0.4652,
      "step": 8909
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9871333241462708,
      "learning_rate": 4.202124546587754e-06,
      "loss": 0.5151,
      "step": 8910
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9857281446456909,
      "learning_rate": 4.201111717006111e-06,
      "loss": 0.5353,
      "step": 8911
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0676463842391968,
      "learning_rate": 4.200098921062875e-06,
      "loss": 0.5313,
      "step": 8912
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0641769170761108,
      "learning_rate": 4.19908615880069e-06,
      "loss": 0.5544,
      "step": 8913
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0252957344055176,
      "learning_rate": 4.198073430262199e-06,
      "loss": 0.5368,
      "step": 8914
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9849558472633362,
      "learning_rate": 4.197060735490048e-06,
      "loss": 0.486,
      "step": 8915
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0962169170379639,
      "learning_rate": 4.196048074526876e-06,
      "loss": 0.5688,
      "step": 8916
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.074508786201477,
      "learning_rate": 4.195035447415324e-06,
      "loss": 0.5516,
      "step": 8917
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.011013388633728,
      "learning_rate": 4.194022854198026e-06,
      "loss": 0.5414,
      "step": 8918
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0027620792388916,
      "learning_rate": 4.193010294917624e-06,
      "loss": 0.4732,
      "step": 8919
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0046014785766602,
      "learning_rate": 4.1919977696167515e-06,
      "loss": 0.4781,
      "step": 8920
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9843215942382812,
      "learning_rate": 4.190985278338042e-06,
      "loss": 0.5117,
      "step": 8921
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0696995258331299,
      "learning_rate": 4.189972821124126e-06,
      "loss": 0.5469,
      "step": 8922
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.970366358757019,
      "learning_rate": 4.188960398017638e-06,
      "loss": 0.5035,
      "step": 8923
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9937092661857605,
      "learning_rate": 4.187948009061207e-06,
      "loss": 0.4778,
      "step": 8924
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9816207885742188,
      "learning_rate": 4.186935654297461e-06,
      "loss": 0.5206,
      "step": 8925
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9615216851234436,
      "learning_rate": 4.1859233337690245e-06,
      "loss": 0.4804,
      "step": 8926
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.017531394958496,
      "learning_rate": 4.1849110475185225e-06,
      "loss": 0.509,
      "step": 8927
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1661837100982666,
      "learning_rate": 4.183898795588584e-06,
      "loss": 0.5104,
      "step": 8928
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.035853624343872,
      "learning_rate": 4.1828865780218285e-06,
      "loss": 0.4876,
      "step": 8929
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9957482218742371,
      "learning_rate": 4.181874394860875e-06,
      "loss": 0.5066,
      "step": 8930
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0787684917449951,
      "learning_rate": 4.180862246148344e-06,
      "loss": 0.5226,
      "step": 8931
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9769288301467896,
      "learning_rate": 4.1798501319268565e-06,
      "loss": 0.4636,
      "step": 8932
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1175696849822998,
      "learning_rate": 4.178838052239027e-06,
      "loss": 0.5323,
      "step": 8933
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0295634269714355,
      "learning_rate": 4.177826007127468e-06,
      "loss": 0.4897,
      "step": 8934
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9559885859489441,
      "learning_rate": 4.176813996634796e-06,
      "loss": 0.5013,
      "step": 8935
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1553454399108887,
      "learning_rate": 4.175802020803624e-06,
      "loss": 0.5116,
      "step": 8936
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1046700477600098,
      "learning_rate": 4.174790079676563e-06,
      "loss": 0.49,
      "step": 8937
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0114251375198364,
      "learning_rate": 4.173778173296219e-06,
      "loss": 0.5084,
      "step": 8938
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.115433692932129,
      "learning_rate": 4.172766301705202e-06,
      "loss": 0.5343,
      "step": 8939
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0423344373703003,
      "learning_rate": 4.171754464946119e-06,
      "loss": 0.576,
      "step": 8940
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9994631409645081,
      "learning_rate": 4.170742663061575e-06,
      "loss": 0.5001,
      "step": 8941
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0563585758209229,
      "learning_rate": 4.169730896094172e-06,
      "loss": 0.486,
      "step": 8942
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0463634729385376,
      "learning_rate": 4.1687191640865135e-06,
      "loss": 0.4895,
      "step": 8943
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0102711915969849,
      "learning_rate": 4.167707467081197e-06,
      "loss": 0.5176,
      "step": 8944
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0874836444854736,
      "learning_rate": 4.166695805120825e-06,
      "loss": 0.6213,
      "step": 8945
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9626545906066895,
      "learning_rate": 4.165684178247993e-06,
      "loss": 0.4792,
      "step": 8946
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0832995176315308,
      "learning_rate": 4.1646725865053005e-06,
      "loss": 0.546,
      "step": 8947
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1358460187911987,
      "learning_rate": 4.163661029935336e-06,
      "loss": 0.4843,
      "step": 8948
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0058339834213257,
      "learning_rate": 4.162649508580698e-06,
      "loss": 0.5078,
      "step": 8949
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0836069583892822,
      "learning_rate": 4.161638022483976e-06,
      "loss": 0.493,
      "step": 8950
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0379528999328613,
      "learning_rate": 4.160626571687761e-06,
      "loss": 0.5035,
      "step": 8951
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9744682312011719,
      "learning_rate": 4.159615156234639e-06,
      "loss": 0.5002,
      "step": 8952
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0467766523361206,
      "learning_rate": 4.158603776167201e-06,
      "loss": 0.4753,
      "step": 8953
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9876313805580139,
      "learning_rate": 4.157592431528031e-06,
      "loss": 0.4733,
      "step": 8954
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9515472054481506,
      "learning_rate": 4.156581122359714e-06,
      "loss": 0.5397,
      "step": 8955
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9919760227203369,
      "learning_rate": 4.15556984870483e-06,
      "loss": 0.5128,
      "step": 8956
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9981586933135986,
      "learning_rate": 4.1545586106059636e-06,
      "loss": 0.517,
      "step": 8957
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0415711402893066,
      "learning_rate": 4.153547408105691e-06,
      "loss": 0.5341,
      "step": 8958
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0257682800292969,
      "learning_rate": 4.152536241246595e-06,
      "loss": 0.5003,
      "step": 8959
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0822460651397705,
      "learning_rate": 4.151525110071248e-06,
      "loss": 0.5218,
      "step": 8960
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0275789499282837,
      "learning_rate": 4.1505140146222276e-06,
      "loss": 0.5335,
      "step": 8961
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0042998790740967,
      "learning_rate": 4.149502954942107e-06,
      "loss": 0.5144,
      "step": 8962
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5982105731964111,
      "learning_rate": 4.148491931073459e-06,
      "loss": 0.53,
      "step": 8963
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9901741743087769,
      "learning_rate": 4.147480943058852e-06,
      "loss": 0.55,
      "step": 8964
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0572997331619263,
      "learning_rate": 4.146469990940858e-06,
      "loss": 0.5193,
      "step": 8965
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0043081045150757,
      "learning_rate": 4.1454590747620424e-06,
      "loss": 0.5609,
      "step": 8966
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0042357444763184,
      "learning_rate": 4.144448194564973e-06,
      "loss": 0.504,
      "step": 8967
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.105394721031189,
      "learning_rate": 4.1434373503922145e-06,
      "loss": 0.5269,
      "step": 8968
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0081567764282227,
      "learning_rate": 4.142426542286329e-06,
      "loss": 0.551,
      "step": 8969
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.062154769897461,
      "learning_rate": 4.141415770289877e-06,
      "loss": 0.5276,
      "step": 8970
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9704147577285767,
      "learning_rate": 4.140405034445423e-06,
      "loss": 0.5222,
      "step": 8971
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0527328252792358,
      "learning_rate": 4.13939433479552e-06,
      "loss": 0.5073,
      "step": 8972
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0498989820480347,
      "learning_rate": 4.13838367138273e-06,
      "loss": 0.5046,
      "step": 8973
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0412909984588623,
      "learning_rate": 4.137373044249604e-06,
      "loss": 0.5062,
      "step": 8974
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0586767196655273,
      "learning_rate": 4.1363624534387e-06,
      "loss": 0.5243,
      "step": 8975
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9737481474876404,
      "learning_rate": 4.135351898992568e-06,
      "loss": 0.4969,
      "step": 8976
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.039725422859192,
      "learning_rate": 4.134341380953761e-06,
      "loss": 0.5123,
      "step": 8977
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9762406349182129,
      "learning_rate": 4.133330899364824e-06,
      "loss": 0.509,
      "step": 8978
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0494922399520874,
      "learning_rate": 4.1323204542683105e-06,
      "loss": 0.5729,
      "step": 8979
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1076762676239014,
      "learning_rate": 4.131310045706763e-06,
      "loss": 0.5347,
      "step": 8980
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0384715795516968,
      "learning_rate": 4.130299673722729e-06,
      "loss": 0.5733,
      "step": 8981
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.998306930065155,
      "learning_rate": 4.129289338358748e-06,
      "loss": 0.498,
      "step": 8982
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0489163398742676,
      "learning_rate": 4.128279039657366e-06,
      "loss": 0.5565,
      "step": 8983
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0348968505859375,
      "learning_rate": 4.127268777661119e-06,
      "loss": 0.5388,
      "step": 8984
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9773428440093994,
      "learning_rate": 4.126258552412551e-06,
      "loss": 0.5214,
      "step": 8985
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0756745338439941,
      "learning_rate": 4.125248363954192e-06,
      "loss": 0.5214,
      "step": 8986
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.950646698474884,
      "learning_rate": 4.124238212328585e-06,
      "loss": 0.5012,
      "step": 8987
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.036226749420166,
      "learning_rate": 4.123228097578258e-06,
      "loss": 0.5613,
      "step": 8988
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.103487253189087,
      "learning_rate": 4.122218019745748e-06,
      "loss": 0.5256,
      "step": 8989
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0424480438232422,
      "learning_rate": 4.121207978873582e-06,
      "loss": 0.4807,
      "step": 8990
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0146026611328125,
      "learning_rate": 4.12019797500429e-06,
      "loss": 0.4948,
      "step": 8991
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.040250301361084,
      "learning_rate": 4.119188008180401e-06,
      "loss": 0.4975,
      "step": 8992
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0362321138381958,
      "learning_rate": 4.118178078444442e-06,
      "loss": 0.5254,
      "step": 8993
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.110710859298706,
      "learning_rate": 4.117168185838936e-06,
      "loss": 0.5201,
      "step": 8994
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0704374313354492,
      "learning_rate": 4.1161583304064055e-06,
      "loss": 0.4986,
      "step": 8995
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2568418979644775,
      "learning_rate": 4.115148512189374e-06,
      "loss": 0.5502,
      "step": 8996
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0175095796585083,
      "learning_rate": 4.114138731230362e-06,
      "loss": 0.5464,
      "step": 8997
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0222601890563965,
      "learning_rate": 4.113128987571885e-06,
      "loss": 0.5629,
      "step": 8998
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0167863368988037,
      "learning_rate": 4.1121192812564595e-06,
      "loss": 0.4754,
      "step": 8999
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1251527070999146,
      "learning_rate": 4.111109612326603e-06,
      "loss": 0.5546,
      "step": 9000
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0255568027496338,
      "learning_rate": 4.110099980824831e-06,
      "loss": 0.5204,
      "step": 9001
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.967444658279419,
      "learning_rate": 4.109090386793652e-06,
      "loss": 0.4901,
      "step": 9002
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9833146333694458,
      "learning_rate": 4.108080830275576e-06,
      "loss": 0.5335,
      "step": 9003
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0646792650222778,
      "learning_rate": 4.107071311313113e-06,
      "loss": 0.501,
      "step": 9004
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0886857509613037,
      "learning_rate": 4.106061829948773e-06,
      "loss": 0.5414,
      "step": 9005
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0678012371063232,
      "learning_rate": 4.10505238622506e-06,
      "loss": 0.5214,
      "step": 9006
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.06692373752594,
      "learning_rate": 4.104042980184476e-06,
      "loss": 0.6,
      "step": 9007
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9910625219345093,
      "learning_rate": 4.103033611869525e-06,
      "loss": 0.4904,
      "step": 9008
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9979705214500427,
      "learning_rate": 4.1020242813227096e-06,
      "loss": 0.5246,
      "step": 9009
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0530836582183838,
      "learning_rate": 4.101014988586528e-06,
      "loss": 0.5236,
      "step": 9010
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9468037486076355,
      "learning_rate": 4.100005733703477e-06,
      "loss": 0.4779,
      "step": 9011
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0236084461212158,
      "learning_rate": 4.0989965167160526e-06,
      "loss": 0.5306,
      "step": 9012
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9787879586219788,
      "learning_rate": 4.097987337666753e-06,
      "loss": 0.5132,
      "step": 9013
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9772061109542847,
      "learning_rate": 4.096978196598068e-06,
      "loss": 0.5285,
      "step": 9014
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0517624616622925,
      "learning_rate": 4.09596909355249e-06,
      "loss": 0.5299,
      "step": 9015
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9787384867668152,
      "learning_rate": 4.094960028572506e-06,
      "loss": 0.5308,
      "step": 9016
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0707100629806519,
      "learning_rate": 4.0939510017006095e-06,
      "loss": 0.4623,
      "step": 9017
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9557934999465942,
      "learning_rate": 4.092942012979285e-06,
      "loss": 0.4952,
      "step": 9018
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0198768377304077,
      "learning_rate": 4.091933062451015e-06,
      "loss": 0.5118,
      "step": 9019
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0764273405075073,
      "learning_rate": 4.0909241501582865e-06,
      "loss": 0.5339,
      "step": 9020
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1788642406463623,
      "learning_rate": 4.089915276143577e-06,
      "loss": 0.5636,
      "step": 9021
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1062636375427246,
      "learning_rate": 4.088906440449371e-06,
      "loss": 0.4871,
      "step": 9022
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.003823161125183,
      "learning_rate": 4.087897643118145e-06,
      "loss": 0.5081,
      "step": 9023
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0966594219207764,
      "learning_rate": 4.086888884192377e-06,
      "loss": 0.524,
      "step": 9024
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0141334533691406,
      "learning_rate": 4.0858801637145395e-06,
      "loss": 0.4818,
      "step": 9025
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.045664668083191,
      "learning_rate": 4.084871481727111e-06,
      "loss": 0.4793,
      "step": 9026
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0357179641723633,
      "learning_rate": 4.083862838272559e-06,
      "loss": 0.5286,
      "step": 9027
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0230491161346436,
      "learning_rate": 4.082854233393358e-06,
      "loss": 0.4954,
      "step": 9028
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1332944631576538,
      "learning_rate": 4.081845667131971e-06,
      "loss": 0.5341,
      "step": 9029
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1222625970840454,
      "learning_rate": 4.080837139530872e-06,
      "loss": 0.52,
      "step": 9030
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0352935791015625,
      "learning_rate": 4.0798286506325225e-06,
      "loss": 0.4609,
      "step": 9031
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.044788122177124,
      "learning_rate": 4.078820200479389e-06,
      "loss": 0.5113,
      "step": 9032
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9968512058258057,
      "learning_rate": 4.077811789113929e-06,
      "loss": 0.5507,
      "step": 9033
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1189789772033691,
      "learning_rate": 4.076803416578608e-06,
      "loss": 0.5566,
      "step": 9034
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9433977007865906,
      "learning_rate": 4.0757950829158855e-06,
      "loss": 0.4937,
      "step": 9035
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9398898482322693,
      "learning_rate": 4.074786788168216e-06,
      "loss": 0.5519,
      "step": 9036
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0074583292007446,
      "learning_rate": 4.073778532378056e-06,
      "loss": 0.5391,
      "step": 9037
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0626059770584106,
      "learning_rate": 4.072770315587858e-06,
      "loss": 0.5198,
      "step": 9038
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0814262628555298,
      "learning_rate": 4.071762137840079e-06,
      "loss": 0.4846,
      "step": 9039
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0414854288101196,
      "learning_rate": 4.070753999177167e-06,
      "loss": 0.5235,
      "step": 9040
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9925733804702759,
      "learning_rate": 4.069745899641571e-06,
      "loss": 0.4721,
      "step": 9041
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0932774543762207,
      "learning_rate": 4.0687378392757374e-06,
      "loss": 0.5368,
      "step": 9042
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0027060508728027,
      "learning_rate": 4.0677298181221155e-06,
      "loss": 0.5164,
      "step": 9043
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0120025873184204,
      "learning_rate": 4.066721836223149e-06,
      "loss": 0.5285,
      "step": 9044
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1019954681396484,
      "learning_rate": 4.065713893621278e-06,
      "loss": 0.5937,
      "step": 9045
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0674412250518799,
      "learning_rate": 4.064705990358943e-06,
      "loss": 0.4762,
      "step": 9046
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0491529703140259,
      "learning_rate": 4.063698126478587e-06,
      "loss": 0.5085,
      "step": 9047
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0456452369689941,
      "learning_rate": 4.062690302022647e-06,
      "loss": 0.5022,
      "step": 9048
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9888370037078857,
      "learning_rate": 4.0616825170335565e-06,
      "loss": 0.5098,
      "step": 9049
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9898523688316345,
      "learning_rate": 4.060674771553751e-06,
      "loss": 0.5218,
      "step": 9050
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9696733951568604,
      "learning_rate": 4.059667065625662e-06,
      "loss": 0.4681,
      "step": 9051
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.947935938835144,
      "learning_rate": 4.058659399291724e-06,
      "loss": 0.4732,
      "step": 9052
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0508413314819336,
      "learning_rate": 4.057651772594362e-06,
      "loss": 0.5909,
      "step": 9053
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9845364689826965,
      "learning_rate": 4.056644185576007e-06,
      "loss": 0.5609,
      "step": 9054
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9722887873649597,
      "learning_rate": 4.055636638279082e-06,
      "loss": 0.4734,
      "step": 9055
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0830838680267334,
      "learning_rate": 4.054629130746015e-06,
      "loss": 0.5326,
      "step": 9056
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0439698696136475,
      "learning_rate": 4.053621663019225e-06,
      "loss": 0.4756,
      "step": 9057
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9679062962532043,
      "learning_rate": 4.052614235141136e-06,
      "loss": 0.4911,
      "step": 9058
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.990229070186615,
      "learning_rate": 4.051606847154164e-06,
      "loss": 0.4859,
      "step": 9059
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0050839185714722,
      "learning_rate": 4.05059949910073e-06,
      "loss": 0.5275,
      "step": 9060
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9835556745529175,
      "learning_rate": 4.049592191023247e-06,
      "loss": 0.4554,
      "step": 9061
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0395163297653198,
      "learning_rate": 4.0485849229641325e-06,
      "loss": 0.5225,
      "step": 9062
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0173500776290894,
      "learning_rate": 4.047577694965794e-06,
      "loss": 0.5247,
      "step": 9063
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1180040836334229,
      "learning_rate": 4.046570507070649e-06,
      "loss": 0.5739,
      "step": 9064
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9771701097488403,
      "learning_rate": 4.045563359321102e-06,
      "loss": 0.4915,
      "step": 9065
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1348835229873657,
      "learning_rate": 4.044556251759562e-06,
      "loss": 0.5332,
      "step": 9066
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9895780086517334,
      "learning_rate": 4.043549184428434e-06,
      "loss": 0.5067,
      "step": 9067
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0886191129684448,
      "learning_rate": 4.042542157370122e-06,
      "loss": 0.5575,
      "step": 9068
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0821690559387207,
      "learning_rate": 4.041535170627029e-06,
      "loss": 0.5179,
      "step": 9069
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0421892404556274,
      "learning_rate": 4.040528224241558e-06,
      "loss": 0.4974,
      "step": 9070
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.916330873966217,
      "learning_rate": 4.039521318256104e-06,
      "loss": 0.4191,
      "step": 9071
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0716793537139893,
      "learning_rate": 4.038514452713065e-06,
      "loss": 0.5112,
      "step": 9072
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9565438628196716,
      "learning_rate": 4.037507627654838e-06,
      "loss": 0.48,
      "step": 9073
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0748134851455688,
      "learning_rate": 4.0365008431238184e-06,
      "loss": 0.5484,
      "step": 9074
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.108292579650879,
      "learning_rate": 4.035494099162396e-06,
      "loss": 0.5484,
      "step": 9075
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9552716612815857,
      "learning_rate": 4.03448739581296e-06,
      "loss": 0.4937,
      "step": 9076
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0210188627243042,
      "learning_rate": 4.033480733117902e-06,
      "loss": 0.5145,
      "step": 9077
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0513836145401,
      "learning_rate": 4.032474111119609e-06,
      "loss": 0.5311,
      "step": 9078
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0993913412094116,
      "learning_rate": 4.031467529860466e-06,
      "loss": 0.5559,
      "step": 9079
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.02354097366333,
      "learning_rate": 4.030460989382853e-06,
      "loss": 0.5015,
      "step": 9080
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9933673739433289,
      "learning_rate": 4.029454489729156e-06,
      "loss": 0.4839,
      "step": 9081
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0586044788360596,
      "learning_rate": 4.028448030941756e-06,
      "loss": 0.5142,
      "step": 9082
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9417781233787537,
      "learning_rate": 4.027441613063029e-06,
      "loss": 0.514,
      "step": 9083
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9901105165481567,
      "learning_rate": 4.026435236135351e-06,
      "loss": 0.5514,
      "step": 9084
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0686466693878174,
      "learning_rate": 4.025428900201098e-06,
      "loss": 0.5488,
      "step": 9085
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0662195682525635,
      "learning_rate": 4.024422605302646e-06,
      "loss": 0.5545,
      "step": 9086
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.995975911617279,
      "learning_rate": 4.023416351482364e-06,
      "loss": 0.4709,
      "step": 9087
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0386031866073608,
      "learning_rate": 4.022410138782621e-06,
      "loss": 0.5448,
      "step": 9088
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9993204474449158,
      "learning_rate": 4.021403967245786e-06,
      "loss": 0.4724,
      "step": 9089
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9913796186447144,
      "learning_rate": 4.020397836914227e-06,
      "loss": 0.4881,
      "step": 9090
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0130491256713867,
      "learning_rate": 4.019391747830307e-06,
      "loss": 0.5029,
      "step": 9091
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1187041997909546,
      "learning_rate": 4.018385700036389e-06,
      "loss": 0.5349,
      "step": 9092
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0950320959091187,
      "learning_rate": 4.017379693574833e-06,
      "loss": 0.5018,
      "step": 9093
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.059096336364746,
      "learning_rate": 4.016373728488002e-06,
      "loss": 0.5471,
      "step": 9094
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9699594974517822,
      "learning_rate": 4.01536780481825e-06,
      "loss": 0.5111,
      "step": 9095
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9474306106567383,
      "learning_rate": 4.014361922607936e-06,
      "loss": 0.5221,
      "step": 9096
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0697035789489746,
      "learning_rate": 4.013356081899412e-06,
      "loss": 0.5109,
      "step": 9097
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0884960889816284,
      "learning_rate": 4.0123502827350295e-06,
      "loss": 0.5052,
      "step": 9098
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0792391300201416,
      "learning_rate": 4.011344525157141e-06,
      "loss": 0.5616,
      "step": 9099
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.062630295753479,
      "learning_rate": 4.010338809208098e-06,
      "loss": 0.5512,
      "step": 9100
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1278208494186401,
      "learning_rate": 4.009333134930244e-06,
      "loss": 0.4883,
      "step": 9101
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.066338062286377,
      "learning_rate": 4.0083275023659236e-06,
      "loss": 0.5633,
      "step": 9102
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0112189054489136,
      "learning_rate": 4.007321911557483e-06,
      "loss": 0.5312,
      "step": 9103
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0003081560134888,
      "learning_rate": 4.0063163625472645e-06,
      "loss": 0.4837,
      "step": 9104
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9883372783660889,
      "learning_rate": 4.005310855377608e-06,
      "loss": 0.5329,
      "step": 9105
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9752151370048523,
      "learning_rate": 4.004305390090848e-06,
      "loss": 0.5131,
      "step": 9106
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.092448353767395,
      "learning_rate": 4.003299966729325e-06,
      "loss": 0.5343,
      "step": 9107
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0620566606521606,
      "learning_rate": 4.002294585335375e-06,
      "loss": 0.4834,
      "step": 9108
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0752540826797485,
      "learning_rate": 4.001289245951329e-06,
      "loss": 0.5103,
      "step": 9109
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0744627714157104,
      "learning_rate": 4.000283948619517e-06,
      "loss": 0.5233,
      "step": 9110
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0267804861068726,
      "learning_rate": 3.99927869338227e-06,
      "loss": 0.4902,
      "step": 9111
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1059128046035767,
      "learning_rate": 3.998273480281919e-06,
      "loss": 0.524,
      "step": 9112
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.12089204788208,
      "learning_rate": 3.997268309360785e-06,
      "loss": 0.5238,
      "step": 9113
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0233542919158936,
      "learning_rate": 3.996263180661194e-06,
      "loss": 0.5276,
      "step": 9114
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0219800472259521,
      "learning_rate": 3.995258094225468e-06,
      "loss": 0.4793,
      "step": 9115
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9956018328666687,
      "learning_rate": 3.99425305009593e-06,
      "loss": 0.4779,
      "step": 9116
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0090950727462769,
      "learning_rate": 3.993248048314897e-06,
      "loss": 0.5031,
      "step": 9117
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0749940872192383,
      "learning_rate": 3.992243088924686e-06,
      "loss": 0.5009,
      "step": 9118
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0072776079177856,
      "learning_rate": 3.991238171967612e-06,
      "loss": 0.5036,
      "step": 9119
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0810630321502686,
      "learning_rate": 3.9902332974859906e-06,
      "loss": 0.5413,
      "step": 9120
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0139497518539429,
      "learning_rate": 3.989228465522133e-06,
      "loss": 0.5441,
      "step": 9121
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0974375009536743,
      "learning_rate": 3.9882236761183476e-06,
      "loss": 0.5376,
      "step": 9122
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0594953298568726,
      "learning_rate": 3.987218929316942e-06,
      "loss": 0.5285,
      "step": 9123
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9792594313621521,
      "learning_rate": 3.986214225160226e-06,
      "loss": 0.4978,
      "step": 9124
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0863070487976074,
      "learning_rate": 3.9852095636905026e-06,
      "loss": 0.551,
      "step": 9125
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0734864473342896,
      "learning_rate": 3.984204944950073e-06,
      "loss": 0.5533,
      "step": 9126
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0189743041992188,
      "learning_rate": 3.983200368981241e-06,
      "loss": 0.5499,
      "step": 9127
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0332967042922974,
      "learning_rate": 3.982195835826302e-06,
      "loss": 0.507,
      "step": 9128
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0456657409667969,
      "learning_rate": 3.981191345527558e-06,
      "loss": 0.5451,
      "step": 9129
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.15138578414917,
      "learning_rate": 3.9801868981273e-06,
      "loss": 0.542,
      "step": 9130
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9716564416885376,
      "learning_rate": 3.979182493667826e-06,
      "loss": 0.5107,
      "step": 9131
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9666895866394043,
      "learning_rate": 3.978178132191424e-06,
      "loss": 0.5195,
      "step": 9132
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1179858446121216,
      "learning_rate": 3.9771738137403885e-06,
      "loss": 0.5324,
      "step": 9133
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0694400072097778,
      "learning_rate": 3.976169538357004e-06,
      "loss": 0.5392,
      "step": 9134
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1132067441940308,
      "learning_rate": 3.97516530608356e-06,
      "loss": 0.5465,
      "step": 9135
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0701768398284912,
      "learning_rate": 3.974161116962337e-06,
      "loss": 0.533,
      "step": 9136
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.993928849697113,
      "learning_rate": 3.973156971035623e-06,
      "loss": 0.525,
      "step": 9137
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0189783573150635,
      "learning_rate": 3.9721528683456966e-06,
      "loss": 0.5099,
      "step": 9138
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.042134404182434,
      "learning_rate": 3.971148808934838e-06,
      "loss": 0.5525,
      "step": 9139
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.031539797782898,
      "learning_rate": 3.970144792845322e-06,
      "loss": 0.4708,
      "step": 9140
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9681317806243896,
      "learning_rate": 3.9691408201194275e-06,
      "loss": 0.5319,
      "step": 9141
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.963118314743042,
      "learning_rate": 3.968136890799426e-06,
      "loss": 0.4821,
      "step": 9142
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9749932289123535,
      "learning_rate": 3.967133004927592e-06,
      "loss": 0.5076,
      "step": 9143
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0764023065567017,
      "learning_rate": 3.9661291625461945e-06,
      "loss": 0.5544,
      "step": 9144
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0290968418121338,
      "learning_rate": 3.965125363697499e-06,
      "loss": 0.5353,
      "step": 9145
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.998972475528717,
      "learning_rate": 3.964121608423775e-06,
      "loss": 0.5439,
      "step": 9146
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.948601245880127,
      "learning_rate": 3.963117896767288e-06,
      "loss": 0.5401,
      "step": 9147
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9426062107086182,
      "learning_rate": 3.962114228770299e-06,
      "loss": 0.5126,
      "step": 9148
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9606030583381653,
      "learning_rate": 3.961110604475067e-06,
      "loss": 0.4788,
      "step": 9149
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9690484404563904,
      "learning_rate": 3.960107023923855e-06,
      "loss": 0.5265,
      "step": 9150
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0060521364212036,
      "learning_rate": 3.959103487158919e-06,
      "loss": 0.5165,
      "step": 9151
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9228886961936951,
      "learning_rate": 3.958099994222515e-06,
      "loss": 0.504,
      "step": 9152
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0803656578063965,
      "learning_rate": 3.957096545156893e-06,
      "loss": 0.5398,
      "step": 9153
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.093693733215332,
      "learning_rate": 3.956093140004308e-06,
      "loss": 0.5192,
      "step": 9154
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.07072114944458,
      "learning_rate": 3.955089778807012e-06,
      "loss": 0.5252,
      "step": 9155
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0248234272003174,
      "learning_rate": 3.954086461607248e-06,
      "loss": 0.479,
      "step": 9156
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9825848937034607,
      "learning_rate": 3.9530831884472655e-06,
      "loss": 0.4851,
      "step": 9157
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0224632024765015,
      "learning_rate": 3.952079959369308e-06,
      "loss": 0.509,
      "step": 9158
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1817082166671753,
      "learning_rate": 3.951076774415619e-06,
      "loss": 0.5363,
      "step": 9159
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.055174469947815,
      "learning_rate": 3.950073633628436e-06,
      "loss": 0.5546,
      "step": 9160
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0198811292648315,
      "learning_rate": 3.949070537050002e-06,
      "loss": 0.5562,
      "step": 9161
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.990945041179657,
      "learning_rate": 3.948067484722549e-06,
      "loss": 0.5661,
      "step": 9162
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0743000507354736,
      "learning_rate": 3.947064476688318e-06,
      "loss": 0.5032,
      "step": 9163
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1395080089569092,
      "learning_rate": 3.946061512989537e-06,
      "loss": 0.5668,
      "step": 9164
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9844474196434021,
      "learning_rate": 3.94505859366844e-06,
      "loss": 0.5251,
      "step": 9165
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0164319276809692,
      "learning_rate": 3.944055718767255e-06,
      "loss": 0.5247,
      "step": 9166
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0355435609817505,
      "learning_rate": 3.943052888328211e-06,
      "loss": 0.4943,
      "step": 9167
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0106382369995117,
      "learning_rate": 3.942050102393533e-06,
      "loss": 0.5792,
      "step": 9168
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9311456680297852,
      "learning_rate": 3.941047361005445e-06,
      "loss": 0.4944,
      "step": 9169
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0806012153625488,
      "learning_rate": 3.940044664206168e-06,
      "loss": 0.5499,
      "step": 9170
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1101220846176147,
      "learning_rate": 3.939042012037924e-06,
      "loss": 0.5134,
      "step": 9171
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.014001488685608,
      "learning_rate": 3.938039404542929e-06,
      "loss": 0.4938,
      "step": 9172
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0506998300552368,
      "learning_rate": 3.937036841763401e-06,
      "loss": 0.5107,
      "step": 9173
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9829044342041016,
      "learning_rate": 3.936034323741555e-06,
      "loss": 0.5064,
      "step": 9174
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.094900131225586,
      "learning_rate": 3.935031850519599e-06,
      "loss": 0.5056,
      "step": 9175
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1232244968414307,
      "learning_rate": 3.934029422139749e-06,
      "loss": 0.5032,
      "step": 9176
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.076108694076538,
      "learning_rate": 3.933027038644213e-06,
      "loss": 0.4857,
      "step": 9177
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0531495809555054,
      "learning_rate": 3.932024700075196e-06,
      "loss": 0.5043,
      "step": 9178
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9696038961410522,
      "learning_rate": 3.931022406474902e-06,
      "loss": 0.5041,
      "step": 9179
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0817495584487915,
      "learning_rate": 3.930020157885537e-06,
      "loss": 0.5031,
      "step": 9180
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.058228850364685,
      "learning_rate": 3.929017954349301e-06,
      "loss": 0.5281,
      "step": 9181
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0713286399841309,
      "learning_rate": 3.928015795908394e-06,
      "loss": 0.5015,
      "step": 9182
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0175244808197021,
      "learning_rate": 3.927013682605011e-06,
      "loss": 0.4855,
      "step": 9183
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0039374828338623,
      "learning_rate": 3.9260116144813495e-06,
      "loss": 0.5395,
      "step": 9184
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9803718328475952,
      "learning_rate": 3.925009591579604e-06,
      "loss": 0.4945,
      "step": 9185
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0349849462509155,
      "learning_rate": 3.9240076139419655e-06,
      "loss": 0.4929,
      "step": 9186
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9854714274406433,
      "learning_rate": 3.92300568161062e-06,
      "loss": 0.529,
      "step": 9187
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0709526538848877,
      "learning_rate": 3.9220037946277606e-06,
      "loss": 0.515,
      "step": 9188
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9761338829994202,
      "learning_rate": 3.921001953035573e-06,
      "loss": 0.4962,
      "step": 9189
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0243232250213623,
      "learning_rate": 3.920000156876238e-06,
      "loss": 0.5189,
      "step": 9190
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0289701223373413,
      "learning_rate": 3.91899840619194e-06,
      "loss": 0.4971,
      "step": 9191
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1676281690597534,
      "learning_rate": 3.9179967010248556e-06,
      "loss": 0.5659,
      "step": 9192
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0041871070861816,
      "learning_rate": 3.91699504141717e-06,
      "loss": 0.4873,
      "step": 9193
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0443731546401978,
      "learning_rate": 3.915993427411054e-06,
      "loss": 0.5586,
      "step": 9194
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0109294652938843,
      "learning_rate": 3.914991859048684e-06,
      "loss": 0.5058,
      "step": 9195
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0021347999572754,
      "learning_rate": 3.913990336372231e-06,
      "loss": 0.4987,
      "step": 9196
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.090026617050171,
      "learning_rate": 3.912988859423869e-06,
      "loss": 0.5192,
      "step": 9197
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.079319715499878,
      "learning_rate": 3.911987428245765e-06,
      "loss": 0.5062,
      "step": 9198
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0323288440704346,
      "learning_rate": 3.9109860428800845e-06,
      "loss": 0.5075,
      "step": 9199
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9536668658256531,
      "learning_rate": 3.909984703368992e-06,
      "loss": 0.4905,
      "step": 9200
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0101124048233032,
      "learning_rate": 3.9089834097546534e-06,
      "loss": 0.5074,
      "step": 9201
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1430697441101074,
      "learning_rate": 3.907982162079229e-06,
      "loss": 0.5154,
      "step": 9202
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0381815433502197,
      "learning_rate": 3.906980960384875e-06,
      "loss": 0.5142,
      "step": 9203
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0313682556152344,
      "learning_rate": 3.90597980471375e-06,
      "loss": 0.49,
      "step": 9204
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1362711191177368,
      "learning_rate": 3.904978695108011e-06,
      "loss": 0.5485,
      "step": 9205
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0760146379470825,
      "learning_rate": 3.9039776316098104e-06,
      "loss": 0.5211,
      "step": 9206
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.040243148803711,
      "learning_rate": 3.902976614261298e-06,
      "loss": 0.4711,
      "step": 9207
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0480488538742065,
      "learning_rate": 3.901975643104625e-06,
      "loss": 0.5126,
      "step": 9208
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9665186405181885,
      "learning_rate": 3.9009747181819355e-06,
      "loss": 0.5173,
      "step": 9209
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1100220680236816,
      "learning_rate": 3.8999738395353795e-06,
      "loss": 0.5552,
      "step": 9210
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0339200496673584,
      "learning_rate": 3.898973007207097e-06,
      "loss": 0.5203,
      "step": 9211
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0091273784637451,
      "learning_rate": 3.897972221239233e-06,
      "loss": 0.5636,
      "step": 9212
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9974884390830994,
      "learning_rate": 3.896971481673923e-06,
      "loss": 0.53,
      "step": 9213
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0038785934448242,
      "learning_rate": 3.895970788553308e-06,
      "loss": 0.4895,
      "step": 9214
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0148288011550903,
      "learning_rate": 3.894970141919522e-06,
      "loss": 0.5236,
      "step": 9215
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0612456798553467,
      "learning_rate": 3.8939695418147e-06,
      "loss": 0.5206,
      "step": 9216
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0391217470169067,
      "learning_rate": 3.892968988280971e-06,
      "loss": 0.5306,
      "step": 9217
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9972529411315918,
      "learning_rate": 3.891968481360469e-06,
      "loss": 0.5359,
      "step": 9218
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0199835300445557,
      "learning_rate": 3.890968021095318e-06,
      "loss": 0.5045,
      "step": 9219
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0816878080368042,
      "learning_rate": 3.889967607527648e-06,
      "loss": 0.5284,
      "step": 9220
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9663817286491394,
      "learning_rate": 3.888967240699578e-06,
      "loss": 0.5057,
      "step": 9221
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0129668712615967,
      "learning_rate": 3.887966920653234e-06,
      "loss": 0.5088,
      "step": 9222
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0308432579040527,
      "learning_rate": 3.886966647430733e-06,
      "loss": 0.4421,
      "step": 9223
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9944412112236023,
      "learning_rate": 3.8859664210741965e-06,
      "loss": 0.4675,
      "step": 9224
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1368075609207153,
      "learning_rate": 3.884966241625737e-06,
      "loss": 0.5579,
      "step": 9225
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0805929899215698,
      "learning_rate": 3.88396610912747e-06,
      "loss": 0.5132,
      "step": 9226
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9060555696487427,
      "learning_rate": 3.882966023621509e-06,
      "loss": 0.4862,
      "step": 9227
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9819164276123047,
      "learning_rate": 3.881965985149962e-06,
      "loss": 0.502,
      "step": 9228
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.055558204650879,
      "learning_rate": 3.880965993754939e-06,
      "loss": 0.5513,
      "step": 9229
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0333936214447021,
      "learning_rate": 3.879966049478544e-06,
      "loss": 0.486,
      "step": 9230
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0895931720733643,
      "learning_rate": 3.878966152362882e-06,
      "loss": 0.5341,
      "step": 9231
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1182314157485962,
      "learning_rate": 3.877966302450057e-06,
      "loss": 0.5111,
      "step": 9232
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9638293981552124,
      "learning_rate": 3.876966499782168e-06,
      "loss": 0.4846,
      "step": 9233
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.008067011833191,
      "learning_rate": 3.875966744401311e-06,
      "loss": 0.5494,
      "step": 9234
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9954712390899658,
      "learning_rate": 3.874967036349585e-06,
      "loss": 0.5261,
      "step": 9235
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9387092590332031,
      "learning_rate": 3.8739673756690845e-06,
      "loss": 0.4703,
      "step": 9236
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0514830350875854,
      "learning_rate": 3.872967762401899e-06,
      "loss": 0.4944,
      "step": 9237
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0541136264801025,
      "learning_rate": 3.8719681965901225e-06,
      "loss": 0.5415,
      "step": 9238
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0214393138885498,
      "learning_rate": 3.870968678275838e-06,
      "loss": 0.559,
      "step": 9239
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0299464464187622,
      "learning_rate": 3.869969207501138e-06,
      "loss": 0.547,
      "step": 9240
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9679032564163208,
      "learning_rate": 3.868969784308101e-06,
      "loss": 0.5373,
      "step": 9241
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0148398876190186,
      "learning_rate": 3.867970408738814e-06,
      "loss": 0.5005,
      "step": 9242
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0126168727874756,
      "learning_rate": 3.866971080835352e-06,
      "loss": 0.528,
      "step": 9243
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1298620700836182,
      "learning_rate": 3.8659718006398e-06,
      "loss": 0.5561,
      "step": 9244
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0348117351531982,
      "learning_rate": 3.864972568194227e-06,
      "loss": 0.5081,
      "step": 9245
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0190778970718384,
      "learning_rate": 3.863973383540714e-06,
      "loss": 0.5419,
      "step": 9246
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9507538080215454,
      "learning_rate": 3.8629742467213266e-06,
      "loss": 0.4728,
      "step": 9247
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.019291639328003,
      "learning_rate": 3.86197515777814e-06,
      "loss": 0.4777,
      "step": 9248
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1635910272598267,
      "learning_rate": 3.860976116753221e-06,
      "loss": 0.5153,
      "step": 9249
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0270839929580688,
      "learning_rate": 3.859977123688636e-06,
      "loss": 0.5348,
      "step": 9250
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.003929615020752,
      "learning_rate": 3.858978178626446e-06,
      "loss": 0.5224,
      "step": 9251
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0090539455413818,
      "learning_rate": 3.8579792816087175e-06,
      "loss": 0.4824,
      "step": 9252
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0115137100219727,
      "learning_rate": 3.856980432677508e-06,
      "loss": 0.5075,
      "step": 9253
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0627245903015137,
      "learning_rate": 3.855981631874877e-06,
      "loss": 0.5505,
      "step": 9254
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0712007284164429,
      "learning_rate": 3.85498287924288e-06,
      "loss": 0.5619,
      "step": 9255
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9710827469825745,
      "learning_rate": 3.853984174823568e-06,
      "loss": 0.5019,
      "step": 9256
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1037195920944214,
      "learning_rate": 3.852985518658997e-06,
      "loss": 0.5162,
      "step": 9257
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0953142642974854,
      "learning_rate": 3.851986910791217e-06,
      "loss": 0.5105,
      "step": 9258
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0179473161697388,
      "learning_rate": 3.850988351262274e-06,
      "loss": 0.4858,
      "step": 9259
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0427842140197754,
      "learning_rate": 3.849989840114213e-06,
      "loss": 0.5628,
      "step": 9260
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0496574640274048,
      "learning_rate": 3.84899137738908e-06,
      "loss": 0.5172,
      "step": 9261
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9854255318641663,
      "learning_rate": 3.847992963128917e-06,
      "loss": 0.5073,
      "step": 9262
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9890566468238831,
      "learning_rate": 3.846994597375763e-06,
      "loss": 0.4825,
      "step": 9263
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.06586754322052,
      "learning_rate": 3.845996280171653e-06,
      "loss": 0.5554,
      "step": 9264
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0632151365280151,
      "learning_rate": 3.844998011558626e-06,
      "loss": 0.5553,
      "step": 9265
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0093785524368286,
      "learning_rate": 3.843999791578716e-06,
      "loss": 0.4919,
      "step": 9266
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0256174802780151,
      "learning_rate": 3.843001620273954e-06,
      "loss": 0.5176,
      "step": 9267
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.087141990661621,
      "learning_rate": 3.842003497686367e-06,
      "loss": 0.5095,
      "step": 9268
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9762932062149048,
      "learning_rate": 3.841005423857984e-06,
      "loss": 0.4938,
      "step": 9269
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0480889081954956,
      "learning_rate": 3.840007398830833e-06,
      "loss": 0.5023,
      "step": 9270
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9900689125061035,
      "learning_rate": 3.839009422646935e-06,
      "loss": 0.5756,
      "step": 9271
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.013576865196228,
      "learning_rate": 3.8380114953483095e-06,
      "loss": 0.5436,
      "step": 9272
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.955690860748291,
      "learning_rate": 3.837013616976977e-06,
      "loss": 0.4558,
      "step": 9273
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0240601301193237,
      "learning_rate": 3.8360157875749575e-06,
      "loss": 0.5591,
      "step": 9274
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9818782806396484,
      "learning_rate": 3.835018007184265e-06,
      "loss": 0.5416,
      "step": 9275
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9765317440032959,
      "learning_rate": 3.834020275846909e-06,
      "loss": 0.4923,
      "step": 9276
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0090014934539795,
      "learning_rate": 3.833022593604902e-06,
      "loss": 0.5279,
      "step": 9277
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9756602048873901,
      "learning_rate": 3.832024960500257e-06,
      "loss": 0.4579,
      "step": 9278
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0843050479888916,
      "learning_rate": 3.8310273765749774e-06,
      "loss": 0.5439,
      "step": 9279
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0157897472381592,
      "learning_rate": 3.830029841871067e-06,
      "loss": 0.5275,
      "step": 9280
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0507923364639282,
      "learning_rate": 3.82903235643053e-06,
      "loss": 0.5352,
      "step": 9281
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.057509183883667,
      "learning_rate": 3.828034920295368e-06,
      "loss": 0.5214,
      "step": 9282
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1005525588989258,
      "learning_rate": 3.827037533507579e-06,
      "loss": 0.5365,
      "step": 9283
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0654922723770142,
      "learning_rate": 3.826040196109158e-06,
      "loss": 0.5105,
      "step": 9284
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0815964937210083,
      "learning_rate": 3.825042908142102e-06,
      "loss": 0.5384,
      "step": 9285
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.967993438243866,
      "learning_rate": 3.824045669648398e-06,
      "loss": 0.5078,
      "step": 9286
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.002392292022705,
      "learning_rate": 3.823048480670044e-06,
      "loss": 0.5284,
      "step": 9287
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.01754629611969,
      "learning_rate": 3.8220513412490215e-06,
      "loss": 0.5216,
      "step": 9288
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.083465337753296,
      "learning_rate": 3.821054251427321e-06,
      "loss": 0.5552,
      "step": 9289
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9693893790245056,
      "learning_rate": 3.820057211246923e-06,
      "loss": 0.4693,
      "step": 9290
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2156462669372559,
      "learning_rate": 3.819060220749813e-06,
      "loss": 0.5517,
      "step": 9291
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1351375579833984,
      "learning_rate": 3.8180632799779675e-06,
      "loss": 0.4979,
      "step": 9292
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1142170429229736,
      "learning_rate": 3.817066388973367e-06,
      "loss": 0.4765,
      "step": 9293
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0561952590942383,
      "learning_rate": 3.816069547777983e-06,
      "loss": 0.5332,
      "step": 9294
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0577905178070068,
      "learning_rate": 3.815072756433794e-06,
      "loss": 0.5367,
      "step": 9295
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.036191701889038,
      "learning_rate": 3.814076014982769e-06,
      "loss": 0.5153,
      "step": 9296
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.082903504371643,
      "learning_rate": 3.8130793234668782e-06,
      "loss": 0.5365,
      "step": 9297
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0335683822631836,
      "learning_rate": 3.812082681928086e-06,
      "loss": 0.5667,
      "step": 9298
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9932186603546143,
      "learning_rate": 3.81108609040836e-06,
      "loss": 0.5059,
      "step": 9299
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.030775547027588,
      "learning_rate": 3.810089548949665e-06,
      "loss": 0.5228,
      "step": 9300
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0719740390777588,
      "learning_rate": 3.8090930575939588e-06,
      "loss": 0.4898,
      "step": 9301
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2146013975143433,
      "learning_rate": 3.8080966163832e-06,
      "loss": 0.5341,
      "step": 9302
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1133536100387573,
      "learning_rate": 3.807100225359346e-06,
      "loss": 0.5073,
      "step": 9303
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.131595253944397,
      "learning_rate": 3.8061038845643535e-06,
      "loss": 0.5494,
      "step": 9304
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0334343910217285,
      "learning_rate": 3.8051075940401727e-06,
      "loss": 0.5115,
      "step": 9305
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0557368993759155,
      "learning_rate": 3.8041113538287537e-06,
      "loss": 0.5398,
      "step": 9306
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.090477705001831,
      "learning_rate": 3.803115163972044e-06,
      "loss": 0.5244,
      "step": 9307
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0973477363586426,
      "learning_rate": 3.8021190245119937e-06,
      "loss": 0.5909,
      "step": 9308
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1514519453048706,
      "learning_rate": 3.8011229354905445e-06,
      "loss": 0.5593,
      "step": 9309
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0729470252990723,
      "learning_rate": 3.8001268969496357e-06,
      "loss": 0.5221,
      "step": 9310
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.132446527481079,
      "learning_rate": 3.799130908931209e-06,
      "loss": 0.5498,
      "step": 9311
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0245213508605957,
      "learning_rate": 3.7981349714772044e-06,
      "loss": 0.496,
      "step": 9312
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.035501480102539,
      "learning_rate": 3.7971390846295546e-06,
      "loss": 0.568,
      "step": 9313
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1167900562286377,
      "learning_rate": 3.7961432484301925e-06,
      "loss": 0.5439,
      "step": 9314
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0059419870376587,
      "learning_rate": 3.7951474629210517e-06,
      "loss": 0.4986,
      "step": 9315
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0720117092132568,
      "learning_rate": 3.7941517281440577e-06,
      "loss": 0.4818,
      "step": 9316
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0264700651168823,
      "learning_rate": 3.7931560441411413e-06,
      "loss": 0.52,
      "step": 9317
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9719229340553284,
      "learning_rate": 3.792160410954225e-06,
      "loss": 0.5183,
      "step": 9318
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9971143007278442,
      "learning_rate": 3.791164828625233e-06,
      "loss": 0.5075,
      "step": 9319
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9525416493415833,
      "learning_rate": 3.7901692971960823e-06,
      "loss": 0.4485,
      "step": 9320
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0481828451156616,
      "learning_rate": 3.7891738167086968e-06,
      "loss": 0.4993,
      "step": 9321
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0023530721664429,
      "learning_rate": 3.7881783872049875e-06,
      "loss": 0.531,
      "step": 9322
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.02839994430542,
      "learning_rate": 3.7871830087268726e-06,
      "loss": 0.4787,
      "step": 9323
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1028426885604858,
      "learning_rate": 3.7861876813162596e-06,
      "loss": 0.5187,
      "step": 9324
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1042375564575195,
      "learning_rate": 3.7851924050150633e-06,
      "loss": 0.5468,
      "step": 9325
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0218197107315063,
      "learning_rate": 3.7841971798651876e-06,
      "loss": 0.4854,
      "step": 9326
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0304937362670898,
      "learning_rate": 3.78320200590854e-06,
      "loss": 0.5644,
      "step": 9327
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9608798027038574,
      "learning_rate": 3.782206883187021e-06,
      "loss": 0.5526,
      "step": 9328
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1082509756088257,
      "learning_rate": 3.7812118117425363e-06,
      "loss": 0.4879,
      "step": 9329
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9273836612701416,
      "learning_rate": 3.7802167916169808e-06,
      "loss": 0.535,
      "step": 9330
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0372880697250366,
      "learning_rate": 3.7792218228522536e-06,
      "loss": 0.5635,
      "step": 9331
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0150055885314941,
      "learning_rate": 3.7782269054902493e-06,
      "loss": 0.5236,
      "step": 9332
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9948717355728149,
      "learning_rate": 3.777232039572858e-06,
      "loss": 0.5182,
      "step": 9333
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0627039670944214,
      "learning_rate": 3.7762372251419722e-06,
      "loss": 0.5691,
      "step": 9334
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9712206721305847,
      "learning_rate": 3.7752424622394807e-06,
      "loss": 0.4835,
      "step": 9335
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0412235260009766,
      "learning_rate": 3.7742477509072684e-06,
      "loss": 0.4699,
      "step": 9336
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0442676544189453,
      "learning_rate": 3.7732530911872177e-06,
      "loss": 0.5041,
      "step": 9337
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0916869640350342,
      "learning_rate": 3.7722584831212127e-06,
      "loss": 0.5264,
      "step": 9338
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.066199541091919,
      "learning_rate": 3.771263926751133e-06,
      "loss": 0.5292,
      "step": 9339
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.045325756072998,
      "learning_rate": 3.7702694221188548e-06,
      "loss": 0.5335,
      "step": 9340
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0770453214645386,
      "learning_rate": 3.769274969266251e-06,
      "loss": 0.5102,
      "step": 9341
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.151835560798645,
      "learning_rate": 3.768280568235198e-06,
      "loss": 0.5352,
      "step": 9342
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9812878966331482,
      "learning_rate": 3.767286219067566e-06,
      "loss": 0.5264,
      "step": 9343
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0495941638946533,
      "learning_rate": 3.766291921805224e-06,
      "loss": 0.5127,
      "step": 9344
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.004805088043213,
      "learning_rate": 3.765297676490035e-06,
      "loss": 0.4805,
      "step": 9345
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0435104370117188,
      "learning_rate": 3.764303483163867e-06,
      "loss": 0.5383,
      "step": 9346
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1722232103347778,
      "learning_rate": 3.7633093418685806e-06,
      "loss": 0.4849,
      "step": 9347
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0799705982208252,
      "learning_rate": 3.7623152526460365e-06,
      "loss": 0.5175,
      "step": 9348
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1457706689834595,
      "learning_rate": 3.7613212155380907e-06,
      "loss": 0.5338,
      "step": 9349
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0496329069137573,
      "learning_rate": 3.760327230586598e-06,
      "loss": 0.4957,
      "step": 9350
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.058712124824524,
      "learning_rate": 3.7593332978334153e-06,
      "loss": 0.51,
      "step": 9351
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0165393352508545,
      "learning_rate": 3.7583394173203913e-06,
      "loss": 0.5161,
      "step": 9352
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0252450704574585,
      "learning_rate": 3.757345589089374e-06,
      "loss": 0.5228,
      "step": 9353
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0472545623779297,
      "learning_rate": 3.75635181318221e-06,
      "loss": 0.5508,
      "step": 9354
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1078425645828247,
      "learning_rate": 3.755358089640747e-06,
      "loss": 0.5382,
      "step": 9355
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1093820333480835,
      "learning_rate": 3.754364418506825e-06,
      "loss": 0.5056,
      "step": 9356
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.99863600730896,
      "learning_rate": 3.7533707998222835e-06,
      "loss": 0.5549,
      "step": 9357
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0610816478729248,
      "learning_rate": 3.7523772336289594e-06,
      "loss": 0.5259,
      "step": 9358
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.961898684501648,
      "learning_rate": 3.751383719968692e-06,
      "loss": 0.4806,
      "step": 9359
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0301170349121094,
      "learning_rate": 3.7503902588833124e-06,
      "loss": 0.5118,
      "step": 9360
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0162115097045898,
      "learning_rate": 3.7493968504146513e-06,
      "loss": 0.4647,
      "step": 9361
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0619779825210571,
      "learning_rate": 3.748403494604539e-06,
      "loss": 0.5069,
      "step": 9362
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0400292873382568,
      "learning_rate": 3.747410191494799e-06,
      "loss": 0.5548,
      "step": 9363
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1837493181228638,
      "learning_rate": 3.74641694112726e-06,
      "loss": 0.4983,
      "step": 9364
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.074021816253662,
      "learning_rate": 3.745423743543744e-06,
      "loss": 0.4983,
      "step": 9365
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.034995675086975,
      "learning_rate": 3.7444305987860698e-06,
      "loss": 0.5463,
      "step": 9366
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1181045770645142,
      "learning_rate": 3.7434375068960528e-06,
      "loss": 0.5069,
      "step": 9367
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0152443647384644,
      "learning_rate": 3.7424444679155126e-06,
      "loss": 0.5617,
      "step": 9368
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0449225902557373,
      "learning_rate": 3.7414514818862613e-06,
      "loss": 0.5603,
      "step": 9369
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0839518308639526,
      "learning_rate": 3.7404585488501106e-06,
      "loss": 0.5406,
      "step": 9370
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9754869937896729,
      "learning_rate": 3.7394656688488663e-06,
      "loss": 0.4874,
      "step": 9371
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0486022233963013,
      "learning_rate": 3.7384728419243386e-06,
      "loss": 0.5203,
      "step": 9372
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0572601556777954,
      "learning_rate": 3.7374800681183334e-06,
      "loss": 0.5349,
      "step": 9373
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0961726903915405,
      "learning_rate": 3.736487347472649e-06,
      "loss": 0.5418,
      "step": 9374
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1418040990829468,
      "learning_rate": 3.735494680029086e-06,
      "loss": 0.5269,
      "step": 9375
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0011223554611206,
      "learning_rate": 3.734502065829443e-06,
      "loss": 0.4951,
      "step": 9376
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0225920677185059,
      "learning_rate": 3.7335095049155173e-06,
      "loss": 0.4769,
      "step": 9377
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.065794587135315,
      "learning_rate": 3.732516997329101e-06,
      "loss": 0.5224,
      "step": 9378
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9985071420669556,
      "learning_rate": 3.731524543111983e-06,
      "loss": 0.5744,
      "step": 9379
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0735899209976196,
      "learning_rate": 3.7305321423059526e-06,
      "loss": 0.5112,
      "step": 9380
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0018068552017212,
      "learning_rate": 3.7295397949528e-06,
      "loss": 0.4992,
      "step": 9381
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.014588475227356,
      "learning_rate": 3.7285475010943067e-06,
      "loss": 0.5304,
      "step": 9382
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0748690366744995,
      "learning_rate": 3.7275552607722544e-06,
      "loss": 0.5171,
      "step": 9383
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0756276845932007,
      "learning_rate": 3.726563074028422e-06,
      "loss": 0.5285,
      "step": 9384
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0047024488449097,
      "learning_rate": 3.7255709409045914e-06,
      "loss": 0.4926,
      "step": 9385
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0440177917480469,
      "learning_rate": 3.724578861442535e-06,
      "loss": 0.4949,
      "step": 9386
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0596648454666138,
      "learning_rate": 3.7235868356840244e-06,
      "loss": 0.5292,
      "step": 9387
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9568275809288025,
      "learning_rate": 3.722594863670831e-06,
      "loss": 0.4839,
      "step": 9388
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.017176628112793,
      "learning_rate": 3.7216029454447262e-06,
      "loss": 0.4838,
      "step": 9389
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0817347764968872,
      "learning_rate": 3.720611081047474e-06,
      "loss": 0.5194,
      "step": 9390
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9253323078155518,
      "learning_rate": 3.7196192705208378e-06,
      "loss": 0.4787,
      "step": 9391
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0608935356140137,
      "learning_rate": 3.7186275139065807e-06,
      "loss": 0.4919,
      "step": 9392
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.075279951095581,
      "learning_rate": 3.7176358112464593e-06,
      "loss": 0.5452,
      "step": 9393
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9955409169197083,
      "learning_rate": 3.716644162582235e-06,
      "loss": 0.5298,
      "step": 9394
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0128940343856812,
      "learning_rate": 3.7156525679556597e-06,
      "loss": 0.5528,
      "step": 9395
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9894576072692871,
      "learning_rate": 3.7146610274084875e-06,
      "loss": 0.538,
      "step": 9396
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1473660469055176,
      "learning_rate": 3.7136695409824665e-06,
      "loss": 0.5343,
      "step": 9397
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.062568187713623,
      "learning_rate": 3.712678108719348e-06,
      "loss": 0.5436,
      "step": 9398
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1151442527770996,
      "learning_rate": 3.711686730660875e-06,
      "loss": 0.5086,
      "step": 9399
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0051631927490234,
      "learning_rate": 3.710695406848794e-06,
      "loss": 0.5591,
      "step": 9400
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1091972589492798,
      "learning_rate": 3.709704137324841e-06,
      "loss": 0.5446,
      "step": 9401
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0302300453186035,
      "learning_rate": 3.7087129221307605e-06,
      "loss": 0.4906,
      "step": 9402
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0200457572937012,
      "learning_rate": 3.7077217613082863e-06,
      "loss": 0.5153,
      "step": 9403
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0734446048736572,
      "learning_rate": 3.7067306548991543e-06,
      "loss": 0.5324,
      "step": 9404
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0424174070358276,
      "learning_rate": 3.7057396029450925e-06,
      "loss": 0.5747,
      "step": 9405
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9913774132728577,
      "learning_rate": 3.7047486054878367e-06,
      "loss": 0.531,
      "step": 9406
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.007533073425293,
      "learning_rate": 3.7037576625691095e-06,
      "loss": 0.5315,
      "step": 9407
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.990408182144165,
      "learning_rate": 3.7027667742306393e-06,
      "loss": 0.5084,
      "step": 9408
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9975370764732361,
      "learning_rate": 3.7017759405141476e-06,
      "loss": 0.4906,
      "step": 9409
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0215646028518677,
      "learning_rate": 3.7007851614613522e-06,
      "loss": 0.5576,
      "step": 9410
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9978448152542114,
      "learning_rate": 3.699794437113975e-06,
      "loss": 0.5561,
      "step": 9411
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9627044796943665,
      "learning_rate": 3.698803767513732e-06,
      "loss": 0.5112,
      "step": 9412
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0477159023284912,
      "learning_rate": 3.6978131527023363e-06,
      "loss": 0.4439,
      "step": 9413
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0030741691589355,
      "learning_rate": 3.696822592721497e-06,
      "loss": 0.5298,
      "step": 9414
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9714114665985107,
      "learning_rate": 3.695832087612925e-06,
      "loss": 0.5228,
      "step": 9415
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1148886680603027,
      "learning_rate": 3.6948416374183287e-06,
      "loss": 0.5417,
      "step": 9416
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.027069330215454,
      "learning_rate": 3.6938512421794103e-06,
      "loss": 0.5612,
      "step": 9417
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0608090162277222,
      "learning_rate": 3.6928609019378702e-06,
      "loss": 0.5397,
      "step": 9418
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0678914785385132,
      "learning_rate": 3.6918706167354125e-06,
      "loss": 0.5306,
      "step": 9419
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9682676792144775,
      "learning_rate": 3.690880386613732e-06,
      "loss": 0.4933,
      "step": 9420
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0711426734924316,
      "learning_rate": 3.689890211614525e-06,
      "loss": 0.5093,
      "step": 9421
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.001579999923706,
      "learning_rate": 3.6889000917794816e-06,
      "loss": 0.5197,
      "step": 9422
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.130233883857727,
      "learning_rate": 3.6879100271502953e-06,
      "loss": 0.5472,
      "step": 9423
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9985414147377014,
      "learning_rate": 3.6869200177686543e-06,
      "loss": 0.4826,
      "step": 9424
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9541159868240356,
      "learning_rate": 3.6859300636762423e-06,
      "loss": 0.4679,
      "step": 9425
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0594643354415894,
      "learning_rate": 3.6849401649147453e-06,
      "loss": 0.5476,
      "step": 9426
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0757240056991577,
      "learning_rate": 3.683950321525841e-06,
      "loss": 0.5004,
      "step": 9427
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9650991559028625,
      "learning_rate": 3.682960533551213e-06,
      "loss": 0.4863,
      "step": 9428
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9989383220672607,
      "learning_rate": 3.681970801032534e-06,
      "loss": 0.5131,
      "step": 9429
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.018264651298523,
      "learning_rate": 3.68098112401148e-06,
      "loss": 0.4824,
      "step": 9430
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0563603639602661,
      "learning_rate": 3.6799915025297206e-06,
      "loss": 0.4889,
      "step": 9431
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0259501934051514,
      "learning_rate": 3.6790019366289293e-06,
      "loss": 0.5,
      "step": 9432
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0241059064865112,
      "learning_rate": 3.67801242635077e-06,
      "loss": 0.5454,
      "step": 9433
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0234817266464233,
      "learning_rate": 3.6770229717369086e-06,
      "loss": 0.5117,
      "step": 9434
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0400785207748413,
      "learning_rate": 3.6760335728290062e-06,
      "loss": 0.5439,
      "step": 9435
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0051709413528442,
      "learning_rate": 3.6750442296687272e-06,
      "loss": 0.5163,
      "step": 9436
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0035061836242676,
      "learning_rate": 3.6740549422977244e-06,
      "loss": 0.5235,
      "step": 9437
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.043760061264038,
      "learning_rate": 3.6730657107576574e-06,
      "loss": 0.5468,
      "step": 9438
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0327634811401367,
      "learning_rate": 3.6720765350901765e-06,
      "loss": 0.5015,
      "step": 9439
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0611823797225952,
      "learning_rate": 3.671087415336931e-06,
      "loss": 0.5478,
      "step": 9440
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.970547616481781,
      "learning_rate": 3.6700983515395726e-06,
      "loss": 0.5038,
      "step": 9441
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0590057373046875,
      "learning_rate": 3.669109343739747e-06,
      "loss": 0.4945,
      "step": 9442
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0782158374786377,
      "learning_rate": 3.668120391979098e-06,
      "loss": 0.4998,
      "step": 9443
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8958311676979065,
      "learning_rate": 3.6671314962992634e-06,
      "loss": 0.4783,
      "step": 9444
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0121147632598877,
      "learning_rate": 3.666142656741886e-06,
      "loss": 0.5356,
      "step": 9445
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0281257629394531,
      "learning_rate": 3.6651538733486027e-06,
      "loss": 0.5264,
      "step": 9446
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9911076426506042,
      "learning_rate": 3.664165146161045e-06,
      "loss": 0.5249,
      "step": 9447
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0183645486831665,
      "learning_rate": 3.663176475220844e-06,
      "loss": 0.532,
      "step": 9448
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9785016775131226,
      "learning_rate": 3.6621878605696338e-06,
      "loss": 0.4689,
      "step": 9449
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0997915267944336,
      "learning_rate": 3.6611993022490383e-06,
      "loss": 0.4984,
      "step": 9450
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.024304986000061,
      "learning_rate": 3.660210800300683e-06,
      "loss": 0.5057,
      "step": 9451
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0579379796981812,
      "learning_rate": 3.6592223547661888e-06,
      "loss": 0.5391,
      "step": 9452
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0190274715423584,
      "learning_rate": 3.6582339656871778e-06,
      "loss": 0.499,
      "step": 9453
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0800539255142212,
      "learning_rate": 3.6572456331052673e-06,
      "loss": 0.5218,
      "step": 9454
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.018898844718933,
      "learning_rate": 3.656257357062073e-06,
      "loss": 0.5727,
      "step": 9455
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1038990020751953,
      "learning_rate": 3.6552691375992056e-06,
      "loss": 0.5848,
      "step": 9456
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0068682432174683,
      "learning_rate": 3.6542809747582755e-06,
      "loss": 0.52,
      "step": 9457
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0212358236312866,
      "learning_rate": 3.6532928685808937e-06,
      "loss": 0.5399,
      "step": 9458
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0951766967773438,
      "learning_rate": 3.6523048191086654e-06,
      "loss": 0.5039,
      "step": 9459
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0452274084091187,
      "learning_rate": 3.6513168263831913e-06,
      "loss": 0.4942,
      "step": 9460
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.118046760559082,
      "learning_rate": 3.6503288904460725e-06,
      "loss": 0.5067,
      "step": 9461
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0423998832702637,
      "learning_rate": 3.6493410113389116e-06,
      "loss": 0.5379,
      "step": 9462
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0723527669906616,
      "learning_rate": 3.648353189103302e-06,
      "loss": 0.5605,
      "step": 9463
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.032225251197815,
      "learning_rate": 3.6473654237808365e-06,
      "loss": 0.4947,
      "step": 9464
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0548571348190308,
      "learning_rate": 3.6463777154131065e-06,
      "loss": 0.5521,
      "step": 9465
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9877645373344421,
      "learning_rate": 3.645390064041704e-06,
      "loss": 0.4899,
      "step": 9466
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9609139561653137,
      "learning_rate": 3.6444024697082137e-06,
      "loss": 0.4671,
      "step": 9467
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0752314329147339,
      "learning_rate": 3.6434149324542185e-06,
      "loss": 0.5313,
      "step": 9468
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9778896570205688,
      "learning_rate": 3.6424274523213e-06,
      "loss": 0.5053,
      "step": 9469
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0713123083114624,
      "learning_rate": 3.641440029351041e-06,
      "loss": 0.5318,
      "step": 9470
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1029250621795654,
      "learning_rate": 3.640452663585017e-06,
      "loss": 0.5373,
      "step": 9471
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0140851736068726,
      "learning_rate": 3.6394653550647996e-06,
      "loss": 0.4839,
      "step": 9472
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9579431414604187,
      "learning_rate": 3.638478103831965e-06,
      "loss": 0.48,
      "step": 9473
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.975917398929596,
      "learning_rate": 3.6374909099280786e-06,
      "loss": 0.5076,
      "step": 9474
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0448685884475708,
      "learning_rate": 3.636503773394713e-06,
      "loss": 0.5353,
      "step": 9475
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0627449750900269,
      "learning_rate": 3.635516694273428e-06,
      "loss": 0.5691,
      "step": 9476
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1303012371063232,
      "learning_rate": 3.63452967260579e-06,
      "loss": 0.5148,
      "step": 9477
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9685790538787842,
      "learning_rate": 3.633542708433355e-06,
      "loss": 0.4675,
      "step": 9478
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0133579969406128,
      "learning_rate": 3.632555801797686e-06,
      "loss": 0.5217,
      "step": 9479
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0786575078964233,
      "learning_rate": 3.631568952740333e-06,
      "loss": 0.5198,
      "step": 9480
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.087209701538086,
      "learning_rate": 3.6305821613028524e-06,
      "loss": 0.5274,
      "step": 9481
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0685940980911255,
      "learning_rate": 3.6295954275267914e-06,
      "loss": 0.5198,
      "step": 9482
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0028527975082397,
      "learning_rate": 3.6286087514537017e-06,
      "loss": 0.4928,
      "step": 9483
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0007100105285645,
      "learning_rate": 3.6276221331251253e-06,
      "loss": 0.4894,
      "step": 9484
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9105003476142883,
      "learning_rate": 3.626635572582608e-06,
      "loss": 0.4802,
      "step": 9485
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1068147420883179,
      "learning_rate": 3.6256490698676884e-06,
      "loss": 0.5292,
      "step": 9486
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0141433477401733,
      "learning_rate": 3.6246626250219047e-06,
      "loss": 0.4763,
      "step": 9487
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9817478060722351,
      "learning_rate": 3.623676238086794e-06,
      "loss": 0.4768,
      "step": 9488
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.070478916168213,
      "learning_rate": 3.6226899091038896e-06,
      "loss": 0.5326,
      "step": 9489
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1199346780776978,
      "learning_rate": 3.6217036381147216e-06,
      "loss": 0.5614,
      "step": 9490
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1189179420471191,
      "learning_rate": 3.620717425160818e-06,
      "loss": 0.5233,
      "step": 9491
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1126006841659546,
      "learning_rate": 3.619731270283705e-06,
      "loss": 0.5454,
      "step": 9492
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.058700442314148,
      "learning_rate": 3.6187451735249085e-06,
      "loss": 0.4868,
      "step": 9493
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1182832717895508,
      "learning_rate": 3.6177591349259465e-06,
      "loss": 0.5634,
      "step": 9494
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1220393180847168,
      "learning_rate": 3.616773154528339e-06,
      "loss": 0.5331,
      "step": 9495
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0078002214431763,
      "learning_rate": 3.6157872323736017e-06,
      "loss": 0.4684,
      "step": 9496
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0165269374847412,
      "learning_rate": 3.61480136850325e-06,
      "loss": 0.5044,
      "step": 9497
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0735347270965576,
      "learning_rate": 3.6138155629587925e-06,
      "loss": 0.5003,
      "step": 9498
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0364093780517578,
      "learning_rate": 3.61282981578174e-06,
      "loss": 0.5305,
      "step": 9499
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9629833698272705,
      "learning_rate": 3.611844127013598e-06,
      "loss": 0.5103,
      "step": 9500
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0671041011810303,
      "learning_rate": 3.6108584966958717e-06,
      "loss": 0.4761,
      "step": 9501
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0553507804870605,
      "learning_rate": 3.6098729248700604e-06,
      "loss": 0.5267,
      "step": 9502
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9224212765693665,
      "learning_rate": 3.6088874115776664e-06,
      "loss": 0.4905,
      "step": 9503
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9481444954872131,
      "learning_rate": 3.6079019568601816e-06,
      "loss": 0.5146,
      "step": 9504
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9659034609794617,
      "learning_rate": 3.606916560759104e-06,
      "loss": 0.484,
      "step": 9505
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0172665119171143,
      "learning_rate": 3.6059312233159237e-06,
      "loss": 0.5203,
      "step": 9506
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0298722982406616,
      "learning_rate": 3.6049459445721303e-06,
      "loss": 0.4974,
      "step": 9507
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0093005895614624,
      "learning_rate": 3.6039607245692086e-06,
      "loss": 0.4608,
      "step": 9508
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9928223490715027,
      "learning_rate": 3.6029755633486464e-06,
      "loss": 0.5323,
      "step": 9509
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9848858714103699,
      "learning_rate": 3.601990460951922e-06,
      "loss": 0.4473,
      "step": 9510
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9341360926628113,
      "learning_rate": 3.6010054174205167e-06,
      "loss": 0.4807,
      "step": 9511
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0773816108703613,
      "learning_rate": 3.6000204327959055e-06,
      "loss": 0.5336,
      "step": 9512
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.973141074180603,
      "learning_rate": 3.599035507119565e-06,
      "loss": 0.4723,
      "step": 9513
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0100555419921875,
      "learning_rate": 3.5980506404329647e-06,
      "loss": 0.5553,
      "step": 9514
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0515456199645996,
      "learning_rate": 3.597065832777576e-06,
      "loss": 0.5559,
      "step": 9515
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9875101447105408,
      "learning_rate": 3.5960810841948622e-06,
      "loss": 0.532,
      "step": 9516
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9792640209197998,
      "learning_rate": 3.595096394726293e-06,
      "loss": 0.4745,
      "step": 9517
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0469679832458496,
      "learning_rate": 3.594111764413326e-06,
      "loss": 0.5014,
      "step": 9518
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.092686414718628,
      "learning_rate": 3.5931271932974227e-06,
      "loss": 0.5121,
      "step": 9519
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9899653196334839,
      "learning_rate": 3.592142681420039e-06,
      "loss": 0.5247,
      "step": 9520
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0317049026489258,
      "learning_rate": 3.5911582288226275e-06,
      "loss": 0.5234,
      "step": 9521
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0484519004821777,
      "learning_rate": 3.5901738355466433e-06,
      "loss": 0.5742,
      "step": 9522
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0550673007965088,
      "learning_rate": 3.5891895016335347e-06,
      "loss": 0.5109,
      "step": 9523
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0651562213897705,
      "learning_rate": 3.588205227124749e-06,
      "loss": 0.529,
      "step": 9524
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0518194437026978,
      "learning_rate": 3.587221012061728e-06,
      "loss": 0.5127,
      "step": 9525
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9867557287216187,
      "learning_rate": 3.586236856485916e-06,
      "loss": 0.5191,
      "step": 9526
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.00824773311615,
      "learning_rate": 3.5852527604387533e-06,
      "loss": 0.5048,
      "step": 9527
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0379291772842407,
      "learning_rate": 3.5842687239616745e-06,
      "loss": 0.4946,
      "step": 9528
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9897534847259521,
      "learning_rate": 3.583284747096114e-06,
      "loss": 0.5173,
      "step": 9529
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9886677265167236,
      "learning_rate": 3.5823008298835044e-06,
      "loss": 0.5686,
      "step": 9530
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1005386114120483,
      "learning_rate": 3.5813169723652763e-06,
      "loss": 0.4842,
      "step": 9531
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1401194334030151,
      "learning_rate": 3.5803331745828558e-06,
      "loss": 0.5507,
      "step": 9532
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9957314133644104,
      "learning_rate": 3.579349436577665e-06,
      "loss": 0.5301,
      "step": 9533
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.05839204788208,
      "learning_rate": 3.5783657583911268e-06,
      "loss": 0.5346,
      "step": 9534
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9779189825057983,
      "learning_rate": 3.5773821400646623e-06,
      "loss": 0.4958,
      "step": 9535
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9635352492332458,
      "learning_rate": 3.5763985816396873e-06,
      "loss": 0.4826,
      "step": 9536
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0159794092178345,
      "learning_rate": 3.575415083157615e-06,
      "loss": 0.4449,
      "step": 9537
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9795805215835571,
      "learning_rate": 3.5744316446598565e-06,
      "loss": 0.4828,
      "step": 9538
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9881680607795715,
      "learning_rate": 3.5734482661878244e-06,
      "loss": 0.5146,
      "step": 9539
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1142942905426025,
      "learning_rate": 3.5724649477829232e-06,
      "loss": 0.5824,
      "step": 9540
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0041018724441528,
      "learning_rate": 3.5714816894865556e-06,
      "loss": 0.4668,
      "step": 9541
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0158956050872803,
      "learning_rate": 3.570498491340124e-06,
      "loss": 0.49,
      "step": 9542
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0179728269577026,
      "learning_rate": 3.5695153533850302e-06,
      "loss": 0.4837,
      "step": 9543
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0044492483139038,
      "learning_rate": 3.5685322756626683e-06,
      "loss": 0.5229,
      "step": 9544
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0525065660476685,
      "learning_rate": 3.5675492582144322e-06,
      "loss": 0.5419,
      "step": 9545
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0244168043136597,
      "learning_rate": 3.566566301081712e-06,
      "loss": 0.485,
      "step": 9546
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0063843727111816,
      "learning_rate": 3.5655834043059e-06,
      "loss": 0.4753,
      "step": 9547
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9746105074882507,
      "learning_rate": 3.5646005679283813e-06,
      "loss": 0.5192,
      "step": 9548
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9663261771202087,
      "learning_rate": 3.5636177919905385e-06,
      "loss": 0.4677,
      "step": 9549
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0138956308364868,
      "learning_rate": 3.5626350765337546e-06,
      "loss": 0.5469,
      "step": 9550
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0008865594863892,
      "learning_rate": 3.5616524215994052e-06,
      "loss": 0.5154,
      "step": 9551
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1205394268035889,
      "learning_rate": 3.560669827228871e-06,
      "loss": 0.552,
      "step": 9552
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0503113269805908,
      "learning_rate": 3.559687293463522e-06,
      "loss": 0.5198,
      "step": 9553
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9706525802612305,
      "learning_rate": 3.5587048203447314e-06,
      "loss": 0.5262,
      "step": 9554
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0099480152130127,
      "learning_rate": 3.557722407913865e-06,
      "loss": 0.518,
      "step": 9555
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.021026611328125,
      "learning_rate": 3.5567400562122934e-06,
      "loss": 0.4408,
      "step": 9556
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.036303997039795,
      "learning_rate": 3.5557577652813758e-06,
      "loss": 0.4817,
      "step": 9557
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0459154844284058,
      "learning_rate": 3.554775535162475e-06,
      "loss": 0.5314,
      "step": 9558
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0108460187911987,
      "learning_rate": 3.5537933658969475e-06,
      "loss": 0.471,
      "step": 9559
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0860623121261597,
      "learning_rate": 3.5528112575261525e-06,
      "loss": 0.5329,
      "step": 9560
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.072257399559021,
      "learning_rate": 3.5518292100914396e-06,
      "loss": 0.5182,
      "step": 9561
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0338189601898193,
      "learning_rate": 3.550847223634162e-06,
      "loss": 0.5529,
      "step": 9562
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.070862889289856,
      "learning_rate": 3.549865298195665e-06,
      "loss": 0.5077,
      "step": 9563
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.977547287940979,
      "learning_rate": 3.5488834338172974e-06,
      "loss": 0.5186,
      "step": 9564
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0770364999771118,
      "learning_rate": 3.547901630540399e-06,
      "loss": 0.5116,
      "step": 9565
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0419204235076904,
      "learning_rate": 3.546919888406313e-06,
      "loss": 0.5073,
      "step": 9566
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1295720338821411,
      "learning_rate": 3.5459382074563737e-06,
      "loss": 0.5172,
      "step": 9567
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9190590381622314,
      "learning_rate": 3.5449565877319175e-06,
      "loss": 0.4753,
      "step": 9568
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0490338802337646,
      "learning_rate": 3.54397502927428e-06,
      "loss": 0.4919,
      "step": 9569
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0051147937774658,
      "learning_rate": 3.5429935321247887e-06,
      "loss": 0.5308,
      "step": 9570
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1259526014328003,
      "learning_rate": 3.5420120963247706e-06,
      "loss": 0.5069,
      "step": 9571
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0539841651916504,
      "learning_rate": 3.5410307219155495e-06,
      "loss": 0.52,
      "step": 9572
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0451796054840088,
      "learning_rate": 3.540049408938452e-06,
      "loss": 0.5172,
      "step": 9573
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0502591133117676,
      "learning_rate": 3.539068157434794e-06,
      "loss": 0.485,
      "step": 9574
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0056434869766235,
      "learning_rate": 3.538086967445894e-06,
      "loss": 0.4762,
      "step": 9575
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.001711368560791,
      "learning_rate": 3.5371058390130643e-06,
      "loss": 0.5313,
      "step": 9576
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9541638493537903,
      "learning_rate": 3.536124772177621e-06,
      "loss": 0.4718,
      "step": 9577
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0295259952545166,
      "learning_rate": 3.535143766980871e-06,
      "loss": 0.493,
      "step": 9578
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0524946451187134,
      "learning_rate": 3.53416282346412e-06,
      "loss": 0.5176,
      "step": 9579
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1142582893371582,
      "learning_rate": 3.533181941668675e-06,
      "loss": 0.4635,
      "step": 9580
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9713674783706665,
      "learning_rate": 3.5322011216358325e-06,
      "loss": 0.4751,
      "step": 9581
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0555380582809448,
      "learning_rate": 3.5312203634068977e-06,
      "loss": 0.5359,
      "step": 9582
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0953872203826904,
      "learning_rate": 3.5302396670231622e-06,
      "loss": 0.5405,
      "step": 9583
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9998186230659485,
      "learning_rate": 3.529259032525923e-06,
      "loss": 0.5104,
      "step": 9584
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.030078649520874,
      "learning_rate": 3.5282784599564667e-06,
      "loss": 0.4584,
      "step": 9585
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9956438541412354,
      "learning_rate": 3.5272979493560877e-06,
      "loss": 0.5021,
      "step": 9586
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0601505041122437,
      "learning_rate": 3.5263175007660676e-06,
      "loss": 0.5392,
      "step": 9587
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9951921701431274,
      "learning_rate": 3.5253371142276915e-06,
      "loss": 0.5092,
      "step": 9588
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.18478262424469,
      "learning_rate": 3.5243567897822382e-06,
      "loss": 0.5594,
      "step": 9589
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0945976972579956,
      "learning_rate": 3.5233765274709885e-06,
      "loss": 0.5258,
      "step": 9590
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0256088972091675,
      "learning_rate": 3.5223963273352157e-06,
      "loss": 0.5315,
      "step": 9591
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0806553363800049,
      "learning_rate": 3.5214161894161948e-06,
      "loss": 0.5353,
      "step": 9592
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0250076055526733,
      "learning_rate": 3.5204361137551924e-06,
      "loss": 0.5281,
      "step": 9593
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0070712566375732,
      "learning_rate": 3.5194561003934798e-06,
      "loss": 0.4873,
      "step": 9594
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0493123531341553,
      "learning_rate": 3.5184761493723197e-06,
      "loss": 0.5123,
      "step": 9595
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9745450019836426,
      "learning_rate": 3.5174962607329755e-06,
      "loss": 0.5053,
      "step": 9596
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.979505181312561,
      "learning_rate": 3.516516434516707e-06,
      "loss": 0.5271,
      "step": 9597
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0419960021972656,
      "learning_rate": 3.5155366707647686e-06,
      "loss": 0.4916,
      "step": 9598
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0558388233184814,
      "learning_rate": 3.514556969518418e-06,
      "loss": 0.5722,
      "step": 9599
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0423963069915771,
      "learning_rate": 3.513577330818907e-06,
      "loss": 0.5179,
      "step": 9600
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0253368616104126,
      "learning_rate": 3.512597754707484e-06,
      "loss": 0.5341,
      "step": 9601
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9791843295097351,
      "learning_rate": 3.511618241225393e-06,
      "loss": 0.5181,
      "step": 9602
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.036952257156372,
      "learning_rate": 3.5106387904138804e-06,
      "loss": 0.4877,
      "step": 9603
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0005004405975342,
      "learning_rate": 3.5096594023141895e-06,
      "loss": 0.5288,
      "step": 9604
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0786027908325195,
      "learning_rate": 3.508680076967556e-06,
      "loss": 0.568,
      "step": 9605
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.108238935470581,
      "learning_rate": 3.507700814415215e-06,
      "loss": 0.5047,
      "step": 9606
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0978314876556396,
      "learning_rate": 3.5067216146984016e-06,
      "loss": 0.5321,
      "step": 9607
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0798311233520508,
      "learning_rate": 3.505742477858348e-06,
      "loss": 0.4769,
      "step": 9608
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0491946935653687,
      "learning_rate": 3.50476340393628e-06,
      "loss": 0.4781,
      "step": 9609
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.044782280921936,
      "learning_rate": 3.5037843929734216e-06,
      "loss": 0.5266,
      "step": 9610
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0892606973648071,
      "learning_rate": 3.502805445010998e-06,
      "loss": 0.4602,
      "step": 9611
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.022353172302246,
      "learning_rate": 3.5018265600902313e-06,
      "loss": 0.5044,
      "step": 9612
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0843284130096436,
      "learning_rate": 3.5008477382523355e-06,
      "loss": 0.5184,
      "step": 9613
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0725637674331665,
      "learning_rate": 3.4998689795385245e-06,
      "loss": 0.5246,
      "step": 9614
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0283935070037842,
      "learning_rate": 3.4988902839900118e-06,
      "loss": 0.5522,
      "step": 9615
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0854668617248535,
      "learning_rate": 3.4979116516480094e-06,
      "loss": 0.5249,
      "step": 9616
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.140599012374878,
      "learning_rate": 3.496933082553722e-06,
      "loss": 0.5659,
      "step": 9617
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0611871480941772,
      "learning_rate": 3.495954576748353e-06,
      "loss": 0.4834,
      "step": 9618
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.074779748916626,
      "learning_rate": 3.4949761342731025e-06,
      "loss": 0.5413,
      "step": 9619
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0762356519699097,
      "learning_rate": 3.493997755169174e-06,
      "loss": 0.5502,
      "step": 9620
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0419806241989136,
      "learning_rate": 3.4930194394777615e-06,
      "loss": 0.5192,
      "step": 9621
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9293636679649353,
      "learning_rate": 3.492041187240056e-06,
      "loss": 0.4888,
      "step": 9622
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0450637340545654,
      "learning_rate": 3.49106299849725e-06,
      "loss": 0.5002,
      "step": 9623
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0150624513626099,
      "learning_rate": 3.4900848732905348e-06,
      "loss": 0.5786,
      "step": 9624
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.049338698387146,
      "learning_rate": 3.4891068116610914e-06,
      "loss": 0.5621,
      "step": 9625
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0134360790252686,
      "learning_rate": 3.4881288136501036e-06,
      "loss": 0.5458,
      "step": 9626
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0989668369293213,
      "learning_rate": 3.487150879298753e-06,
      "loss": 0.5159,
      "step": 9627
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9743390083312988,
      "learning_rate": 3.486173008648215e-06,
      "loss": 0.5264,
      "step": 9628
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9863070249557495,
      "learning_rate": 3.485195201739665e-06,
      "loss": 0.499,
      "step": 9629
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9773442149162292,
      "learning_rate": 3.4842174586142772e-06,
      "loss": 0.4861,
      "step": 9630
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0338678359985352,
      "learning_rate": 3.4832397793132187e-06,
      "loss": 0.565,
      "step": 9631
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0463933944702148,
      "learning_rate": 3.4822621638776555e-06,
      "loss": 0.5112,
      "step": 9632
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9847338795661926,
      "learning_rate": 3.4812846123487532e-06,
      "loss": 0.4575,
      "step": 9633
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1015348434448242,
      "learning_rate": 3.4803071247676735e-06,
      "loss": 0.5258,
      "step": 9634
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9594420194625854,
      "learning_rate": 3.4793297011755746e-06,
      "loss": 0.4941,
      "step": 9635
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0367622375488281,
      "learning_rate": 3.4783523416136096e-06,
      "loss": 0.5378,
      "step": 9636
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9925249218940735,
      "learning_rate": 3.477375046122935e-06,
      "loss": 0.5082,
      "step": 9637
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.014980673789978,
      "learning_rate": 3.476397814744702e-06,
      "loss": 0.5395,
      "step": 9638
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.002394199371338,
      "learning_rate": 3.4754206475200556e-06,
      "loss": 0.5063,
      "step": 9639
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1005549430847168,
      "learning_rate": 3.4744435444901412e-06,
      "loss": 0.5004,
      "step": 9640
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0560725927352905,
      "learning_rate": 3.473466505696103e-06,
      "loss": 0.4656,
      "step": 9641
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0724444389343262,
      "learning_rate": 3.4724895311790806e-06,
      "loss": 0.4971,
      "step": 9642
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0155487060546875,
      "learning_rate": 3.4715126209802104e-06,
      "loss": 0.5048,
      "step": 9643
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.043113350868225,
      "learning_rate": 3.4705357751406256e-06,
      "loss": 0.5237,
      "step": 9644
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9849514365196228,
      "learning_rate": 3.469558993701457e-06,
      "loss": 0.471,
      "step": 9645
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0826847553253174,
      "learning_rate": 3.468582276703838e-06,
      "loss": 0.5301,
      "step": 9646
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0258899927139282,
      "learning_rate": 3.467605624188891e-06,
      "loss": 0.4994,
      "step": 9647
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9913710355758667,
      "learning_rate": 3.46662903619774e-06,
      "loss": 0.4979,
      "step": 9648
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1202266216278076,
      "learning_rate": 3.4656525127715045e-06,
      "loss": 0.5012,
      "step": 9649
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0036853551864624,
      "learning_rate": 3.464676053951307e-06,
      "loss": 0.5242,
      "step": 9650
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0133076906204224,
      "learning_rate": 3.463699659778259e-06,
      "loss": 0.5275,
      "step": 9651
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.055254578590393,
      "learning_rate": 3.4627233302934737e-06,
      "loss": 0.5477,
      "step": 9652
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0198588371276855,
      "learning_rate": 3.4617470655380597e-06,
      "loss": 0.5041,
      "step": 9653
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0016930103302002,
      "learning_rate": 3.460770865553128e-06,
      "loss": 0.4829,
      "step": 9654
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0819693803787231,
      "learning_rate": 3.4597947303797795e-06,
      "loss": 0.5676,
      "step": 9655
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0289030075073242,
      "learning_rate": 3.458818660059117e-06,
      "loss": 0.5057,
      "step": 9656
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0696585178375244,
      "learning_rate": 3.4578426546322403e-06,
      "loss": 0.5772,
      "step": 9657
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9678754806518555,
      "learning_rate": 3.4568667141402425e-06,
      "loss": 0.5178,
      "step": 9658
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0327085256576538,
      "learning_rate": 3.4558908386242208e-06,
      "loss": 0.4855,
      "step": 9659
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0984344482421875,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.5505,
      "step": 9660
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1058754920959473,
      "learning_rate": 3.4539392826844607e-06,
      "loss": 0.5344,
      "step": 9661
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0699716806411743,
      "learning_rate": 3.4529636023428935e-06,
      "loss": 0.4925,
      "step": 9662
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0517855882644653,
      "learning_rate": 3.4519879871416505e-06,
      "loss": 0.4727,
      "step": 9663
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0027638673782349,
      "learning_rate": 3.451012437121806e-06,
      "loss": 0.4895,
      "step": 9664
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0605578422546387,
      "learning_rate": 3.4500369523244414e-06,
      "loss": 0.4929,
      "step": 9665
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9826477766036987,
      "learning_rate": 3.4490615327906264e-06,
      "loss": 0.5018,
      "step": 9666
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.012223482131958,
      "learning_rate": 3.448086178561436e-06,
      "loss": 0.49,
      "step": 9667
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0503581762313843,
      "learning_rate": 3.447110889677938e-06,
      "loss": 0.5748,
      "step": 9668
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1112549304962158,
      "learning_rate": 3.4461356661811997e-06,
      "loss": 0.504,
      "step": 9669
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9363526701927185,
      "learning_rate": 3.4451605081122797e-06,
      "loss": 0.4459,
      "step": 9670
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1454665660858154,
      "learning_rate": 3.4441854155122446e-06,
      "loss": 0.5197,
      "step": 9671
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0244606733322144,
      "learning_rate": 3.443210388422148e-06,
      "loss": 0.5044,
      "step": 9672
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9871229529380798,
      "learning_rate": 3.4422354268830473e-06,
      "loss": 0.5346,
      "step": 9673
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0190272331237793,
      "learning_rate": 3.441260530935994e-06,
      "loss": 0.5098,
      "step": 9674
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0977753400802612,
      "learning_rate": 3.4402857006220353e-06,
      "loss": 0.5826,
      "step": 9675
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1146341562271118,
      "learning_rate": 3.439310935982221e-06,
      "loss": 0.5425,
      "step": 9676
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0122137069702148,
      "learning_rate": 3.4383362370575947e-06,
      "loss": 0.4889,
      "step": 9677
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0664081573486328,
      "learning_rate": 3.4373616038891966e-06,
      "loss": 0.5433,
      "step": 9678
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9652659893035889,
      "learning_rate": 3.4363870365180634e-06,
      "loss": 0.4762,
      "step": 9679
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0801855325698853,
      "learning_rate": 3.435412534985234e-06,
      "loss": 0.5382,
      "step": 9680
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0500061511993408,
      "learning_rate": 3.4344380993317404e-06,
      "loss": 0.5247,
      "step": 9681
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1069306135177612,
      "learning_rate": 3.433463729598613e-06,
      "loss": 0.5384,
      "step": 9682
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0150129795074463,
      "learning_rate": 3.432489425826876e-06,
      "loss": 0.5031,
      "step": 9683
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.047875165939331,
      "learning_rate": 3.431515188057557e-06,
      "loss": 0.4805,
      "step": 9684
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9312357306480408,
      "learning_rate": 3.4305410163316788e-06,
      "loss": 0.4782,
      "step": 9685
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0885319709777832,
      "learning_rate": 3.429566910690258e-06,
      "loss": 0.5127,
      "step": 9686
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1217483282089233,
      "learning_rate": 3.42859287117431e-06,
      "loss": 0.5348,
      "step": 9687
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1635698080062866,
      "learning_rate": 3.42761889782485e-06,
      "loss": 0.5381,
      "step": 9688
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1344828605651855,
      "learning_rate": 3.4266449906828897e-06,
      "loss": 0.5175,
      "step": 9689
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1053919792175293,
      "learning_rate": 3.4256711497894346e-06,
      "loss": 0.53,
      "step": 9690
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9443258047103882,
      "learning_rate": 3.4246973751854917e-06,
      "loss": 0.4928,
      "step": 9691
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0636018514633179,
      "learning_rate": 3.42372366691206e-06,
      "loss": 0.5219,
      "step": 9692
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1260120868682861,
      "learning_rate": 3.422750025010143e-06,
      "loss": 0.4891,
      "step": 9693
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0381945371627808,
      "learning_rate": 3.421776449520735e-06,
      "loss": 0.5239,
      "step": 9694
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.145067572593689,
      "learning_rate": 3.4208029404848315e-06,
      "loss": 0.524,
      "step": 9695
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0171116590499878,
      "learning_rate": 3.4198294979434207e-06,
      "loss": 0.5243,
      "step": 9696
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0491117238998413,
      "learning_rate": 3.418856121937494e-06,
      "loss": 0.4638,
      "step": 9697
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0508257150650024,
      "learning_rate": 3.4178828125080354e-06,
      "loss": 0.5367,
      "step": 9698
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.080831527709961,
      "learning_rate": 3.4169095696960287e-06,
      "loss": 0.573,
      "step": 9699
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9865775108337402,
      "learning_rate": 3.4159363935424505e-06,
      "loss": 0.5196,
      "step": 9700
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.059741497039795,
      "learning_rate": 3.4149632840882817e-06,
      "loss": 0.5221,
      "step": 9701
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9895997643470764,
      "learning_rate": 3.413990241374495e-06,
      "loss": 0.5056,
      "step": 9702
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0225650072097778,
      "learning_rate": 3.4130172654420623e-06,
      "loss": 0.4945,
      "step": 9703
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0705265998840332,
      "learning_rate": 3.4120443563319527e-06,
      "loss": 0.4667,
      "step": 9704
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0751066207885742,
      "learning_rate": 3.4110715140851286e-06,
      "loss": 0.4826,
      "step": 9705
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0618090629577637,
      "learning_rate": 3.4100987387425566e-06,
      "loss": 0.5378,
      "step": 9706
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0707107782363892,
      "learning_rate": 3.4091260303451967e-06,
      "loss": 0.5091,
      "step": 9707
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.029998540878296,
      "learning_rate": 3.4081533889340056e-06,
      "loss": 0.4905,
      "step": 9708
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0424970388412476,
      "learning_rate": 3.407180814549935e-06,
      "loss": 0.5162,
      "step": 9709
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.09152352809906,
      "learning_rate": 3.4062083072339415e-06,
      "loss": 0.5531,
      "step": 9710
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0747696161270142,
      "learning_rate": 3.4052358670269727e-06,
      "loss": 0.556,
      "step": 9711
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9668502807617188,
      "learning_rate": 3.4042634939699728e-06,
      "loss": 0.4954,
      "step": 9712
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0281851291656494,
      "learning_rate": 3.4032911881038842e-06,
      "loss": 0.5478,
      "step": 9713
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.090576410293579,
      "learning_rate": 3.4023189494696506e-06,
      "loss": 0.5851,
      "step": 9714
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1164906024932861,
      "learning_rate": 3.401346778108209e-06,
      "loss": 0.5551,
      "step": 9715
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0226433277130127,
      "learning_rate": 3.4003746740604925e-06,
      "loss": 0.53,
      "step": 9716
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0139816999435425,
      "learning_rate": 3.399402637367433e-06,
      "loss": 0.5212,
      "step": 9717
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0398943424224854,
      "learning_rate": 3.398430668069961e-06,
      "loss": 0.5091,
      "step": 9718
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9936200380325317,
      "learning_rate": 3.3974587662090026e-06,
      "loss": 0.5545,
      "step": 9719
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1318458318710327,
      "learning_rate": 3.396486931825481e-06,
      "loss": 0.5478,
      "step": 9720
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9907205700874329,
      "learning_rate": 3.395515164960316e-06,
      "loss": 0.4836,
      "step": 9721
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0219179391860962,
      "learning_rate": 3.394543465654424e-06,
      "loss": 0.53,
      "step": 9722
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0675764083862305,
      "learning_rate": 3.393571833948724e-06,
      "loss": 0.5186,
      "step": 9723
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0911437273025513,
      "learning_rate": 3.3926002698841253e-06,
      "loss": 0.4721,
      "step": 9724
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0146348476409912,
      "learning_rate": 3.3916287735015375e-06,
      "loss": 0.517,
      "step": 9725
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0276731252670288,
      "learning_rate": 3.390657344841865e-06,
      "loss": 0.459,
      "step": 9726
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1111414432525635,
      "learning_rate": 3.3896859839460155e-06,
      "loss": 0.5204,
      "step": 9727
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0293622016906738,
      "learning_rate": 3.3887146908548875e-06,
      "loss": 0.558,
      "step": 9728
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0552021265029907,
      "learning_rate": 3.3877434656093777e-06,
      "loss": 0.4724,
      "step": 9729
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.110007405281067,
      "learning_rate": 3.3867723082503807e-06,
      "loss": 0.5886,
      "step": 9730
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.009547472000122,
      "learning_rate": 3.385801218818792e-06,
      "loss": 0.4899,
      "step": 9731
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1629246473312378,
      "learning_rate": 3.384830197355499e-06,
      "loss": 0.5559,
      "step": 9732
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0362046957015991,
      "learning_rate": 3.383859243901385e-06,
      "loss": 0.4975,
      "step": 9733
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9874945878982544,
      "learning_rate": 3.3828883584973364e-06,
      "loss": 0.5235,
      "step": 9734
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0950794219970703,
      "learning_rate": 3.3819175411842353e-06,
      "loss": 0.5257,
      "step": 9735
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1203292608261108,
      "learning_rate": 3.3809467920029574e-06,
      "loss": 0.5148,
      "step": 9736
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.057608962059021,
      "learning_rate": 3.3799761109943775e-06,
      "loss": 0.4833,
      "step": 9737
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9889419078826904,
      "learning_rate": 3.3790054981993683e-06,
      "loss": 0.5243,
      "step": 9738
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0515849590301514,
      "learning_rate": 3.3780349536587965e-06,
      "loss": 0.4974,
      "step": 9739
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.059637427330017,
      "learning_rate": 3.377064477413533e-06,
      "loss": 0.5346,
      "step": 9740
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9503380656242371,
      "learning_rate": 3.376094069504437e-06,
      "loss": 0.4479,
      "step": 9741
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1286183595657349,
      "learning_rate": 3.3751237299723715e-06,
      "loss": 0.5201,
      "step": 9742
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0307422876358032,
      "learning_rate": 3.3741534588581915e-06,
      "loss": 0.542,
      "step": 9743
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.067309856414795,
      "learning_rate": 3.3731832562027555e-06,
      "loss": 0.5095,
      "step": 9744
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1034531593322754,
      "learning_rate": 3.372213122046912e-06,
      "loss": 0.504,
      "step": 9745
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.020670771598816,
      "learning_rate": 3.3712430564315124e-06,
      "loss": 0.4971,
      "step": 9746
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0703192949295044,
      "learning_rate": 3.3702730593974e-06,
      "loss": 0.5267,
      "step": 9747
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0400869846343994,
      "learning_rate": 3.3693031309854214e-06,
      "loss": 0.5108,
      "step": 9748
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9968366622924805,
      "learning_rate": 3.3683332712364138e-06,
      "loss": 0.488,
      "step": 9749
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0771602392196655,
      "learning_rate": 3.3673634801912175e-06,
      "loss": 0.5276,
      "step": 9750
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9836703538894653,
      "learning_rate": 3.3663937578906642e-06,
      "loss": 0.4721,
      "step": 9751
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.000105381011963,
      "learning_rate": 3.365424104375587e-06,
      "loss": 0.4941,
      "step": 9752
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9447736740112305,
      "learning_rate": 3.3644545196868146e-06,
      "loss": 0.5152,
      "step": 9753
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9786331057548523,
      "learning_rate": 3.3634850038651734e-06,
      "loss": 0.5285,
      "step": 9754
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0418307781219482,
      "learning_rate": 3.362515556951485e-06,
      "loss": 0.488,
      "step": 9755
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9895498752593994,
      "learning_rate": 3.36154617898657e-06,
      "loss": 0.4815,
      "step": 9756
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0580096244812012,
      "learning_rate": 3.360576870011246e-06,
      "loss": 0.4725,
      "step": 9757
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0222854614257812,
      "learning_rate": 3.3596076300663273e-06,
      "loss": 0.5171,
      "step": 9758
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0051594972610474,
      "learning_rate": 3.3586384591926235e-06,
      "loss": 0.5213,
      "step": 9759
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0701817274093628,
      "learning_rate": 3.3576693574309447e-06,
      "loss": 0.5393,
      "step": 9760
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0076082944869995,
      "learning_rate": 3.3567003248220966e-06,
      "loss": 0.5361,
      "step": 9761
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0037952661514282,
      "learning_rate": 3.355731361406882e-06,
      "loss": 0.4991,
      "step": 9762
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0817643404006958,
      "learning_rate": 3.354762467226098e-06,
      "loss": 0.5282,
      "step": 9763
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0783755779266357,
      "learning_rate": 3.3537936423205435e-06,
      "loss": 0.5244,
      "step": 9764
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0131877660751343,
      "learning_rate": 3.352824886731012e-06,
      "loss": 0.4696,
      "step": 9765
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0318191051483154,
      "learning_rate": 3.351856200498296e-06,
      "loss": 0.546,
      "step": 9766
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9506867527961731,
      "learning_rate": 3.3508875836631806e-06,
      "loss": 0.5074,
      "step": 9767
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9747241735458374,
      "learning_rate": 3.3499190362664523e-06,
      "loss": 0.4476,
      "step": 9768
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9470990896224976,
      "learning_rate": 3.3489505583488925e-06,
      "loss": 0.4749,
      "step": 9769
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0915932655334473,
      "learning_rate": 3.3479821499512823e-06,
      "loss": 0.5607,
      "step": 9770
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0824368000030518,
      "learning_rate": 3.347013811114396e-06,
      "loss": 0.5394,
      "step": 9771
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0586308240890503,
      "learning_rate": 3.346045541879009e-06,
      "loss": 0.4506,
      "step": 9772
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9798187017440796,
      "learning_rate": 3.3450773422858886e-06,
      "loss": 0.5374,
      "step": 9773
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9897297024726868,
      "learning_rate": 3.3441092123758055e-06,
      "loss": 0.4729,
      "step": 9774
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.115580677986145,
      "learning_rate": 3.3431411521895228e-06,
      "loss": 0.4852,
      "step": 9775
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.088579773902893,
      "learning_rate": 3.342173161767803e-06,
      "loss": 0.5668,
      "step": 9776
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.986916184425354,
      "learning_rate": 3.341205241151403e-06,
      "loss": 0.4979,
      "step": 9777
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0729726552963257,
      "learning_rate": 3.3402373903810807e-06,
      "loss": 0.5086,
      "step": 9778
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9873821139335632,
      "learning_rate": 3.3392696094975875e-06,
      "loss": 0.4812,
      "step": 9779
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0625243186950684,
      "learning_rate": 3.338301898541675e-06,
      "loss": 0.5295,
      "step": 9780
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0633163452148438,
      "learning_rate": 3.337334257554086e-06,
      "loss": 0.5408,
      "step": 9781
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9733723402023315,
      "learning_rate": 3.3363666865755708e-06,
      "loss": 0.5166,
      "step": 9782
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1456708908081055,
      "learning_rate": 3.335399185646865e-06,
      "loss": 0.5323,
      "step": 9783
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9787518382072449,
      "learning_rate": 3.33443175480871e-06,
      "loss": 0.4762,
      "step": 9784
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0292288064956665,
      "learning_rate": 3.3334643941018398e-06,
      "loss": 0.4775,
      "step": 9785
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0584862232208252,
      "learning_rate": 3.3324971035669844e-06,
      "loss": 0.5477,
      "step": 9786
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0264787673950195,
      "learning_rate": 3.3315298832448762e-06,
      "loss": 0.4961,
      "step": 9787
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.108717918395996,
      "learning_rate": 3.330562733176242e-06,
      "loss": 0.4933,
      "step": 9788
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1448596715927124,
      "learning_rate": 3.3295956534018033e-06,
      "loss": 0.5257,
      "step": 9789
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0974916219711304,
      "learning_rate": 3.328628643962278e-06,
      "loss": 0.5133,
      "step": 9790
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9968658685684204,
      "learning_rate": 3.3276617048983876e-06,
      "loss": 0.4774,
      "step": 9791
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.061164140701294,
      "learning_rate": 3.326694836250847e-06,
      "loss": 0.5387,
      "step": 9792
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0636687278747559,
      "learning_rate": 3.325728038060365e-06,
      "loss": 0.5125,
      "step": 9793
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0595797300338745,
      "learning_rate": 3.324761310367649e-06,
      "loss": 0.5199,
      "step": 9794
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.101511001586914,
      "learning_rate": 3.3237946532134077e-06,
      "loss": 0.5327,
      "step": 9795
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.987757682800293,
      "learning_rate": 3.322828066638343e-06,
      "loss": 0.4707,
      "step": 9796
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0710266828536987,
      "learning_rate": 3.321861550683154e-06,
      "loss": 0.5579,
      "step": 9797
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.121856927871704,
      "learning_rate": 3.3208951053885367e-06,
      "loss": 0.5582,
      "step": 9798
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0450007915496826,
      "learning_rate": 3.3199287307951844e-06,
      "loss": 0.528,
      "step": 9799
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0158655643463135,
      "learning_rate": 3.31896242694379e-06,
      "loss": 0.5187,
      "step": 9800
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2039495706558228,
      "learning_rate": 3.317996193875041e-06,
      "loss": 0.5636,
      "step": 9801
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0527689456939697,
      "learning_rate": 3.3170300316296194e-06,
      "loss": 0.5362,
      "step": 9802
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9969490170478821,
      "learning_rate": 3.3160639402482077e-06,
      "loss": 0.4939,
      "step": 9803
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1049686670303345,
      "learning_rate": 3.3150979197714874e-06,
      "loss": 0.5056,
      "step": 9804
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0586570501327515,
      "learning_rate": 3.314131970240132e-06,
      "loss": 0.5425,
      "step": 9805
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0372644662857056,
      "learning_rate": 3.3131660916948137e-06,
      "loss": 0.5566,
      "step": 9806
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.983227550983429,
      "learning_rate": 3.3122002841762023e-06,
      "loss": 0.5616,
      "step": 9807
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0370455980300903,
      "learning_rate": 3.311234547724968e-06,
      "loss": 0.5537,
      "step": 9808
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0073283910751343,
      "learning_rate": 3.310268882381772e-06,
      "loss": 0.5028,
      "step": 9809
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0266960859298706,
      "learning_rate": 3.3093032881872738e-06,
      "loss": 0.4729,
      "step": 9810
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.022322177886963,
      "learning_rate": 3.3083377651821314e-06,
      "loss": 0.4925,
      "step": 9811
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9509308934211731,
      "learning_rate": 3.3073723134070033e-06,
      "loss": 0.5048,
      "step": 9812
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0213093757629395,
      "learning_rate": 3.3064069329025394e-06,
      "loss": 0.531,
      "step": 9813
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0336143970489502,
      "learning_rate": 3.3054416237093863e-06,
      "loss": 0.5419,
      "step": 9814
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9689751863479614,
      "learning_rate": 3.3044763858681928e-06,
      "loss": 0.4507,
      "step": 9815
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0528861284255981,
      "learning_rate": 3.303511219419598e-06,
      "loss": 0.5247,
      "step": 9816
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0158718824386597,
      "learning_rate": 3.3025461244042462e-06,
      "loss": 0.4837,
      "step": 9817
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0771912336349487,
      "learning_rate": 3.3015811008627707e-06,
      "loss": 0.5347,
      "step": 9818
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1466127634048462,
      "learning_rate": 3.3006161488358084e-06,
      "loss": 0.5163,
      "step": 9819
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0220280885696411,
      "learning_rate": 3.299651268363986e-06,
      "loss": 0.5152,
      "step": 9820
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0260629653930664,
      "learning_rate": 3.298686459487936e-06,
      "loss": 0.4703,
      "step": 9821
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0086876153945923,
      "learning_rate": 3.2977217222482794e-06,
      "loss": 0.4937,
      "step": 9822
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.048424482345581,
      "learning_rate": 3.29675705668564e-06,
      "loss": 0.5067,
      "step": 9823
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9417296648025513,
      "learning_rate": 3.2957924628406347e-06,
      "loss": 0.4311,
      "step": 9824
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0503716468811035,
      "learning_rate": 3.2948279407538813e-06,
      "loss": 0.474,
      "step": 9825
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0028961896896362,
      "learning_rate": 3.2938634904659903e-06,
      "loss": 0.4969,
      "step": 9826
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0006153583526611,
      "learning_rate": 3.2928991120175747e-06,
      "loss": 0.5085,
      "step": 9827
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.036969542503357,
      "learning_rate": 3.2919348054492363e-06,
      "loss": 0.4478,
      "step": 9828
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0671666860580444,
      "learning_rate": 3.2909705708015834e-06,
      "loss": 0.5363,
      "step": 9829
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0576975345611572,
      "learning_rate": 3.290006408115213e-06,
      "loss": 0.5515,
      "step": 9830
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.024174451828003,
      "learning_rate": 3.289042317430726e-06,
      "loss": 0.5003,
      "step": 9831
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9663321375846863,
      "learning_rate": 3.2880782987887128e-06,
      "loss": 0.489,
      "step": 9832
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9781193137168884,
      "learning_rate": 3.2871143522297672e-06,
      "loss": 0.5053,
      "step": 9833
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.021514892578125,
      "learning_rate": 3.286150477794479e-06,
      "loss": 0.517,
      "step": 9834
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9754778742790222,
      "learning_rate": 3.2851866755234324e-06,
      "loss": 0.5128,
      "step": 9835
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9535737633705139,
      "learning_rate": 3.2842229454572084e-06,
      "loss": 0.4931,
      "step": 9836
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0680464506149292,
      "learning_rate": 3.2832592876363866e-06,
      "loss": 0.5357,
      "step": 9837
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0458132028579712,
      "learning_rate": 3.2822957021015455e-06,
      "loss": 0.4679,
      "step": 9838
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0636483430862427,
      "learning_rate": 3.2813321888932573e-06,
      "loss": 0.4727,
      "step": 9839
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9563536643981934,
      "learning_rate": 3.2803687480520905e-06,
      "loss": 0.4623,
      "step": 9840
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0053445100784302,
      "learning_rate": 3.279405379618613e-06,
      "loss": 0.4782,
      "step": 9841
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0068389177322388,
      "learning_rate": 3.278442083633392e-06,
      "loss": 0.4693,
      "step": 9842
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0501048564910889,
      "learning_rate": 3.277478860136985e-06,
      "loss": 0.5249,
      "step": 9843
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0450025796890259,
      "learning_rate": 3.276515709169951e-06,
      "loss": 0.5085,
      "step": 9844
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9735075831413269,
      "learning_rate": 3.2755526307728447e-06,
      "loss": 0.509,
      "step": 9845
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9778958559036255,
      "learning_rate": 3.2745896249862166e-06,
      "loss": 0.4734,
      "step": 9846
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1158515214920044,
      "learning_rate": 3.2736266918506195e-06,
      "loss": 0.5633,
      "step": 9847
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.112127661705017,
      "learning_rate": 3.272663831406595e-06,
      "loss": 0.5784,
      "step": 9848
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0029457807540894,
      "learning_rate": 3.2717010436946894e-06,
      "loss": 0.5044,
      "step": 9849
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.359940767288208,
      "learning_rate": 3.270738328755438e-06,
      "loss": 0.5601,
      "step": 9850
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.055325984954834,
      "learning_rate": 3.269775686629383e-06,
      "loss": 0.5604,
      "step": 9851
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.098065733909607,
      "learning_rate": 3.2688131173570523e-06,
      "loss": 0.548,
      "step": 9852
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0677322149276733,
      "learning_rate": 3.2678506209789805e-06,
      "loss": 0.5502,
      "step": 9853
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.995072603225708,
      "learning_rate": 3.2668881975356915e-06,
      "loss": 0.4901,
      "step": 9854
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.11382257938385,
      "learning_rate": 3.2659258470677137e-06,
      "loss": 0.5686,
      "step": 9855
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9235460758209229,
      "learning_rate": 3.2649635696155646e-06,
      "loss": 0.4439,
      "step": 9856
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0282461643218994,
      "learning_rate": 3.2640013652197654e-06,
      "loss": 0.5014,
      "step": 9857
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0538965463638306,
      "learning_rate": 3.263039233920827e-06,
      "loss": 0.5484,
      "step": 9858
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.136786699295044,
      "learning_rate": 3.2620771757592663e-06,
      "loss": 0.4908,
      "step": 9859
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2098864316940308,
      "learning_rate": 3.261115190775589e-06,
      "loss": 0.4867,
      "step": 9860
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1039862632751465,
      "learning_rate": 3.2601532790103026e-06,
      "loss": 0.4787,
      "step": 9861
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1275825500488281,
      "learning_rate": 3.259191440503909e-06,
      "loss": 0.5362,
      "step": 9862
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0742462873458862,
      "learning_rate": 3.258229675296907e-06,
      "loss": 0.5357,
      "step": 9863
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.030516266822815,
      "learning_rate": 3.257267983429794e-06,
      "loss": 0.5373,
      "step": 9864
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0194627046585083,
      "learning_rate": 3.2563063649430647e-06,
      "loss": 0.5076,
      "step": 9865
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0631455183029175,
      "learning_rate": 3.25534481987721e-06,
      "loss": 0.5389,
      "step": 9866
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0659403800964355,
      "learning_rate": 3.2543833482727123e-06,
      "loss": 0.5412,
      "step": 9867
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9998225569725037,
      "learning_rate": 3.2534219501700603e-06,
      "loss": 0.4862,
      "step": 9868
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.070391058921814,
      "learning_rate": 3.252460625609736e-06,
      "loss": 0.4995,
      "step": 9869
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0833836793899536,
      "learning_rate": 3.251499374632214e-06,
      "loss": 0.5161,
      "step": 9870
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0146968364715576,
      "learning_rate": 3.2505381972779704e-06,
      "loss": 0.4949,
      "step": 9871
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0704102516174316,
      "learning_rate": 3.249577093587477e-06,
      "loss": 0.4979,
      "step": 9872
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0711511373519897,
      "learning_rate": 3.2486160636012054e-06,
      "loss": 0.5424,
      "step": 9873
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9882245659828186,
      "learning_rate": 3.2476551073596173e-06,
      "loss": 0.5129,
      "step": 9874
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.091054081916809,
      "learning_rate": 3.246694224903175e-06,
      "loss": 0.4935,
      "step": 9875
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0546417236328125,
      "learning_rate": 3.245733416272341e-06,
      "loss": 0.5292,
      "step": 9876
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0956344604492188,
      "learning_rate": 3.2447726815075715e-06,
      "loss": 0.5185,
      "step": 9877
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0548691749572754,
      "learning_rate": 3.243812020649318e-06,
      "loss": 0.4968,
      "step": 9878
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1286178827285767,
      "learning_rate": 3.24285143373803e-06,
      "loss": 0.5511,
      "step": 9879
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0500680208206177,
      "learning_rate": 3.241890920814154e-06,
      "loss": 0.4643,
      "step": 9880
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0624537467956543,
      "learning_rate": 3.2409304819181377e-06,
      "loss": 0.5021,
      "step": 9881
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0358778238296509,
      "learning_rate": 3.2399701170904197e-06,
      "loss": 0.4912,
      "step": 9882
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9780733585357666,
      "learning_rate": 3.239009826371436e-06,
      "loss": 0.5024,
      "step": 9883
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0151331424713135,
      "learning_rate": 3.238049609801621e-06,
      "loss": 0.5204,
      "step": 9884
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0676971673965454,
      "learning_rate": 3.2370894674214102e-06,
      "loss": 0.5266,
      "step": 9885
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9793708920478821,
      "learning_rate": 3.2361293992712295e-06,
      "loss": 0.4243,
      "step": 9886
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9515348672866821,
      "learning_rate": 3.2351694053915027e-06,
      "loss": 0.502,
      "step": 9887
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0956379175186157,
      "learning_rate": 3.2342094858226514e-06,
      "loss": 0.5517,
      "step": 9888
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9976474642753601,
      "learning_rate": 3.233249640605098e-06,
      "loss": 0.4876,
      "step": 9889
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0082616806030273,
      "learning_rate": 3.232289869779256e-06,
      "loss": 0.48,
      "step": 9890
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0102382898330688,
      "learning_rate": 3.231330173385537e-06,
      "loss": 0.5028,
      "step": 9891
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9744532108306885,
      "learning_rate": 3.2303705514643537e-06,
      "loss": 0.5359,
      "step": 9892
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0318796634674072,
      "learning_rate": 3.229411004056108e-06,
      "loss": 0.5089,
      "step": 9893
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0769177675247192,
      "learning_rate": 3.2284515312012056e-06,
      "loss": 0.4983,
      "step": 9894
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0182427167892456,
      "learning_rate": 3.2274921329400484e-06,
      "loss": 0.5398,
      "step": 9895
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9979657530784607,
      "learning_rate": 3.226532809313031e-06,
      "loss": 0.4801,
      "step": 9896
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0415728092193604,
      "learning_rate": 3.2255735603605454e-06,
      "loss": 0.5321,
      "step": 9897
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.053540825843811,
      "learning_rate": 3.2246143861229857e-06,
      "loss": 0.4922,
      "step": 9898
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0729095935821533,
      "learning_rate": 3.223655286640739e-06,
      "loss": 0.5117,
      "step": 9899
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0364995002746582,
      "learning_rate": 3.2226962619541885e-06,
      "loss": 0.5212,
      "step": 9900
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0050946474075317,
      "learning_rate": 3.221737312103714e-06,
      "loss": 0.4442,
      "step": 9901
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0963795185089111,
      "learning_rate": 3.2207784371296957e-06,
      "loss": 0.5423,
      "step": 9902
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.034480333328247,
      "learning_rate": 3.2198196370725095e-06,
      "loss": 0.5377,
      "step": 9903
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0671687126159668,
      "learning_rate": 3.218860911972525e-06,
      "loss": 0.5646,
      "step": 9904
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.066817283630371,
      "learning_rate": 3.2179022618701093e-06,
      "loss": 0.5549,
      "step": 9905
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.04256272315979,
      "learning_rate": 3.2169436868056316e-06,
      "loss": 0.5206,
      "step": 9906
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0120028257369995,
      "learning_rate": 3.215985186819453e-06,
      "loss": 0.5276,
      "step": 9907
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9782642722129822,
      "learning_rate": 3.2150267619519326e-06,
      "loss": 0.5158,
      "step": 9908
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0254771709442139,
      "learning_rate": 3.214068412243424e-06,
      "loss": 0.4962,
      "step": 9909
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0787056684494019,
      "learning_rate": 3.213110137734281e-06,
      "loss": 0.5418,
      "step": 9910
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0449023246765137,
      "learning_rate": 3.2121519384648558e-06,
      "loss": 0.5456,
      "step": 9911
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9803979992866516,
      "learning_rate": 3.211193814475494e-06,
      "loss": 0.5253,
      "step": 9912
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.007240653038025,
      "learning_rate": 3.2102357658065357e-06,
      "loss": 0.5351,
      "step": 9913
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0950753688812256,
      "learning_rate": 3.2092777924983224e-06,
      "loss": 0.56,
      "step": 9914
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.057042121887207,
      "learning_rate": 3.208319894591194e-06,
      "loss": 0.534,
      "step": 9915
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.131855845451355,
      "learning_rate": 3.207362072125482e-06,
      "loss": 0.5138,
      "step": 9916
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0062309503555298,
      "learning_rate": 3.2064043251415166e-06,
      "loss": 0.4759,
      "step": 9917
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0769048929214478,
      "learning_rate": 3.2054466536796236e-06,
      "loss": 0.5153,
      "step": 9918
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0601894855499268,
      "learning_rate": 3.2044890577801317e-06,
      "loss": 0.5159,
      "step": 9919
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.018688440322876,
      "learning_rate": 3.2035315374833596e-06,
      "loss": 0.5008,
      "step": 9920
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0696890354156494,
      "learning_rate": 3.2025740928296235e-06,
      "loss": 0.5138,
      "step": 9921
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0606672763824463,
      "learning_rate": 3.201616723859241e-06,
      "loss": 0.5422,
      "step": 9922
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.055463194847107,
      "learning_rate": 3.20065943061252e-06,
      "loss": 0.5375,
      "step": 9923
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9518321752548218,
      "learning_rate": 3.199702213129773e-06,
      "loss": 0.4758,
      "step": 9924
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0561015605926514,
      "learning_rate": 3.1987450714513018e-06,
      "loss": 0.5227,
      "step": 9925
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0529154539108276,
      "learning_rate": 3.1977880056174105e-06,
      "loss": 0.5044,
      "step": 9926
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9994805455207825,
      "learning_rate": 3.196831015668396e-06,
      "loss": 0.5001,
      "step": 9927
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1532816886901855,
      "learning_rate": 3.195874101644555e-06,
      "loss": 0.5132,
      "step": 9928
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9695237874984741,
      "learning_rate": 3.194917263586179e-06,
      "loss": 0.4726,
      "step": 9929
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9463463425636292,
      "learning_rate": 3.1939605015335588e-06,
      "loss": 0.4794,
      "step": 9930
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0079615116119385,
      "learning_rate": 3.193003815526977e-06,
      "loss": 0.512,
      "step": 9931
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.112295150756836,
      "learning_rate": 3.192047205606721e-06,
      "loss": 0.5726,
      "step": 9932
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0682371854782104,
      "learning_rate": 3.1910906718130665e-06,
      "loss": 0.5121,
      "step": 9933
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0254143476486206,
      "learning_rate": 3.1901342141862917e-06,
      "loss": 0.5264,
      "step": 9934
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0990660190582275,
      "learning_rate": 3.1891778327666673e-06,
      "loss": 0.5597,
      "step": 9935
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0980420112609863,
      "learning_rate": 3.1882215275944673e-06,
      "loss": 0.5718,
      "step": 9936
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0230224132537842,
      "learning_rate": 3.187265298709954e-06,
      "loss": 0.502,
      "step": 9937
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0240975618362427,
      "learning_rate": 3.1863091461533945e-06,
      "loss": 0.4951,
      "step": 9938
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.000030755996704,
      "learning_rate": 3.1853530699650483e-06,
      "loss": 0.5177,
      "step": 9939
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0125668048858643,
      "learning_rate": 3.184397070185169e-06,
      "loss": 0.5508,
      "step": 9940
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9873566627502441,
      "learning_rate": 3.183441146854014e-06,
      "loss": 0.4808,
      "step": 9941
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.15785813331604,
      "learning_rate": 3.182485300011834e-06,
      "loss": 0.5752,
      "step": 9942
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.069769024848938,
      "learning_rate": 3.181529529698875e-06,
      "loss": 0.4809,
      "step": 9943
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9738884568214417,
      "learning_rate": 3.1805738359553796e-06,
      "loss": 0.4969,
      "step": 9944
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9629955291748047,
      "learning_rate": 3.1796182188215917e-06,
      "loss": 0.4698,
      "step": 9945
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0184857845306396,
      "learning_rate": 3.1786626783377494e-06,
      "loss": 0.5016,
      "step": 9946
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0259008407592773,
      "learning_rate": 3.177707214544086e-06,
      "loss": 0.5089,
      "step": 9947
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9904293417930603,
      "learning_rate": 3.1767518274808298e-06,
      "loss": 0.477,
      "step": 9948
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0364230871200562,
      "learning_rate": 3.175796517188212e-06,
      "loss": 0.5233,
      "step": 9949
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.961456298828125,
      "learning_rate": 3.174841283706459e-06,
      "loss": 0.483,
      "step": 9950
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9019849300384521,
      "learning_rate": 3.17388612707579e-06,
      "loss": 0.5053,
      "step": 9951
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0838817358016968,
      "learning_rate": 3.172931047336421e-06,
      "loss": 0.5164,
      "step": 9952
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0342077016830444,
      "learning_rate": 3.1719760445285712e-06,
      "loss": 0.4517,
      "step": 9953
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0971794128417969,
      "learning_rate": 3.1710211186924524e-06,
      "loss": 0.4854,
      "step": 9954
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.061440348625183,
      "learning_rate": 3.170066269868271e-06,
      "loss": 0.5214,
      "step": 9955
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1033830642700195,
      "learning_rate": 3.169111498096232e-06,
      "loss": 0.5152,
      "step": 9956
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9737071394920349,
      "learning_rate": 3.1681568034165383e-06,
      "loss": 0.4773,
      "step": 9957
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9737476110458374,
      "learning_rate": 3.167202185869391e-06,
      "loss": 0.5052,
      "step": 9958
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.07955002784729,
      "learning_rate": 3.166247645494982e-06,
      "loss": 0.5298,
      "step": 9959
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0375590324401855,
      "learning_rate": 3.1652931823335074e-06,
      "loss": 0.4933,
      "step": 9960
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0839290618896484,
      "learning_rate": 3.164338796425152e-06,
      "loss": 0.5463,
      "step": 9961
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0975672006607056,
      "learning_rate": 3.163384487810106e-06,
      "loss": 0.4767,
      "step": 9962
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0232704877853394,
      "learning_rate": 3.162430256528549e-06,
      "loss": 0.5154,
      "step": 9963
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0841917991638184,
      "learning_rate": 3.161476102620663e-06,
      "loss": 0.5422,
      "step": 9964
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0752722024917603,
      "learning_rate": 3.16052202612662e-06,
      "loss": 0.4876,
      "step": 9965
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0556458234786987,
      "learning_rate": 3.159568027086598e-06,
      "loss": 0.5505,
      "step": 9966
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0755393505096436,
      "learning_rate": 3.1586141055407627e-06,
      "loss": 0.5475,
      "step": 9967
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9437804818153381,
      "learning_rate": 3.157660261529283e-06,
      "loss": 0.4961,
      "step": 9968
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0762423276901245,
      "learning_rate": 3.15670649509232e-06,
      "loss": 0.4954,
      "step": 9969
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0257775783538818,
      "learning_rate": 3.155752806270033e-06,
      "loss": 0.5933,
      "step": 9970
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.016419768333435,
      "learning_rate": 3.1547991951025795e-06,
      "loss": 0.4811,
      "step": 9971
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0622209310531616,
      "learning_rate": 3.153845661630115e-06,
      "loss": 0.5471,
      "step": 9972
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0601319074630737,
      "learning_rate": 3.152892205892787e-06,
      "loss": 0.5176,
      "step": 9973
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0306676626205444,
      "learning_rate": 3.15193882793074e-06,
      "loss": 0.5243,
      "step": 9974
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0149081945419312,
      "learning_rate": 3.150985527784122e-06,
      "loss": 0.4713,
      "step": 9975
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9692007303237915,
      "learning_rate": 3.1500323054930715e-06,
      "loss": 0.4944,
      "step": 9976
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1371957063674927,
      "learning_rate": 3.149079161097725e-06,
      "loss": 0.5151,
      "step": 9977
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0571235418319702,
      "learning_rate": 3.1481260946382143e-06,
      "loss": 0.476,
      "step": 9978
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.071311116218567,
      "learning_rate": 3.147173106154673e-06,
      "loss": 0.5701,
      "step": 9979
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0023607015609741,
      "learning_rate": 3.146220195687227e-06,
      "loss": 0.5251,
      "step": 9980
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0234299898147583,
      "learning_rate": 3.145267363276e-06,
      "loss": 0.5122,
      "step": 9981
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0225727558135986,
      "learning_rate": 3.1443146089611102e-06,
      "loss": 0.4988,
      "step": 9982
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9956499338150024,
      "learning_rate": 3.143361932782678e-06,
      "loss": 0.5068,
      "step": 9983
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0327447652816772,
      "learning_rate": 3.142409334780817e-06,
      "loss": 0.4816,
      "step": 9984
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9761244058609009,
      "learning_rate": 3.1414568149956366e-06,
      "loss": 0.4235,
      "step": 9985
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0141386985778809,
      "learning_rate": 3.1405043734672436e-06,
      "loss": 0.511,
      "step": 9986
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9915059208869934,
      "learning_rate": 3.1395520102357413e-06,
      "loss": 0.5017,
      "step": 9987
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.140318512916565,
      "learning_rate": 3.1385997253412336e-06,
      "loss": 0.4719,
      "step": 9988
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1061350107192993,
      "learning_rate": 3.137647518823817e-06,
      "loss": 0.5217,
      "step": 9989
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1086782217025757,
      "learning_rate": 3.136695390723583e-06,
      "loss": 0.5753,
      "step": 9990
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1140615940093994,
      "learning_rate": 3.135743341080624e-06,
      "loss": 0.5298,
      "step": 9991
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0715811252593994,
      "learning_rate": 3.1347913699350286e-06,
      "loss": 0.5055,
      "step": 9992
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9809359908103943,
      "learning_rate": 3.1338394773268805e-06,
      "loss": 0.4604,
      "step": 9993
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.063254714012146,
      "learning_rate": 3.132887663296259e-06,
      "loss": 0.5345,
      "step": 9994
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.029570460319519,
      "learning_rate": 3.131935927883242e-06,
      "loss": 0.4974,
      "step": 9995
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1471205949783325,
      "learning_rate": 3.1309842711279066e-06,
      "loss": 0.5022,
      "step": 9996
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.062389612197876,
      "learning_rate": 3.130032693070322e-06,
      "loss": 0.4802,
      "step": 9997
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.061064600944519,
      "learning_rate": 3.129081193750554e-06,
      "loss": 0.4954,
      "step": 9998
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0565736293792725,
      "learning_rate": 3.1281297732086666e-06,
      "loss": 0.5121,
      "step": 9999
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1841918230056763,
      "learning_rate": 3.1271784314847266e-06,
      "loss": 0.5414,
      "step": 10000
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9984389543533325,
      "learning_rate": 3.126227168618786e-06,
      "loss": 0.4735,
      "step": 10001
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0396091938018799,
      "learning_rate": 3.1252759846509013e-06,
      "loss": 0.5111,
      "step": 10002
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9318031072616577,
      "learning_rate": 3.1243248796211234e-06,
      "loss": 0.5002,
      "step": 10003
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0406665802001953,
      "learning_rate": 3.123373853569498e-06,
      "loss": 0.521,
      "step": 10004
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0957379341125488,
      "learning_rate": 3.1224229065360734e-06,
      "loss": 0.5221,
      "step": 10005
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0628998279571533,
      "learning_rate": 3.1214720385608875e-06,
      "loss": 0.5577,
      "step": 10006
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1218875646591187,
      "learning_rate": 3.120521249683981e-06,
      "loss": 0.5252,
      "step": 10007
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1864256858825684,
      "learning_rate": 3.1195705399453833e-06,
      "loss": 0.4871,
      "step": 10008
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0815975666046143,
      "learning_rate": 3.118619909385131e-06,
      "loss": 0.5513,
      "step": 10009
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0236462354660034,
      "learning_rate": 3.117669358043248e-06,
      "loss": 0.5042,
      "step": 10010
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1469367742538452,
      "learning_rate": 3.116718885959762e-06,
      "loss": 0.5255,
      "step": 10011
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0528572797775269,
      "learning_rate": 3.1157684931746902e-06,
      "loss": 0.4901,
      "step": 10012
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0513113737106323,
      "learning_rate": 3.1148181797280543e-06,
      "loss": 0.5343,
      "step": 10013
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9985837936401367,
      "learning_rate": 3.1138679456598654e-06,
      "loss": 0.5021,
      "step": 10014
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0434377193450928,
      "learning_rate": 3.112917791010137e-06,
      "loss": 0.5342,
      "step": 10015
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0435587167739868,
      "learning_rate": 3.111967715818876e-06,
      "loss": 0.5156,
      "step": 10016
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.067971110343933,
      "learning_rate": 3.1110177201260845e-06,
      "loss": 0.5037,
      "step": 10017
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0365383625030518,
      "learning_rate": 3.1100678039717665e-06,
      "loss": 0.4851,
      "step": 10018
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.078518033027649,
      "learning_rate": 3.1091179673959194e-06,
      "loss": 0.5094,
      "step": 10019
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.158864974975586,
      "learning_rate": 3.108168210438536e-06,
      "loss": 0.528,
      "step": 10020
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0493173599243164,
      "learning_rate": 3.1072185331396083e-06,
      "loss": 0.5069,
      "step": 10021
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.00165593624115,
      "learning_rate": 3.106268935539123e-06,
      "loss": 0.5226,
      "step": 10022
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1848684549331665,
      "learning_rate": 3.1053194176770662e-06,
      "loss": 0.5295,
      "step": 10023
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9629892706871033,
      "learning_rate": 3.1043699795934172e-06,
      "loss": 0.4802,
      "step": 10024
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0899156332015991,
      "learning_rate": 3.1034206213281536e-06,
      "loss": 0.5688,
      "step": 10025
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0079987049102783,
      "learning_rate": 3.10247134292125e-06,
      "loss": 0.4579,
      "step": 10026
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0822067260742188,
      "learning_rate": 3.1015221444126776e-06,
      "loss": 0.5051,
      "step": 10027
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9707460999488831,
      "learning_rate": 3.1005730258424025e-06,
      "loss": 0.5169,
      "step": 10028
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9045226573944092,
      "learning_rate": 3.099623987250391e-06,
      "loss": 0.4893,
      "step": 10029
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0418764352798462,
      "learning_rate": 3.098675028676601e-06,
      "loss": 0.5289,
      "step": 10030
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0181958675384521,
      "learning_rate": 3.0977261501609924e-06,
      "loss": 0.512,
      "step": 10031
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1507220268249512,
      "learning_rate": 3.0967773517435173e-06,
      "loss": 0.5234,
      "step": 10032
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0432597398757935,
      "learning_rate": 3.0958286334641284e-06,
      "loss": 0.4601,
      "step": 10033
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1117364168167114,
      "learning_rate": 3.0948799953627696e-06,
      "loss": 0.5066,
      "step": 10034
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0182552337646484,
      "learning_rate": 3.093931437479388e-06,
      "loss": 0.4733,
      "step": 10035
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9871261119842529,
      "learning_rate": 3.092982959853923e-06,
      "loss": 0.5096,
      "step": 10036
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.086436152458191,
      "learning_rate": 3.092034562526312e-06,
      "loss": 0.5515,
      "step": 10037
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0634336471557617,
      "learning_rate": 3.0910862455364864e-06,
      "loss": 0.4862,
      "step": 10038
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1165075302124023,
      "learning_rate": 3.09013800892438e-06,
      "loss": 0.5504,
      "step": 10039
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9831963181495667,
      "learning_rate": 3.0891898527299167e-06,
      "loss": 0.477,
      "step": 10040
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.017632246017456,
      "learning_rate": 3.088241776993024e-06,
      "loss": 0.527,
      "step": 10041
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0667827129364014,
      "learning_rate": 3.0872937817536165e-06,
      "loss": 0.5569,
      "step": 10042
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0934066772460938,
      "learning_rate": 3.0863458670516157e-06,
      "loss": 0.5066,
      "step": 10043
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0351758003234863,
      "learning_rate": 3.085398032926933e-06,
      "loss": 0.5125,
      "step": 10044
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0830355882644653,
      "learning_rate": 3.0844502794194795e-06,
      "loss": 0.5458,
      "step": 10045
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0485659837722778,
      "learning_rate": 3.083502606569159e-06,
      "loss": 0.4988,
      "step": 10046
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9413774609565735,
      "learning_rate": 3.0825550144158788e-06,
      "loss": 0.4749,
      "step": 10047
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0139778852462769,
      "learning_rate": 3.081607502999536e-06,
      "loss": 0.4913,
      "step": 10048
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0455811023712158,
      "learning_rate": 3.0806600723600275e-06,
      "loss": 0.511,
      "step": 10049
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.008416771888733,
      "learning_rate": 3.0797127225372477e-06,
      "loss": 0.5121,
      "step": 10050
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.03008234500885,
      "learning_rate": 3.078765453571082e-06,
      "loss": 0.47,
      "step": 10051
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.091876745223999,
      "learning_rate": 3.077818265501421e-06,
      "loss": 0.5114,
      "step": 10052
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0145922899246216,
      "learning_rate": 3.0768711583681475e-06,
      "loss": 0.5052,
      "step": 10053
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9602000117301941,
      "learning_rate": 3.075924132211139e-06,
      "loss": 0.4515,
      "step": 10054
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.020117163658142,
      "learning_rate": 3.07497718707027e-06,
      "loss": 0.4649,
      "step": 10055
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1142258644104004,
      "learning_rate": 3.074030322985416e-06,
      "loss": 0.5001,
      "step": 10056
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.013404130935669,
      "learning_rate": 3.073083539996446e-06,
      "loss": 0.5161,
      "step": 10057
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9824324250221252,
      "learning_rate": 3.072136838143225e-06,
      "loss": 0.4878,
      "step": 10058
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9934601187705994,
      "learning_rate": 3.0711902174656126e-06,
      "loss": 0.4888,
      "step": 10059
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.08644437789917,
      "learning_rate": 3.070243678003472e-06,
      "loss": 0.5309,
      "step": 10060
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.105520248413086,
      "learning_rate": 3.069297219796658e-06,
      "loss": 0.5485,
      "step": 10061
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.015161156654358,
      "learning_rate": 3.068350842885022e-06,
      "loss": 0.498,
      "step": 10062
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0500092506408691,
      "learning_rate": 3.0674045473084103e-06,
      "loss": 0.5216,
      "step": 10063
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0675020217895508,
      "learning_rate": 3.0664583331066695e-06,
      "loss": 0.4649,
      "step": 10064
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0304772853851318,
      "learning_rate": 3.0655122003196443e-06,
      "loss": 0.4777,
      "step": 10065
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0809754133224487,
      "learning_rate": 3.06456614898717e-06,
      "loss": 0.5531,
      "step": 10066
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0226454734802246,
      "learning_rate": 3.0636201791490823e-06,
      "loss": 0.5415,
      "step": 10067
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0683984756469727,
      "learning_rate": 3.062674290845211e-06,
      "loss": 0.4793,
      "step": 10068
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9812599420547485,
      "learning_rate": 3.061728484115388e-06,
      "loss": 0.4737,
      "step": 10069
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9435461759567261,
      "learning_rate": 3.0607827589994353e-06,
      "loss": 0.4736,
      "step": 10070
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0279617309570312,
      "learning_rate": 3.0598371155371747e-06,
      "loss": 0.4617,
      "step": 10071
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0339475870132446,
      "learning_rate": 3.058891553768422e-06,
      "loss": 0.4999,
      "step": 10072
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1052274703979492,
      "learning_rate": 3.0579460737329958e-06,
      "loss": 0.5335,
      "step": 10073
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.117714285850525,
      "learning_rate": 3.0570006754707044e-06,
      "loss": 0.5456,
      "step": 10074
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.066931128501892,
      "learning_rate": 3.056055359021354e-06,
      "loss": 0.5208,
      "step": 10075
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9597610831260681,
      "learning_rate": 3.0551101244247494e-06,
      "loss": 0.4912,
      "step": 10076
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0383501052856445,
      "learning_rate": 3.0541649717206933e-06,
      "loss": 0.4777,
      "step": 10077
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.014605164527893,
      "learning_rate": 3.0532199009489814e-06,
      "loss": 0.5084,
      "step": 10078
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0429009199142456,
      "learning_rate": 3.052274912149406e-06,
      "loss": 0.5059,
      "step": 10079
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.058207631111145,
      "learning_rate": 3.0513300053617595e-06,
      "loss": 0.4925,
      "step": 10080
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9927064776420593,
      "learning_rate": 3.0503851806258257e-06,
      "loss": 0.4984,
      "step": 10081
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9581083059310913,
      "learning_rate": 3.0494404379813914e-06,
      "loss": 0.5451,
      "step": 10082
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0245808362960815,
      "learning_rate": 3.048495777468234e-06,
      "loss": 0.5594,
      "step": 10083
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1457056999206543,
      "learning_rate": 3.047551199126131e-06,
      "loss": 0.5133,
      "step": 10084
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0019303560256958,
      "learning_rate": 3.046606702994854e-06,
      "loss": 0.5592,
      "step": 10085
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0403997898101807,
      "learning_rate": 3.0456622891141748e-06,
      "loss": 0.4771,
      "step": 10086
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0224168300628662,
      "learning_rate": 3.0447179575238565e-06,
      "loss": 0.5306,
      "step": 10087
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1169382333755493,
      "learning_rate": 3.0437737082636647e-06,
      "loss": 0.5728,
      "step": 10088
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.010927677154541,
      "learning_rate": 3.0428295413733546e-06,
      "loss": 0.4344,
      "step": 10089
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.061489224433899,
      "learning_rate": 3.0418854568926866e-06,
      "loss": 0.5116,
      "step": 10090
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0159995555877686,
      "learning_rate": 3.0409414548614086e-06,
      "loss": 0.4819,
      "step": 10091
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.984320342540741,
      "learning_rate": 3.039997535319272e-06,
      "loss": 0.5023,
      "step": 10092
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.054763674736023,
      "learning_rate": 3.039053698306019e-06,
      "loss": 0.4756,
      "step": 10093
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0427004098892212,
      "learning_rate": 3.0381099438613948e-06,
      "loss": 0.4761,
      "step": 10094
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.049312949180603,
      "learning_rate": 3.037166272025135e-06,
      "loss": 0.4976,
      "step": 10095
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0668352842330933,
      "learning_rate": 3.0362226828369767e-06,
      "loss": 0.5331,
      "step": 10096
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.065746784210205,
      "learning_rate": 3.0352791763366484e-06,
      "loss": 0.5142,
      "step": 10097
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9855623841285706,
      "learning_rate": 3.0343357525638787e-06,
      "loss": 0.4476,
      "step": 10098
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0520395040512085,
      "learning_rate": 3.0333924115583935e-06,
      "loss": 0.5118,
      "step": 10099
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0070985555648804,
      "learning_rate": 3.032449153359913e-06,
      "loss": 0.4616,
      "step": 10100
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.040044903755188,
      "learning_rate": 3.031505978008153e-06,
      "loss": 0.517,
      "step": 10101
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0617992877960205,
      "learning_rate": 3.030562885542827e-06,
      "loss": 0.4896,
      "step": 10102
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.01646888256073,
      "learning_rate": 3.0296198760036493e-06,
      "loss": 0.4899,
      "step": 10103
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.137224793434143,
      "learning_rate": 3.0286769494303237e-06,
      "loss": 0.5234,
      "step": 10104
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.110205054283142,
      "learning_rate": 3.0277341058625537e-06,
      "loss": 0.5328,
      "step": 10105
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.036597490310669,
      "learning_rate": 3.026791345340038e-06,
      "loss": 0.5172,
      "step": 10106
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9912294149398804,
      "learning_rate": 3.0258486679024767e-06,
      "loss": 0.4618,
      "step": 10107
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1212050914764404,
      "learning_rate": 3.0249060735895603e-06,
      "loss": 0.5229,
      "step": 10108
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0347286462783813,
      "learning_rate": 3.0239635624409767e-06,
      "loss": 0.4913,
      "step": 10109
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0135654211044312,
      "learning_rate": 3.0230211344964154e-06,
      "loss": 0.4797,
      "step": 10110
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1035380363464355,
      "learning_rate": 3.0220787897955544e-06,
      "loss": 0.5162,
      "step": 10111
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0869108438491821,
      "learning_rate": 3.021136528378077e-06,
      "loss": 0.5043,
      "step": 10112
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.066138744354248,
      "learning_rate": 3.020194350283655e-06,
      "loss": 0.5352,
      "step": 10113
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.148913025856018,
      "learning_rate": 3.019252255551963e-06,
      "loss": 0.5744,
      "step": 10114
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9562913179397583,
      "learning_rate": 3.0183102442226653e-06,
      "loss": 0.4837,
      "step": 10115
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9955688118934631,
      "learning_rate": 3.017368316335432e-06,
      "loss": 0.5081,
      "step": 10116
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.027421236038208,
      "learning_rate": 3.0164264719299207e-06,
      "loss": 0.5122,
      "step": 10117
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0766228437423706,
      "learning_rate": 3.0154847110457918e-06,
      "loss": 0.4476,
      "step": 10118
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0523964166641235,
      "learning_rate": 3.0145430337226955e-06,
      "loss": 0.5192,
      "step": 10119
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1346592903137207,
      "learning_rate": 3.013601440000288e-06,
      "loss": 0.5259,
      "step": 10120
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.127166509628296,
      "learning_rate": 3.0126599299182114e-06,
      "loss": 0.5195,
      "step": 10121
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1691970825195312,
      "learning_rate": 3.0117185035161135e-06,
      "loss": 0.522,
      "step": 10122
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0456876754760742,
      "learning_rate": 3.01077716083363e-06,
      "loss": 0.5068,
      "step": 10123
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0364118814468384,
      "learning_rate": 3.009835901910403e-06,
      "loss": 0.4758,
      "step": 10124
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0276139974594116,
      "learning_rate": 3.008894726786062e-06,
      "loss": 0.4957,
      "step": 10125
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1089617013931274,
      "learning_rate": 3.007953635500238e-06,
      "loss": 0.4891,
      "step": 10126
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0690889358520508,
      "learning_rate": 3.0070126280925564e-06,
      "loss": 0.5857,
      "step": 10127
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0123674869537354,
      "learning_rate": 3.0060717046026387e-06,
      "loss": 0.531,
      "step": 10128
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0811229944229126,
      "learning_rate": 3.0051308650701054e-06,
      "loss": 0.5572,
      "step": 10129
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0825634002685547,
      "learning_rate": 3.004190109534573e-06,
      "loss": 0.5139,
      "step": 10130
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.081278920173645,
      "learning_rate": 3.0032494380356523e-06,
      "loss": 0.508,
      "step": 10131
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9947942495346069,
      "learning_rate": 3.002308850612949e-06,
      "loss": 0.5034,
      "step": 10132
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0024405717849731,
      "learning_rate": 3.001368347306073e-06,
      "loss": 0.4934,
      "step": 10133
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0109182596206665,
      "learning_rate": 3.0004279281546235e-06,
      "loss": 0.5127,
      "step": 10134
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0295737981796265,
      "learning_rate": 2.999487593198197e-06,
      "loss": 0.4838,
      "step": 10135
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0023385286331177,
      "learning_rate": 2.9985473424763876e-06,
      "loss": 0.4832,
      "step": 10136
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0790435075759888,
      "learning_rate": 2.9976071760287874e-06,
      "loss": 0.5331,
      "step": 10137
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.053161859512329,
      "learning_rate": 2.9966670938949847e-06,
      "loss": 0.4834,
      "step": 10138
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.980963408946991,
      "learning_rate": 2.995727096114561e-06,
      "loss": 0.5004,
      "step": 10139
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0529303550720215,
      "learning_rate": 2.9947871827270956e-06,
      "loss": 0.5491,
      "step": 10140
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0220396518707275,
      "learning_rate": 2.993847353772168e-06,
      "loss": 0.5115,
      "step": 10141
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9761862754821777,
      "learning_rate": 2.9929076092893496e-06,
      "loss": 0.4773,
      "step": 10142
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0252829790115356,
      "learning_rate": 2.991967949318209e-06,
      "loss": 0.5256,
      "step": 10143
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.00223970413208,
      "learning_rate": 2.9910283738983125e-06,
      "loss": 0.4909,
      "step": 10144
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0818865299224854,
      "learning_rate": 2.9900888830692208e-06,
      "loss": 0.5353,
      "step": 10145
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0524858236312866,
      "learning_rate": 2.9891494768704964e-06,
      "loss": 0.5401,
      "step": 10146
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0927537679672241,
      "learning_rate": 2.9882101553416932e-06,
      "loss": 0.4795,
      "step": 10147
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0539554357528687,
      "learning_rate": 2.9872709185223596e-06,
      "loss": 0.5344,
      "step": 10148
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.976118266582489,
      "learning_rate": 2.9863317664520453e-06,
      "loss": 0.4826,
      "step": 10149
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1192395687103271,
      "learning_rate": 2.9853926991702974e-06,
      "loss": 0.5249,
      "step": 10150
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0809307098388672,
      "learning_rate": 2.984453716716655e-06,
      "loss": 0.5024,
      "step": 10151
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9954077005386353,
      "learning_rate": 2.9835148191306535e-06,
      "loss": 0.479,
      "step": 10152
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.067521572113037,
      "learning_rate": 2.9825760064518273e-06,
      "loss": 0.5265,
      "step": 10153
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8902660012245178,
      "learning_rate": 2.981637278719709e-06,
      "loss": 0.4432,
      "step": 10154
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0645134449005127,
      "learning_rate": 2.9806986359738244e-06,
      "loss": 0.5195,
      "step": 10155
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9383110404014587,
      "learning_rate": 2.979760078253694e-06,
      "loss": 0.4828,
      "step": 10156
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1661232709884644,
      "learning_rate": 2.9788216055988397e-06,
      "loss": 0.5346,
      "step": 10157
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0412225723266602,
      "learning_rate": 2.977883218048775e-06,
      "loss": 0.5077,
      "step": 10158
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0863960981369019,
      "learning_rate": 2.9769449156430147e-06,
      "loss": 0.5489,
      "step": 10159
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0391831398010254,
      "learning_rate": 2.9760066984210655e-06,
      "loss": 0.5354,
      "step": 10160
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9662297964096069,
      "learning_rate": 2.975068566422434e-06,
      "loss": 0.501,
      "step": 10161
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.076661229133606,
      "learning_rate": 2.97413051968662e-06,
      "loss": 0.5495,
      "step": 10162
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1024425029754639,
      "learning_rate": 2.9731925582531227e-06,
      "loss": 0.5142,
      "step": 10163
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0270276069641113,
      "learning_rate": 2.9722546821614373e-06,
      "loss": 0.504,
      "step": 10164
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0266375541687012,
      "learning_rate": 2.9713168914510533e-06,
      "loss": 0.472,
      "step": 10165
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0487178564071655,
      "learning_rate": 2.970379186161455e-06,
      "loss": 0.5039,
      "step": 10166
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0786352157592773,
      "learning_rate": 2.96944156633213e-06,
      "loss": 0.555,
      "step": 10167
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0177651643753052,
      "learning_rate": 2.9685040320025583e-06,
      "loss": 0.4892,
      "step": 10168
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.00448739528656,
      "learning_rate": 2.9675665832122146e-06,
      "loss": 0.5229,
      "step": 10169
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0026636123657227,
      "learning_rate": 2.966629220000569e-06,
      "loss": 0.4948,
      "step": 10170
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0992698669433594,
      "learning_rate": 2.965691942407095e-06,
      "loss": 0.4946,
      "step": 10171
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9647083282470703,
      "learning_rate": 2.9647547504712577e-06,
      "loss": 0.535,
      "step": 10172
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.021139144897461,
      "learning_rate": 2.9638176442325173e-06,
      "loss": 0.5139,
      "step": 10173
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0401581525802612,
      "learning_rate": 2.962880623730332e-06,
      "loss": 0.5098,
      "step": 10174
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1423335075378418,
      "learning_rate": 2.9619436890041555e-06,
      "loss": 0.5268,
      "step": 10175
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.057899832725525,
      "learning_rate": 2.961006840093442e-06,
      "loss": 0.4863,
      "step": 10176
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9457885026931763,
      "learning_rate": 2.9600700770376384e-06,
      "loss": 0.4911,
      "step": 10177
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0614778995513916,
      "learning_rate": 2.959133399876186e-06,
      "loss": 0.4978,
      "step": 10178
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0491783618927002,
      "learning_rate": 2.958196808648525e-06,
      "loss": 0.5372,
      "step": 10179
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0236880779266357,
      "learning_rate": 2.957260303394096e-06,
      "loss": 0.5064,
      "step": 10180
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0263805389404297,
      "learning_rate": 2.9563238841523293e-06,
      "loss": 0.519,
      "step": 10181
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.969336986541748,
      "learning_rate": 2.955387550962654e-06,
      "loss": 0.4496,
      "step": 10182
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.028093934059143,
      "learning_rate": 2.954451303864494e-06,
      "loss": 0.5405,
      "step": 10183
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0766602754592896,
      "learning_rate": 2.9535151428972762e-06,
      "loss": 0.5221,
      "step": 10184
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0071594715118408,
      "learning_rate": 2.9525790681004172e-06,
      "loss": 0.4856,
      "step": 10185
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9100344181060791,
      "learning_rate": 2.9516430795133294e-06,
      "loss": 0.4717,
      "step": 10186
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.075358510017395,
      "learning_rate": 2.950707177175427e-06,
      "loss": 0.4949,
      "step": 10187
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9946542978286743,
      "learning_rate": 2.9497713611261146e-06,
      "loss": 0.4821,
      "step": 10188
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1402562856674194,
      "learning_rate": 2.9488356314047994e-06,
      "loss": 0.5511,
      "step": 10189
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.995521605014801,
      "learning_rate": 2.94789998805088e-06,
      "loss": 0.5295,
      "step": 10190
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1219182014465332,
      "learning_rate": 2.9469644311037545e-06,
      "loss": 0.5539,
      "step": 10191
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0393383502960205,
      "learning_rate": 2.946028960602812e-06,
      "loss": 0.5071,
      "step": 10192
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0298091173171997,
      "learning_rate": 2.9450935765874474e-06,
      "loss": 0.5163,
      "step": 10193
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0860475301742554,
      "learning_rate": 2.9441582790970425e-06,
      "loss": 0.5379,
      "step": 10194
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0946840047836304,
      "learning_rate": 2.9432230681709815e-06,
      "loss": 0.5529,
      "step": 10195
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9496505856513977,
      "learning_rate": 2.942287943848641e-06,
      "loss": 0.4892,
      "step": 10196
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1770808696746826,
      "learning_rate": 2.941352906169398e-06,
      "loss": 0.5534,
      "step": 10197
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.992904782295227,
      "learning_rate": 2.9404179551726214e-06,
      "loss": 0.4744,
      "step": 10198
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0817198753356934,
      "learning_rate": 2.939483090897681e-06,
      "loss": 0.54,
      "step": 10199
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0625017881393433,
      "learning_rate": 2.9385483133839386e-06,
      "loss": 0.4746,
      "step": 10200
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2074081897735596,
      "learning_rate": 2.937613622670756e-06,
      "loss": 0.579,
      "step": 10201
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.994379997253418,
      "learning_rate": 2.9366790187974897e-06,
      "loss": 0.4887,
      "step": 10202
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9682003855705261,
      "learning_rate": 2.9357445018034926e-06,
      "loss": 0.516,
      "step": 10203
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0344123840332031,
      "learning_rate": 2.934810071728114e-06,
      "loss": 0.514,
      "step": 10204
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.05597984790802,
      "learning_rate": 2.9338757286106955e-06,
      "loss": 0.5176,
      "step": 10205
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9796289801597595,
      "learning_rate": 2.9329414724905845e-06,
      "loss": 0.514,
      "step": 10206
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9932934045791626,
      "learning_rate": 2.9320073034071187e-06,
      "loss": 0.4862,
      "step": 10207
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.98027104139328,
      "learning_rate": 2.9310732213996305e-06,
      "loss": 0.5024,
      "step": 10208
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9642611742019653,
      "learning_rate": 2.9301392265074506e-06,
      "loss": 0.5061,
      "step": 10209
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0452041625976562,
      "learning_rate": 2.9292053187699075e-06,
      "loss": 0.4624,
      "step": 10210
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0322476625442505,
      "learning_rate": 2.9282714982263265e-06,
      "loss": 0.4644,
      "step": 10211
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.046685814857483,
      "learning_rate": 2.927337764916025e-06,
      "loss": 0.5169,
      "step": 10212
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.061316728591919,
      "learning_rate": 2.926404118878319e-06,
      "loss": 0.5334,
      "step": 10213
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.000592827796936,
      "learning_rate": 2.925470560152522e-06,
      "loss": 0.5085,
      "step": 10214
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9926203489303589,
      "learning_rate": 2.924537088777944e-06,
      "loss": 0.5286,
      "step": 10215
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.045668601989746,
      "learning_rate": 2.9236037047938894e-06,
      "loss": 0.4923,
      "step": 10216
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0347856283187866,
      "learning_rate": 2.922670408239657e-06,
      "loss": 0.5097,
      "step": 10217
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0084527730941772,
      "learning_rate": 2.921737199154549e-06,
      "loss": 0.5545,
      "step": 10218
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0063661336898804,
      "learning_rate": 2.920804077577859e-06,
      "loss": 0.4822,
      "step": 10219
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9825767874717712,
      "learning_rate": 2.919871043548875e-06,
      "loss": 0.4867,
      "step": 10220
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.020814061164856,
      "learning_rate": 2.9189380971068864e-06,
      "loss": 0.4682,
      "step": 10221
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1582943201065063,
      "learning_rate": 2.918005238291172e-06,
      "loss": 0.5218,
      "step": 10222
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0233937501907349,
      "learning_rate": 2.9170724671410155e-06,
      "loss": 0.5315,
      "step": 10223
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9818440675735474,
      "learning_rate": 2.916139783695694e-06,
      "loss": 0.4892,
      "step": 10224
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0906521081924438,
      "learning_rate": 2.9152071879944743e-06,
      "loss": 0.5153,
      "step": 10225
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1499855518341064,
      "learning_rate": 2.914274680076628e-06,
      "loss": 0.5607,
      "step": 10226
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0200022459030151,
      "learning_rate": 2.913342259981419e-06,
      "loss": 0.5322,
      "step": 10227
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0477135181427002,
      "learning_rate": 2.9124099277481088e-06,
      "loss": 0.4779,
      "step": 10228
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9612158536911011,
      "learning_rate": 2.9114776834159563e-06,
      "loss": 0.4984,
      "step": 10229
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0411784648895264,
      "learning_rate": 2.910545527024209e-06,
      "loss": 0.5143,
      "step": 10230
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0339840650558472,
      "learning_rate": 2.9096134586121227e-06,
      "loss": 0.5487,
      "step": 10231
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.057961106300354,
      "learning_rate": 2.908681478218944e-06,
      "loss": 0.4985,
      "step": 10232
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0351730585098267,
      "learning_rate": 2.907749585883911e-06,
      "loss": 0.5014,
      "step": 10233
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.06086003780365,
      "learning_rate": 2.906817781646264e-06,
      "loss": 0.5022,
      "step": 10234
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0551403760910034,
      "learning_rate": 2.905886065545239e-06,
      "loss": 0.5281,
      "step": 10235
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.962441086769104,
      "learning_rate": 2.9049544376200674e-06,
      "loss": 0.5143,
      "step": 10236
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1584631204605103,
      "learning_rate": 2.9040228979099777e-06,
      "loss": 0.5067,
      "step": 10237
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9825107455253601,
      "learning_rate": 2.9030914464541904e-06,
      "loss": 0.4701,
      "step": 10238
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0211668014526367,
      "learning_rate": 2.902160083291926e-06,
      "loss": 0.4898,
      "step": 10239
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1130526065826416,
      "learning_rate": 2.9012288084624065e-06,
      "loss": 0.474,
      "step": 10240
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.003676176071167,
      "learning_rate": 2.9002976220048383e-06,
      "loss": 0.4768,
      "step": 10241
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0888468027114868,
      "learning_rate": 2.899366523958434e-06,
      "loss": 0.546,
      "step": 10242
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0272295475006104,
      "learning_rate": 2.898435514362397e-06,
      "loss": 0.5243,
      "step": 10243
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9819967150688171,
      "learning_rate": 2.89750459325593e-06,
      "loss": 0.484,
      "step": 10244
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.091143012046814,
      "learning_rate": 2.896573760678232e-06,
      "loss": 0.5206,
      "step": 10245
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0350513458251953,
      "learning_rate": 2.8956430166684945e-06,
      "loss": 0.4974,
      "step": 10246
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0358695983886719,
      "learning_rate": 2.8947123612659068e-06,
      "loss": 0.5276,
      "step": 10247
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.046087622642517,
      "learning_rate": 2.8937817945096614e-06,
      "loss": 0.4793,
      "step": 10248
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1108758449554443,
      "learning_rate": 2.8928513164389353e-06,
      "loss": 0.5576,
      "step": 10249
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.047646403312683,
      "learning_rate": 2.8919209270929106e-06,
      "loss": 0.5497,
      "step": 10250
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0258592367172241,
      "learning_rate": 2.8909906265107647e-06,
      "loss": 0.487,
      "step": 10251
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1011090278625488,
      "learning_rate": 2.890060414731662e-06,
      "loss": 0.4945,
      "step": 10252
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.131653070449829,
      "learning_rate": 2.8891302917947794e-06,
      "loss": 0.5597,
      "step": 10253
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1027075052261353,
      "learning_rate": 2.8882002577392752e-06,
      "loss": 0.5561,
      "step": 10254
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1370142698287964,
      "learning_rate": 2.8872703126043116e-06,
      "loss": 0.6032,
      "step": 10255
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.10757315158844,
      "learning_rate": 2.8863404564290455e-06,
      "loss": 0.4975,
      "step": 10256
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0993229150772095,
      "learning_rate": 2.88541068925263e-06,
      "loss": 0.5404,
      "step": 10257
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9537004828453064,
      "learning_rate": 2.8844810111142143e-06,
      "loss": 0.4863,
      "step": 10258
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.100672960281372,
      "learning_rate": 2.883551422052946e-06,
      "loss": 0.4964,
      "step": 10259
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0630990266799927,
      "learning_rate": 2.8826219221079597e-06,
      "loss": 0.5286,
      "step": 10260
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.017022728919983,
      "learning_rate": 2.8816925113184034e-06,
      "loss": 0.4967,
      "step": 10261
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0680783987045288,
      "learning_rate": 2.8807631897234045e-06,
      "loss": 0.5264,
      "step": 10262
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1391302347183228,
      "learning_rate": 2.8798339573620953e-06,
      "loss": 0.5474,
      "step": 10263
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0025023221969604,
      "learning_rate": 2.8789048142736026e-06,
      "loss": 0.4627,
      "step": 10264
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1172410249710083,
      "learning_rate": 2.8779757604970495e-06,
      "loss": 0.5047,
      "step": 10265
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9674686789512634,
      "learning_rate": 2.877046796071554e-06,
      "loss": 0.5257,
      "step": 10266
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0824980735778809,
      "learning_rate": 2.8761179210362365e-06,
      "loss": 0.5243,
      "step": 10267
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1241073608398438,
      "learning_rate": 2.8751891354302018e-06,
      "loss": 0.5312,
      "step": 10268
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9392661452293396,
      "learning_rate": 2.8742604392925587e-06,
      "loss": 0.4949,
      "step": 10269
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0615390539169312,
      "learning_rate": 2.8733318326624182e-06,
      "loss": 0.5227,
      "step": 10270
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0059072971343994,
      "learning_rate": 2.8724033155788743e-06,
      "loss": 0.5227,
      "step": 10271
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0599112510681152,
      "learning_rate": 2.871474888081025e-06,
      "loss": 0.543,
      "step": 10272
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0321950912475586,
      "learning_rate": 2.870546550207964e-06,
      "loss": 0.4835,
      "step": 10273
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0996196269989014,
      "learning_rate": 2.8696183019987796e-06,
      "loss": 0.4924,
      "step": 10274
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0861026048660278,
      "learning_rate": 2.868690143492559e-06,
      "loss": 0.5298,
      "step": 10275
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1031099557876587,
      "learning_rate": 2.8677620747283807e-06,
      "loss": 0.5125,
      "step": 10276
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0979763269424438,
      "learning_rate": 2.8668340957453224e-06,
      "loss": 0.4891,
      "step": 10277
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9893242716789246,
      "learning_rate": 2.865906206582463e-06,
      "loss": 0.5146,
      "step": 10278
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9797508716583252,
      "learning_rate": 2.8649784072788668e-06,
      "loss": 0.4693,
      "step": 10279
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.061628818511963,
      "learning_rate": 2.8640506978736027e-06,
      "loss": 0.568,
      "step": 10280
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.185620665550232,
      "learning_rate": 2.8631230784057362e-06,
      "loss": 0.5479,
      "step": 10281
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0274667739868164,
      "learning_rate": 2.862195548914318e-06,
      "loss": 0.5336,
      "step": 10282
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1475064754486084,
      "learning_rate": 2.8612681094384135e-06,
      "loss": 0.5158,
      "step": 10283
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0287840366363525,
      "learning_rate": 2.8603407600170664e-06,
      "loss": 0.5117,
      "step": 10284
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9745176434516907,
      "learning_rate": 2.8594135006893264e-06,
      "loss": 0.5212,
      "step": 10285
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0082731246948242,
      "learning_rate": 2.858486331494238e-06,
      "loss": 0.5245,
      "step": 10286
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9963515400886536,
      "learning_rate": 2.8575592524708397e-06,
      "loss": 0.5252,
      "step": 10287
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9823704957962036,
      "learning_rate": 2.856632263658169e-06,
      "loss": 0.4971,
      "step": 10288
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0706892013549805,
      "learning_rate": 2.855705365095258e-06,
      "loss": 0.5358,
      "step": 10289
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9489477872848511,
      "learning_rate": 2.854778556821132e-06,
      "loss": 0.4582,
      "step": 10290
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0349112749099731,
      "learning_rate": 2.8538518388748214e-06,
      "loss": 0.5261,
      "step": 10291
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.108465313911438,
      "learning_rate": 2.8529252112953434e-06,
      "loss": 0.5271,
      "step": 10292
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.125195860862732,
      "learning_rate": 2.8519986741217144e-06,
      "loss": 0.5316,
      "step": 10293
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9881192445755005,
      "learning_rate": 2.8510722273929486e-06,
      "loss": 0.5166,
      "step": 10294
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0663776397705078,
      "learning_rate": 2.8501458711480564e-06,
      "loss": 0.5127,
      "step": 10295
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0565035343170166,
      "learning_rate": 2.8492196054260424e-06,
      "loss": 0.5389,
      "step": 10296
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0431386232376099,
      "learning_rate": 2.848293430265911e-06,
      "loss": 0.5097,
      "step": 10297
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9941859245300293,
      "learning_rate": 2.8473673457066564e-06,
      "loss": 0.479,
      "step": 10298
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9996705055236816,
      "learning_rate": 2.8464413517872737e-06,
      "loss": 0.492,
      "step": 10299
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.066678524017334,
      "learning_rate": 2.845515448546754e-06,
      "loss": 0.5115,
      "step": 10300
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9481530785560608,
      "learning_rate": 2.8445896360240845e-06,
      "loss": 0.4736,
      "step": 10301
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.095447301864624,
      "learning_rate": 2.843663914258249e-06,
      "loss": 0.5133,
      "step": 10302
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9487720131874084,
      "learning_rate": 2.8427382832882207e-06,
      "loss": 0.494,
      "step": 10303
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.069291353225708,
      "learning_rate": 2.8418127431529807e-06,
      "loss": 0.5182,
      "step": 10304
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0383696556091309,
      "learning_rate": 2.8408872938915e-06,
      "loss": 0.4865,
      "step": 10305
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0746654272079468,
      "learning_rate": 2.8399619355427427e-06,
      "loss": 0.5395,
      "step": 10306
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0153661966323853,
      "learning_rate": 2.839036668145674e-06,
      "loss": 0.5271,
      "step": 10307
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0152404308319092,
      "learning_rate": 2.8381114917392538e-06,
      "loss": 0.5079,
      "step": 10308
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0082815885543823,
      "learning_rate": 2.8371864063624375e-06,
      "loss": 0.4782,
      "step": 10309
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0504685640335083,
      "learning_rate": 2.836261412054181e-06,
      "loss": 0.4719,
      "step": 10310
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9775835871696472,
      "learning_rate": 2.8353365088534247e-06,
      "loss": 0.5359,
      "step": 10311
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0367127656936646,
      "learning_rate": 2.8344116967991197e-06,
      "loss": 0.4911,
      "step": 10312
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0500134229660034,
      "learning_rate": 2.8334869759302064e-06,
      "loss": 0.5102,
      "step": 10313
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0859054327011108,
      "learning_rate": 2.8325623462856176e-06,
      "loss": 0.518,
      "step": 10314
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0743653774261475,
      "learning_rate": 2.8316378079042887e-06,
      "loss": 0.5542,
      "step": 10315
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0794223546981812,
      "learning_rate": 2.8307133608251486e-06,
      "loss": 0.4962,
      "step": 10316
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0531331300735474,
      "learning_rate": 2.8297890050871222e-06,
      "loss": 0.5037,
      "step": 10317
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.021409273147583,
      "learning_rate": 2.8288647407291337e-06,
      "loss": 0.5426,
      "step": 10318
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0144912004470825,
      "learning_rate": 2.827940567790096e-06,
      "loss": 0.5216,
      "step": 10319
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0451698303222656,
      "learning_rate": 2.8270164863089227e-06,
      "loss": 0.53,
      "step": 10320
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0527116060256958,
      "learning_rate": 2.82609249632453e-06,
      "loss": 0.5234,
      "step": 10321
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.05509352684021,
      "learning_rate": 2.825168597875818e-06,
      "loss": 0.5229,
      "step": 10322
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0506893396377563,
      "learning_rate": 2.82424479100169e-06,
      "loss": 0.4694,
      "step": 10323
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0868808031082153,
      "learning_rate": 2.8233210757410454e-06,
      "loss": 0.5533,
      "step": 10324
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.011303186416626,
      "learning_rate": 2.8223974521327787e-06,
      "loss": 0.5178,
      "step": 10325
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0738871097564697,
      "learning_rate": 2.8214739202157794e-06,
      "loss": 0.5024,
      "step": 10326
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.006366491317749,
      "learning_rate": 2.820550480028937e-06,
      "loss": 0.5213,
      "step": 10327
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1595299243927002,
      "learning_rate": 2.81962713161113e-06,
      "loss": 0.5485,
      "step": 10328
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.003004789352417,
      "learning_rate": 2.8187038750012396e-06,
      "loss": 0.5179,
      "step": 10329
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9998498558998108,
      "learning_rate": 2.8177807102381404e-06,
      "loss": 0.4903,
      "step": 10330
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9879375696182251,
      "learning_rate": 2.816857637360705e-06,
      "loss": 0.4948,
      "step": 10331
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.04654061794281,
      "learning_rate": 2.8159346564078006e-06,
      "loss": 0.5101,
      "step": 10332
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0643385648727417,
      "learning_rate": 2.815011767418287e-06,
      "loss": 0.5383,
      "step": 10333
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0637271404266357,
      "learning_rate": 2.8140889704310287e-06,
      "loss": 0.5227,
      "step": 10334
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0037415027618408,
      "learning_rate": 2.8131662654848814e-06,
      "loss": 0.5071,
      "step": 10335
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0276168584823608,
      "learning_rate": 2.8122436526186935e-06,
      "loss": 0.5532,
      "step": 10336
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9621335864067078,
      "learning_rate": 2.8113211318713146e-06,
      "loss": 0.536,
      "step": 10337
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9891295433044434,
      "learning_rate": 2.810398703281589e-06,
      "loss": 0.491,
      "step": 10338
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.081217646598816,
      "learning_rate": 2.8094763668883567e-06,
      "loss": 0.4953,
      "step": 10339
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0589220523834229,
      "learning_rate": 2.808554122730457e-06,
      "loss": 0.4989,
      "step": 10340
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0248987674713135,
      "learning_rate": 2.8076319708467146e-06,
      "loss": 0.5145,
      "step": 10341
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1088663339614868,
      "learning_rate": 2.8067099112759665e-06,
      "loss": 0.522,
      "step": 10342
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0120095014572144,
      "learning_rate": 2.8057879440570356e-06,
      "loss": 0.5174,
      "step": 10343
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9366309642791748,
      "learning_rate": 2.804866069228739e-06,
      "loss": 0.472,
      "step": 10344
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1038411855697632,
      "learning_rate": 2.803944286829896e-06,
      "loss": 0.4936,
      "step": 10345
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0267481803894043,
      "learning_rate": 2.8030225968993198e-06,
      "loss": 0.5221,
      "step": 10346
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0624173879623413,
      "learning_rate": 2.802100999475819e-06,
      "loss": 0.4932,
      "step": 10347
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2405089139938354,
      "learning_rate": 2.8011794945982013e-06,
      "loss": 0.5502,
      "step": 10348
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9911314249038696,
      "learning_rate": 2.8002580823052638e-06,
      "loss": 0.4974,
      "step": 10349
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0014399290084839,
      "learning_rate": 2.7993367626358047e-06,
      "loss": 0.4675,
      "step": 10350
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9858644008636475,
      "learning_rate": 2.7984155356286224e-06,
      "loss": 0.4446,
      "step": 10351
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.096359372138977,
      "learning_rate": 2.7974944013225013e-06,
      "loss": 0.4912,
      "step": 10352
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0251573324203491,
      "learning_rate": 2.796573359756229e-06,
      "loss": 0.514,
      "step": 10353
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1570641994476318,
      "learning_rate": 2.7956524109685874e-06,
      "loss": 0.5704,
      "step": 10354
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9382094144821167,
      "learning_rate": 2.7947315549983545e-06,
      "loss": 0.514,
      "step": 10355
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0530364513397217,
      "learning_rate": 2.793810791884306e-06,
      "loss": 0.5164,
      "step": 10356
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0562663078308105,
      "learning_rate": 2.792890121665208e-06,
      "loss": 0.5073,
      "step": 10357
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1291313171386719,
      "learning_rate": 2.791969544379828e-06,
      "loss": 0.5317,
      "step": 10358
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0145524740219116,
      "learning_rate": 2.7910490600669327e-06,
      "loss": 0.5066,
      "step": 10359
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0272448062896729,
      "learning_rate": 2.790128668765275e-06,
      "loss": 0.5488,
      "step": 10360
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0094681978225708,
      "learning_rate": 2.789208370513612e-06,
      "loss": 0.4549,
      "step": 10361
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1560615301132202,
      "learning_rate": 2.7882881653506947e-06,
      "loss": 0.5397,
      "step": 10362
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9382747411727905,
      "learning_rate": 2.787368053315266e-06,
      "loss": 0.4965,
      "step": 10363
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0615712404251099,
      "learning_rate": 2.7864480344460743e-06,
      "loss": 0.4694,
      "step": 10364
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.137567162513733,
      "learning_rate": 2.7855281087818543e-06,
      "loss": 0.5259,
      "step": 10365
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0025049448013306,
      "learning_rate": 2.7846082763613412e-06,
      "loss": 0.4573,
      "step": 10366
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1223913431167603,
      "learning_rate": 2.783688537223268e-06,
      "loss": 0.58,
      "step": 10367
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0727946758270264,
      "learning_rate": 2.7827688914063596e-06,
      "loss": 0.5462,
      "step": 10368
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0876458883285522,
      "learning_rate": 2.78184933894934e-06,
      "loss": 0.4887,
      "step": 10369
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9600358009338379,
      "learning_rate": 2.780929879890931e-06,
      "loss": 0.5089,
      "step": 10370
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0861115455627441,
      "learning_rate": 2.780010514269841e-06,
      "loss": 0.5004,
      "step": 10371
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0740866661071777,
      "learning_rate": 2.7790912421247883e-06,
      "loss": 0.5236,
      "step": 10372
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0634615421295166,
      "learning_rate": 2.7781720634944766e-06,
      "loss": 0.5436,
      "step": 10373
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0422475337982178,
      "learning_rate": 2.77725297841761e-06,
      "loss": 0.4598,
      "step": 10374
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0295344591140747,
      "learning_rate": 2.7763339869328897e-06,
      "loss": 0.5102,
      "step": 10375
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0481268167495728,
      "learning_rate": 2.7754150890790067e-06,
      "loss": 0.5244,
      "step": 10376
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0282061100006104,
      "learning_rate": 2.7744962848946565e-06,
      "loss": 0.4872,
      "step": 10377
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0922186374664307,
      "learning_rate": 2.7735775744185276e-06,
      "loss": 0.5161,
      "step": 10378
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9676588773727417,
      "learning_rate": 2.7726589576893004e-06,
      "loss": 0.4734,
      "step": 10379
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0311994552612305,
      "learning_rate": 2.7717404347456567e-06,
      "loss": 0.4676,
      "step": 10380
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9735383987426758,
      "learning_rate": 2.7708220056262706e-06,
      "loss": 0.5032,
      "step": 10381
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.048998236656189,
      "learning_rate": 2.7699036703698158e-06,
      "loss": 0.5058,
      "step": 10382
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0708740949630737,
      "learning_rate": 2.7689854290149608e-06,
      "loss": 0.5323,
      "step": 10383
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0217273235321045,
      "learning_rate": 2.768067281600365e-06,
      "loss": 0.5148,
      "step": 10384
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.074256181716919,
      "learning_rate": 2.7671492281646937e-06,
      "loss": 0.4863,
      "step": 10385
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0781575441360474,
      "learning_rate": 2.7662312687466026e-06,
      "loss": 0.4688,
      "step": 10386
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9867082238197327,
      "learning_rate": 2.7653134033847393e-06,
      "loss": 0.5157,
      "step": 10387
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0689009428024292,
      "learning_rate": 2.7643956321177558e-06,
      "loss": 0.5145,
      "step": 10388
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9699030518531799,
      "learning_rate": 2.763477954984295e-06,
      "loss": 0.5254,
      "step": 10389
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0327718257904053,
      "learning_rate": 2.7625603720229964e-06,
      "loss": 0.5171,
      "step": 10390
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0824087858200073,
      "learning_rate": 2.7616428832724983e-06,
      "loss": 0.5744,
      "step": 10391
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0552942752838135,
      "learning_rate": 2.760725488771433e-06,
      "loss": 0.5646,
      "step": 10392
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0860564708709717,
      "learning_rate": 2.7598081885584237e-06,
      "loss": 0.5154,
      "step": 10393
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.087424635887146,
      "learning_rate": 2.758890982672102e-06,
      "loss": 0.5118,
      "step": 10394
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0438737869262695,
      "learning_rate": 2.757973871151083e-06,
      "loss": 0.5584,
      "step": 10395
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0562806129455566,
      "learning_rate": 2.757056854033985e-06,
      "loss": 0.5106,
      "step": 10396
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0467734336853027,
      "learning_rate": 2.7561399313594205e-06,
      "loss": 0.5218,
      "step": 10397
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2227981090545654,
      "learning_rate": 2.7552231031659972e-06,
      "loss": 0.5171,
      "step": 10398
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0784660577774048,
      "learning_rate": 2.75430636949232e-06,
      "loss": 0.5352,
      "step": 10399
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9642545580863953,
      "learning_rate": 2.753389730376992e-06,
      "loss": 0.5041,
      "step": 10400
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0255966186523438,
      "learning_rate": 2.752473185858603e-06,
      "loss": 0.51,
      "step": 10401
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1170417070388794,
      "learning_rate": 2.7515567359757526e-06,
      "loss": 0.5409,
      "step": 10402
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0311150550842285,
      "learning_rate": 2.750640380767025e-06,
      "loss": 0.5213,
      "step": 10403
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0396472215652466,
      "learning_rate": 2.7497241202710056e-06,
      "loss": 0.5205,
      "step": 10404
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0209033489227295,
      "learning_rate": 2.7488079545262757e-06,
      "loss": 0.534,
      "step": 10405
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1066263914108276,
      "learning_rate": 2.747891883571412e-06,
      "loss": 0.5341,
      "step": 10406
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.022965669631958,
      "learning_rate": 2.746975907444986e-06,
      "loss": 0.4647,
      "step": 10407
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0523531436920166,
      "learning_rate": 2.7460600261855687e-06,
      "loss": 0.551,
      "step": 10408
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0454198122024536,
      "learning_rate": 2.7451442398317206e-06,
      "loss": 0.5579,
      "step": 10409
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.031033992767334,
      "learning_rate": 2.7442285484220055e-06,
      "loss": 0.4956,
      "step": 10410
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.082212209701538,
      "learning_rate": 2.7433129519949784e-06,
      "loss": 0.5204,
      "step": 10411
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0533207654953003,
      "learning_rate": 2.742397450589193e-06,
      "loss": 0.5319,
      "step": 10412
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9724899530410767,
      "learning_rate": 2.7414820442431976e-06,
      "loss": 0.5085,
      "step": 10413
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1080093383789062,
      "learning_rate": 2.7405667329955344e-06,
      "loss": 0.546,
      "step": 10414
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.017476201057434,
      "learning_rate": 2.739651516884747e-06,
      "loss": 0.454,
      "step": 10415
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0047162771224976,
      "learning_rate": 2.7387363959493733e-06,
      "loss": 0.5082,
      "step": 10416
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0568748712539673,
      "learning_rate": 2.737821370227942e-06,
      "loss": 0.5333,
      "step": 10417
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1608695983886719,
      "learning_rate": 2.7369064397589828e-06,
      "loss": 0.5212,
      "step": 10418
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9917661547660828,
      "learning_rate": 2.7359916045810207e-06,
      "loss": 0.4802,
      "step": 10419
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.007771372795105,
      "learning_rate": 2.7350768647325766e-06,
      "loss": 0.5001,
      "step": 10420
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9977226257324219,
      "learning_rate": 2.734162220252168e-06,
      "loss": 0.447,
      "step": 10421
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.012022852897644,
      "learning_rate": 2.7332476711783044e-06,
      "loss": 0.5321,
      "step": 10422
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0312007665634155,
      "learning_rate": 2.732333217549494e-06,
      "loss": 0.5302,
      "step": 10423
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0605684518814087,
      "learning_rate": 2.7314188594042466e-06,
      "loss": 0.4972,
      "step": 10424
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0762418508529663,
      "learning_rate": 2.7305045967810585e-06,
      "loss": 0.4942,
      "step": 10425
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0472568273544312,
      "learning_rate": 2.7295904297184262e-06,
      "loss": 0.5072,
      "step": 10426
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0426957607269287,
      "learning_rate": 2.7286763582548424e-06,
      "loss": 0.5053,
      "step": 10427
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.085888147354126,
      "learning_rate": 2.7277623824287957e-06,
      "loss": 0.4957,
      "step": 10428
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0257681608200073,
      "learning_rate": 2.726848502278773e-06,
      "loss": 0.5649,
      "step": 10429
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0250338315963745,
      "learning_rate": 2.7259347178432493e-06,
      "loss": 0.4895,
      "step": 10430
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0389498472213745,
      "learning_rate": 2.7250210291607026e-06,
      "loss": 0.5052,
      "step": 10431
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0429414510726929,
      "learning_rate": 2.7241074362696108e-06,
      "loss": 0.4572,
      "step": 10432
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.091538667678833,
      "learning_rate": 2.7231939392084347e-06,
      "loss": 0.5382,
      "step": 10433
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0113235712051392,
      "learning_rate": 2.7222805380156414e-06,
      "loss": 0.5437,
      "step": 10434
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0240751504898071,
      "learning_rate": 2.7213672327296914e-06,
      "loss": 0.5194,
      "step": 10435
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.016080379486084,
      "learning_rate": 2.72045402338904e-06,
      "loss": 0.5137,
      "step": 10436
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0559788942337036,
      "learning_rate": 2.719540910032142e-06,
      "loss": 0.4918,
      "step": 10437
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0363986492156982,
      "learning_rate": 2.7186278926974406e-06,
      "loss": 0.5293,
      "step": 10438
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0203218460083008,
      "learning_rate": 2.717714971423383e-06,
      "loss": 0.4876,
      "step": 10439
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.032476544380188,
      "learning_rate": 2.7168021462484084e-06,
      "loss": 0.4804,
      "step": 10440
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0136157274246216,
      "learning_rate": 2.715889417210953e-06,
      "loss": 0.5163,
      "step": 10441
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0525606870651245,
      "learning_rate": 2.714976784349448e-06,
      "loss": 0.51,
      "step": 10442
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0758986473083496,
      "learning_rate": 2.7140642477023237e-06,
      "loss": 0.5201,
      "step": 10443
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0525035858154297,
      "learning_rate": 2.7131518073079976e-06,
      "loss": 0.5402,
      "step": 10444
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0384174585342407,
      "learning_rate": 2.7122394632048974e-06,
      "loss": 0.5201,
      "step": 10445
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0647464990615845,
      "learning_rate": 2.7113272154314328e-06,
      "loss": 0.5336,
      "step": 10446
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0159748792648315,
      "learning_rate": 2.710415064026018e-06,
      "loss": 0.4487,
      "step": 10447
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0917627811431885,
      "learning_rate": 2.7095030090270596e-06,
      "loss": 0.4907,
      "step": 10448
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9589986801147461,
      "learning_rate": 2.7085910504729617e-06,
      "loss": 0.5268,
      "step": 10449
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.976966142654419,
      "learning_rate": 2.7076791884021236e-06,
      "loss": 0.4598,
      "step": 10450
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0308228731155396,
      "learning_rate": 2.7067674228529417e-06,
      "loss": 0.5131,
      "step": 10451
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0818092823028564,
      "learning_rate": 2.7058557538638026e-06,
      "loss": 0.493,
      "step": 10452
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0554735660552979,
      "learning_rate": 2.7049441814731007e-06,
      "loss": 0.5117,
      "step": 10453
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9924067854881287,
      "learning_rate": 2.704032705719214e-06,
      "loss": 0.5183,
      "step": 10454
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1226341724395752,
      "learning_rate": 2.703121326640522e-06,
      "loss": 0.5464,
      "step": 10455
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9563577771186829,
      "learning_rate": 2.702210044275401e-06,
      "loss": 0.4453,
      "step": 10456
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0308963060379028,
      "learning_rate": 2.7012988586622224e-06,
      "loss": 0.5021,
      "step": 10457
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0026469230651855,
      "learning_rate": 2.7003877698393512e-06,
      "loss": 0.4859,
      "step": 10458
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0531280040740967,
      "learning_rate": 2.6994767778451535e-06,
      "loss": 0.4827,
      "step": 10459
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0416374206542969,
      "learning_rate": 2.6985658827179845e-06,
      "loss": 0.519,
      "step": 10460
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9877068400382996,
      "learning_rate": 2.6976550844961992e-06,
      "loss": 0.4707,
      "step": 10461
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0973079204559326,
      "learning_rate": 2.6967443832181496e-06,
      "loss": 0.5126,
      "step": 10462
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.960408091545105,
      "learning_rate": 2.6958337789221813e-06,
      "loss": 0.4921,
      "step": 10463
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1162091493606567,
      "learning_rate": 2.694923271646637e-06,
      "loss": 0.4924,
      "step": 10464
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9959160089492798,
      "learning_rate": 2.694012861429855e-06,
      "loss": 0.5197,
      "step": 10465
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.061828851699829,
      "learning_rate": 2.693102548310169e-06,
      "loss": 0.5235,
      "step": 10466
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0121952295303345,
      "learning_rate": 2.6921923323259124e-06,
      "loss": 0.5247,
      "step": 10467
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.118021845817566,
      "learning_rate": 2.691282213515406e-06,
      "loss": 0.5105,
      "step": 10468
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.962216317653656,
      "learning_rate": 2.690372191916974e-06,
      "loss": 0.499,
      "step": 10469
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0699259042739868,
      "learning_rate": 2.6894622675689345e-06,
      "loss": 0.5279,
      "step": 10470
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0497336387634277,
      "learning_rate": 2.6885524405096007e-06,
      "loss": 0.5426,
      "step": 10471
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1183973550796509,
      "learning_rate": 2.687642710777284e-06,
      "loss": 0.5466,
      "step": 10472
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0451266765594482,
      "learning_rate": 2.6867330784102896e-06,
      "loss": 0.4896,
      "step": 10473
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0753915309906006,
      "learning_rate": 2.6858235434469138e-06,
      "loss": 0.5291,
      "step": 10474
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9498035907745361,
      "learning_rate": 2.684914105925463e-06,
      "loss": 0.4851,
      "step": 10475
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0795294046401978,
      "learning_rate": 2.6840047658842226e-06,
      "loss": 0.4516,
      "step": 10476
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.12007737159729,
      "learning_rate": 2.683095523361486e-06,
      "loss": 0.5383,
      "step": 10477
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1048475503921509,
      "learning_rate": 2.682186378395536e-06,
      "loss": 0.5598,
      "step": 10478
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.258007526397705,
      "learning_rate": 2.6812773310246547e-06,
      "loss": 0.5166,
      "step": 10479
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.981302797794342,
      "learning_rate": 2.680368381287119e-06,
      "loss": 0.4954,
      "step": 10480
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0412334203720093,
      "learning_rate": 2.6794595292212035e-06,
      "loss": 0.4819,
      "step": 10481
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0630468130111694,
      "learning_rate": 2.67855077486517e-06,
      "loss": 0.5235,
      "step": 10482
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1821410655975342,
      "learning_rate": 2.677642118257292e-06,
      "loss": 0.5377,
      "step": 10483
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.074578046798706,
      "learning_rate": 2.6767335594358234e-06,
      "loss": 0.5425,
      "step": 10484
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9681217670440674,
      "learning_rate": 2.675825098439023e-06,
      "loss": 0.4821,
      "step": 10485
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0614813566207886,
      "learning_rate": 2.6749167353051443e-06,
      "loss": 0.5295,
      "step": 10486
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0384877920150757,
      "learning_rate": 2.674008470072429e-06,
      "loss": 0.482,
      "step": 10487
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1568013429641724,
      "learning_rate": 2.673100302779128e-06,
      "loss": 0.5158,
      "step": 10488
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9850515723228455,
      "learning_rate": 2.6721922334634804e-06,
      "loss": 0.4893,
      "step": 10489
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0520575046539307,
      "learning_rate": 2.671284262163718e-06,
      "loss": 0.5214,
      "step": 10490
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.033778190612793,
      "learning_rate": 2.6703763889180746e-06,
      "loss": 0.4863,
      "step": 10491
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0140942335128784,
      "learning_rate": 2.6694686137647767e-06,
      "loss": 0.5278,
      "step": 10492
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0416064262390137,
      "learning_rate": 2.668560936742048e-06,
      "loss": 0.48,
      "step": 10493
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0666289329528809,
      "learning_rate": 2.6676533578881102e-06,
      "loss": 0.4862,
      "step": 10494
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9985889196395874,
      "learning_rate": 2.6667458772411724e-06,
      "loss": 0.488,
      "step": 10495
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0429118871688843,
      "learning_rate": 2.66583849483945e-06,
      "loss": 0.5285,
      "step": 10496
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9838635325431824,
      "learning_rate": 2.664931210721151e-06,
      "loss": 0.5122,
      "step": 10497
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9577754735946655,
      "learning_rate": 2.6640240249244744e-06,
      "loss": 0.4817,
      "step": 10498
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9911318421363831,
      "learning_rate": 2.6631169374876185e-06,
      "loss": 0.4981,
      "step": 10499
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.989579439163208,
      "learning_rate": 2.6622099484487794e-06,
      "loss": 0.5132,
      "step": 10500
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.030277967453003,
      "learning_rate": 2.6613030578461476e-06,
      "loss": 0.4955,
      "step": 10501
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0554670095443726,
      "learning_rate": 2.6603962657179094e-06,
      "loss": 0.5048,
      "step": 10502
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0888514518737793,
      "learning_rate": 2.6594895721022436e-06,
      "loss": 0.5773,
      "step": 10503
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0285956859588623,
      "learning_rate": 2.6585829770373286e-06,
      "loss": 0.5055,
      "step": 10504
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.151036262512207,
      "learning_rate": 2.657676480561342e-06,
      "loss": 0.5474,
      "step": 10505
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9403602480888367,
      "learning_rate": 2.6567700827124494e-06,
      "loss": 0.4994,
      "step": 10506
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0743473768234253,
      "learning_rate": 2.655863783528817e-06,
      "loss": 0.5122,
      "step": 10507
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1450196504592896,
      "learning_rate": 2.6549575830486053e-06,
      "loss": 0.5577,
      "step": 10508
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0517287254333496,
      "learning_rate": 2.6540514813099728e-06,
      "loss": 0.5208,
      "step": 10509
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0516631603240967,
      "learning_rate": 2.6531454783510736e-06,
      "loss": 0.5507,
      "step": 10510
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1319580078125,
      "learning_rate": 2.6522395742100514e-06,
      "loss": 0.546,
      "step": 10511
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0354925394058228,
      "learning_rate": 2.651333768925052e-06,
      "loss": 0.5001,
      "step": 10512
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1074974536895752,
      "learning_rate": 2.6504280625342203e-06,
      "loss": 0.5803,
      "step": 10513
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0434691905975342,
      "learning_rate": 2.6495224550756888e-06,
      "loss": 0.5396,
      "step": 10514
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.013555884361267,
      "learning_rate": 2.6486169465875887e-06,
      "loss": 0.5151,
      "step": 10515
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0723886489868164,
      "learning_rate": 2.647711537108052e-06,
      "loss": 0.5255,
      "step": 10516
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.099655270576477,
      "learning_rate": 2.6468062266751955e-06,
      "loss": 0.5143,
      "step": 10517
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9594826698303223,
      "learning_rate": 2.6459010153271456e-06,
      "loss": 0.482,
      "step": 10518
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0790048837661743,
      "learning_rate": 2.6449959031020134e-06,
      "loss": 0.5033,
      "step": 10519
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1070122718811035,
      "learning_rate": 2.6440908900379115e-06,
      "loss": 0.4909,
      "step": 10520
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9774593114852905,
      "learning_rate": 2.6431859761729462e-06,
      "loss": 0.5239,
      "step": 10521
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0332831144332886,
      "learning_rate": 2.6422811615452205e-06,
      "loss": 0.4888,
      "step": 10522
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0701346397399902,
      "learning_rate": 2.6413764461928335e-06,
      "loss": 0.4977,
      "step": 10523
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9489225149154663,
      "learning_rate": 2.6404718301538814e-06,
      "loss": 0.5256,
      "step": 10524
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9808928370475769,
      "learning_rate": 2.639567313466448e-06,
      "loss": 0.4806,
      "step": 10525
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9877341985702515,
      "learning_rate": 2.6386628961686277e-06,
      "loss": 0.4937,
      "step": 10526
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.076948642730713,
      "learning_rate": 2.6377585782984972e-06,
      "loss": 0.4969,
      "step": 10527
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0355870723724365,
      "learning_rate": 2.636854359894134e-06,
      "loss": 0.4837,
      "step": 10528
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.047590732574463,
      "learning_rate": 2.635950240993614e-06,
      "loss": 0.5303,
      "step": 10529
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0804356336593628,
      "learning_rate": 2.635046221635005e-06,
      "loss": 0.5481,
      "step": 10530
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9647104144096375,
      "learning_rate": 2.6341423018563727e-06,
      "loss": 0.5039,
      "step": 10531
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0416531562805176,
      "learning_rate": 2.633238481695779e-06,
      "loss": 0.4922,
      "step": 10532
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0723743438720703,
      "learning_rate": 2.6323347611912786e-06,
      "loss": 0.5032,
      "step": 10533
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1137053966522217,
      "learning_rate": 2.6314311403809224e-06,
      "loss": 0.5648,
      "step": 10534
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9235475063323975,
      "learning_rate": 2.630527619302765e-06,
      "loss": 0.4404,
      "step": 10535
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0233099460601807,
      "learning_rate": 2.6296241979948455e-06,
      "loss": 0.5304,
      "step": 10536
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0859589576721191,
      "learning_rate": 2.6287208764952045e-06,
      "loss": 0.5345,
      "step": 10537
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0127158164978027,
      "learning_rate": 2.6278176548418783e-06,
      "loss": 0.5178,
      "step": 10538
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0827937126159668,
      "learning_rate": 2.6269145330728985e-06,
      "loss": 0.4957,
      "step": 10539
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.040254831314087,
      "learning_rate": 2.626011511226294e-06,
      "loss": 0.5154,
      "step": 10540
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0404975414276123,
      "learning_rate": 2.625108589340085e-06,
      "loss": 0.5102,
      "step": 10541
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9493241906166077,
      "learning_rate": 2.624205767452289e-06,
      "loss": 0.4422,
      "step": 10542
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9997432827949524,
      "learning_rate": 2.623303045600928e-06,
      "loss": 0.4639,
      "step": 10543
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1698287725448608,
      "learning_rate": 2.622400423824005e-06,
      "loss": 0.5274,
      "step": 10544
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0547444820404053,
      "learning_rate": 2.62149790215953e-06,
      "loss": 0.4699,
      "step": 10545
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9943386912345886,
      "learning_rate": 2.6205954806455057e-06,
      "loss": 0.5104,
      "step": 10546
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.054709553718567,
      "learning_rate": 2.6196931593199247e-06,
      "loss": 0.5321,
      "step": 10547
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1206477880477905,
      "learning_rate": 2.618790938220788e-06,
      "loss": 0.51,
      "step": 10548
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0720930099487305,
      "learning_rate": 2.617888817386079e-06,
      "loss": 0.4765,
      "step": 10549
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0825947523117065,
      "learning_rate": 2.6169867968537856e-06,
      "loss": 0.5156,
      "step": 10550
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0014879703521729,
      "learning_rate": 2.616084876661888e-06,
      "loss": 0.468,
      "step": 10551
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.965927243232727,
      "learning_rate": 2.6151830568483627e-06,
      "loss": 0.5206,
      "step": 10552
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0722675323486328,
      "learning_rate": 2.614281337451183e-06,
      "loss": 0.5103,
      "step": 10553
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0606327056884766,
      "learning_rate": 2.61337971850832e-06,
      "loss": 0.5609,
      "step": 10554
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9931026101112366,
      "learning_rate": 2.6124782000577296e-06,
      "loss": 0.4632,
      "step": 10555
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0780014991760254,
      "learning_rate": 2.6115767821373807e-06,
      "loss": 0.4643,
      "step": 10556
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0118858814239502,
      "learning_rate": 2.610675464785223e-06,
      "loss": 0.5311,
      "step": 10557
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9681828022003174,
      "learning_rate": 2.6097742480392097e-06,
      "loss": 0.4892,
      "step": 10558
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9976950287818909,
      "learning_rate": 2.6088731319372874e-06,
      "loss": 0.5262,
      "step": 10559
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0612802505493164,
      "learning_rate": 2.6079721165173994e-06,
      "loss": 0.5053,
      "step": 10560
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2253762483596802,
      "learning_rate": 2.6070712018174847e-06,
      "loss": 0.4711,
      "step": 10561
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0154792070388794,
      "learning_rate": 2.6061703878754784e-06,
      "loss": 0.5201,
      "step": 10562
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0816336870193481,
      "learning_rate": 2.6052696747293087e-06,
      "loss": 0.488,
      "step": 10563
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0748910903930664,
      "learning_rate": 2.6043690624169014e-06,
      "loss": 0.5206,
      "step": 10564
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0948655605316162,
      "learning_rate": 2.6034685509761803e-06,
      "loss": 0.5116,
      "step": 10565
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0533651113510132,
      "learning_rate": 2.602568140445061e-06,
      "loss": 0.5126,
      "step": 10566
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.120794653892517,
      "learning_rate": 2.6016678308614583e-06,
      "loss": 0.4874,
      "step": 10567
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9894031882286072,
      "learning_rate": 2.600767622263277e-06,
      "loss": 0.4943,
      "step": 10568
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0337748527526855,
      "learning_rate": 2.599867514688427e-06,
      "loss": 0.5128,
      "step": 10569
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9865636229515076,
      "learning_rate": 2.598967508174808e-06,
      "loss": 0.4648,
      "step": 10570
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0098962783813477,
      "learning_rate": 2.598067602760313e-06,
      "loss": 0.4785,
      "step": 10571
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9975713491439819,
      "learning_rate": 2.597167798482835e-06,
      "loss": 0.5205,
      "step": 10572
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1072999238967896,
      "learning_rate": 2.596268095380263e-06,
      "loss": 0.5235,
      "step": 10573
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0465779304504395,
      "learning_rate": 2.5953684934904788e-06,
      "loss": 0.4864,
      "step": 10574
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.051168441772461,
      "learning_rate": 2.5944689928513643e-06,
      "loss": 0.4652,
      "step": 10575
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0790677070617676,
      "learning_rate": 2.593569593500789e-06,
      "loss": 0.4975,
      "step": 10576
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0252071619033813,
      "learning_rate": 2.592670295476628e-06,
      "loss": 0.4697,
      "step": 10577
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.033689260482788,
      "learning_rate": 2.591771098816749e-06,
      "loss": 0.5521,
      "step": 10578
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0405455827713013,
      "learning_rate": 2.5908720035590085e-06,
      "loss": 0.5088,
      "step": 10579
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0239884853363037,
      "learning_rate": 2.5899730097412678e-06,
      "loss": 0.5009,
      "step": 10580
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.033341646194458,
      "learning_rate": 2.58907411740138e-06,
      "loss": 0.5109,
      "step": 10581
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9961441159248352,
      "learning_rate": 2.5881753265771938e-06,
      "loss": 0.5134,
      "step": 10582
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9867962598800659,
      "learning_rate": 2.587276637306556e-06,
      "loss": 0.448,
      "step": 10583
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0330294370651245,
      "learning_rate": 2.586378049627304e-06,
      "loss": 0.5084,
      "step": 10584
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0358963012695312,
      "learning_rate": 2.5854795635772743e-06,
      "loss": 0.5136,
      "step": 10585
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.01925790309906,
      "learning_rate": 2.584581179194304e-06,
      "loss": 0.5117,
      "step": 10586
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0224287509918213,
      "learning_rate": 2.5836828965162167e-06,
      "loss": 0.518,
      "step": 10587
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0976794958114624,
      "learning_rate": 2.582784715580836e-06,
      "loss": 0.5151,
      "step": 10588
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0719406604766846,
      "learning_rate": 2.581886636425983e-06,
      "loss": 0.5174,
      "step": 10589
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2258132696151733,
      "learning_rate": 2.580988659089471e-06,
      "loss": 0.533,
      "step": 10590
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.126112699508667,
      "learning_rate": 2.580090783609114e-06,
      "loss": 0.545,
      "step": 10591
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0415563583374023,
      "learning_rate": 2.5791930100227133e-06,
      "loss": 0.4935,
      "step": 10592
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.033074140548706,
      "learning_rate": 2.5782953383680733e-06,
      "loss": 0.4644,
      "step": 10593
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.00439453125,
      "learning_rate": 2.5773977686829928e-06,
      "loss": 0.5112,
      "step": 10594
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0228551626205444,
      "learning_rate": 2.5765003010052643e-06,
      "loss": 0.4834,
      "step": 10595
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.091952919960022,
      "learning_rate": 2.5756029353726777e-06,
      "loss": 0.5088,
      "step": 10596
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0610514879226685,
      "learning_rate": 2.574705671823019e-06,
      "loss": 0.538,
      "step": 10597
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.093294620513916,
      "learning_rate": 2.5738085103940634e-06,
      "loss": 0.5479,
      "step": 10598
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0571091175079346,
      "learning_rate": 2.572911451123594e-06,
      "loss": 0.5387,
      "step": 10599
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0951002836227417,
      "learning_rate": 2.572014494049382e-06,
      "loss": 0.5231,
      "step": 10600
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.038913369178772,
      "learning_rate": 2.571117639209191e-06,
      "loss": 0.5439,
      "step": 10601
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0661230087280273,
      "learning_rate": 2.5702208866407873e-06,
      "loss": 0.4539,
      "step": 10602
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9565131664276123,
      "learning_rate": 2.5693242363819292e-06,
      "loss": 0.4747,
      "step": 10603
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0338586568832397,
      "learning_rate": 2.5684276884703717e-06,
      "loss": 0.496,
      "step": 10604
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1044895648956299,
      "learning_rate": 2.567531242943867e-06,
      "loss": 0.5165,
      "step": 10605
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1141607761383057,
      "learning_rate": 2.5666348998401565e-06,
      "loss": 0.499,
      "step": 10606
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0243730545043945,
      "learning_rate": 2.565738659196987e-06,
      "loss": 0.4852,
      "step": 10607
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0987368822097778,
      "learning_rate": 2.5648425210520967e-06,
      "loss": 0.5726,
      "step": 10608
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0123798847198486,
      "learning_rate": 2.563946485443214e-06,
      "loss": 0.4462,
      "step": 10609
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0435576438903809,
      "learning_rate": 2.5630505524080707e-06,
      "loss": 0.488,
      "step": 10610
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0640788078308105,
      "learning_rate": 2.5621547219843905e-06,
      "loss": 0.5194,
      "step": 10611
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.094903826713562,
      "learning_rate": 2.5612589942098952e-06,
      "loss": 0.4869,
      "step": 10612
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0041868686676025,
      "learning_rate": 2.560363369122301e-06,
      "loss": 0.5348,
      "step": 10613
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.072270154953003,
      "learning_rate": 2.559467846759317e-06,
      "loss": 0.5234,
      "step": 10614
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0632047653198242,
      "learning_rate": 2.5585724271586505e-06,
      "loss": 0.5144,
      "step": 10615
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0923739671707153,
      "learning_rate": 2.557677110358009e-06,
      "loss": 0.5269,
      "step": 10616
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0472354888916016,
      "learning_rate": 2.556781896395087e-06,
      "loss": 0.4982,
      "step": 10617
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.073376178741455,
      "learning_rate": 2.55588678530758e-06,
      "loss": 0.4807,
      "step": 10618
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.096312165260315,
      "learning_rate": 2.5549917771331767e-06,
      "loss": 0.4738,
      "step": 10619
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0125393867492676,
      "learning_rate": 2.5540968719095656e-06,
      "loss": 0.4835,
      "step": 10620
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0725840330123901,
      "learning_rate": 2.5532020696744277e-06,
      "loss": 0.4868,
      "step": 10621
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1230672597885132,
      "learning_rate": 2.5523073704654374e-06,
      "loss": 0.5352,
      "step": 10622
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0457652807235718,
      "learning_rate": 2.5514127743202668e-06,
      "loss": 0.5059,
      "step": 10623
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0251890420913696,
      "learning_rate": 2.5505182812765894e-06,
      "loss": 0.5243,
      "step": 10624
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0891318321228027,
      "learning_rate": 2.549623891372065e-06,
      "loss": 0.539,
      "step": 10625
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1247227191925049,
      "learning_rate": 2.5487296046443537e-06,
      "loss": 0.4829,
      "step": 10626
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.102414608001709,
      "learning_rate": 2.547835421131114e-06,
      "loss": 0.5509,
      "step": 10627
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.024851679801941,
      "learning_rate": 2.5469413408699894e-06,
      "loss": 0.4753,
      "step": 10628
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.136244773864746,
      "learning_rate": 2.546047363898636e-06,
      "loss": 0.5328,
      "step": 10629
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0303579568862915,
      "learning_rate": 2.545153490254689e-06,
      "loss": 0.5493,
      "step": 10630
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9925675392150879,
      "learning_rate": 2.5442597199757896e-06,
      "loss": 0.4668,
      "step": 10631
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0880279541015625,
      "learning_rate": 2.5433660530995696e-06,
      "loss": 0.5369,
      "step": 10632
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.025112271308899,
      "learning_rate": 2.54247248966366e-06,
      "loss": 0.4766,
      "step": 10633
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0779856443405151,
      "learning_rate": 2.5415790297056843e-06,
      "loss": 0.477,
      "step": 10634
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0276527404785156,
      "learning_rate": 2.5406856732632647e-06,
      "loss": 0.5265,
      "step": 10635
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0492390394210815,
      "learning_rate": 2.539792420374013e-06,
      "loss": 0.5449,
      "step": 10636
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0722205638885498,
      "learning_rate": 2.5388992710755477e-06,
      "loss": 0.4909,
      "step": 10637
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9178640842437744,
      "learning_rate": 2.5380062254054706e-06,
      "loss": 0.4667,
      "step": 10638
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.117708683013916,
      "learning_rate": 2.5371132834013867e-06,
      "loss": 0.5095,
      "step": 10639
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9836210012435913,
      "learning_rate": 2.5362204451008963e-06,
      "loss": 0.4706,
      "step": 10640
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0415635108947754,
      "learning_rate": 2.5353277105415887e-06,
      "loss": 0.536,
      "step": 10641
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0077011585235596,
      "learning_rate": 2.5344350797610597e-06,
      "loss": 0.4601,
      "step": 10642
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0329313278198242,
      "learning_rate": 2.533542552796893e-06,
      "loss": 0.4972,
      "step": 10643
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1889619827270508,
      "learning_rate": 2.5326501296866677e-06,
      "loss": 0.5188,
      "step": 10644
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.997260570526123,
      "learning_rate": 2.531757810467963e-06,
      "loss": 0.4805,
      "step": 10645
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9727004766464233,
      "learning_rate": 2.53086559517835e-06,
      "loss": 0.504,
      "step": 10646
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0125079154968262,
      "learning_rate": 2.529973483855397e-06,
      "loss": 0.4527,
      "step": 10647
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.019005537033081,
      "learning_rate": 2.52908147653667e-06,
      "loss": 0.4595,
      "step": 10648
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0392253398895264,
      "learning_rate": 2.5281895732597227e-06,
      "loss": 0.5012,
      "step": 10649
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.103070855140686,
      "learning_rate": 2.527297774062115e-06,
      "loss": 0.4936,
      "step": 10650
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1274436712265015,
      "learning_rate": 2.5264060789813994e-06,
      "loss": 0.4775,
      "step": 10651
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9909786581993103,
      "learning_rate": 2.525514488055116e-06,
      "loss": 0.507,
      "step": 10652
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0899275541305542,
      "learning_rate": 2.5246230013208093e-06,
      "loss": 0.5599,
      "step": 10653
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9369357824325562,
      "learning_rate": 2.5237316188160165e-06,
      "loss": 0.4331,
      "step": 10654
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0517666339874268,
      "learning_rate": 2.522840340578272e-06,
      "loss": 0.5221,
      "step": 10655
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.086459994316101,
      "learning_rate": 2.521949166645102e-06,
      "loss": 0.5641,
      "step": 10656
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0164384841918945,
      "learning_rate": 2.5210580970540354e-06,
      "loss": 0.4767,
      "step": 10657
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0612106323242188,
      "learning_rate": 2.5201671318425834e-06,
      "loss": 0.5062,
      "step": 10658
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0751674175262451,
      "learning_rate": 2.519276271048272e-06,
      "loss": 0.5299,
      "step": 10659
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.109610915184021,
      "learning_rate": 2.5183855147086045e-06,
      "loss": 0.5302,
      "step": 10660
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0094627141952515,
      "learning_rate": 2.51749486286109e-06,
      "loss": 0.4559,
      "step": 10661
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9121614098548889,
      "learning_rate": 2.516604315543231e-06,
      "loss": 0.4979,
      "step": 10662
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1115944385528564,
      "learning_rate": 2.515713872792525e-06,
      "loss": 0.5419,
      "step": 10663
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0575385093688965,
      "learning_rate": 2.5148235346464654e-06,
      "loss": 0.4703,
      "step": 10664
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1535428762435913,
      "learning_rate": 2.5139333011425435e-06,
      "loss": 0.5102,
      "step": 10665
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9647922515869141,
      "learning_rate": 2.5130431723182386e-06,
      "loss": 0.4526,
      "step": 10666
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1463685035705566,
      "learning_rate": 2.512153148211038e-06,
      "loss": 0.5403,
      "step": 10667
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0591135025024414,
      "learning_rate": 2.5112632288584116e-06,
      "loss": 0.5183,
      "step": 10668
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.058384895324707,
      "learning_rate": 2.5103734142978325e-06,
      "loss": 0.5627,
      "step": 10669
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0416806936264038,
      "learning_rate": 2.5094837045667684e-06,
      "loss": 0.4947,
      "step": 10670
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9496021270751953,
      "learning_rate": 2.508594099702682e-06,
      "loss": 0.5067,
      "step": 10671
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.076542854309082,
      "learning_rate": 2.5077045997430304e-06,
      "loss": 0.5172,
      "step": 10672
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0082660913467407,
      "learning_rate": 2.5068152047252702e-06,
      "loss": 0.529,
      "step": 10673
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1301501989364624,
      "learning_rate": 2.5059259146868474e-06,
      "loss": 0.5218,
      "step": 10674
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0519551038742065,
      "learning_rate": 2.5050367296652075e-06,
      "loss": 0.4938,
      "step": 10675
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0395547151565552,
      "learning_rate": 2.504147649697791e-06,
      "loss": 0.5311,
      "step": 10676
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9280543327331543,
      "learning_rate": 2.5032586748220354e-06,
      "loss": 0.5124,
      "step": 10677
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0799845457077026,
      "learning_rate": 2.5023698050753732e-06,
      "loss": 0.5323,
      "step": 10678
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9525831937789917,
      "learning_rate": 2.5014810404952262e-06,
      "loss": 0.4677,
      "step": 10679
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9634466767311096,
      "learning_rate": 2.5005923811190226e-06,
      "loss": 0.499,
      "step": 10680
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0382812023162842,
      "learning_rate": 2.4997038269841804e-06,
      "loss": 0.4953,
      "step": 10681
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0796928405761719,
      "learning_rate": 2.498815378128111e-06,
      "loss": 0.5341,
      "step": 10682
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9890086054801941,
      "learning_rate": 2.497927034588225e-06,
      "loss": 0.4597,
      "step": 10683
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.052189826965332,
      "learning_rate": 2.497038796401927e-06,
      "loss": 0.4985,
      "step": 10684
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0046368837356567,
      "learning_rate": 2.4961506636066185e-06,
      "loss": 0.5485,
      "step": 10685
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.116420865058899,
      "learning_rate": 2.495262636239697e-06,
      "loss": 0.5483,
      "step": 10686
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1046727895736694,
      "learning_rate": 2.4943747143385503e-06,
      "loss": 0.5414,
      "step": 10687
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0905431509017944,
      "learning_rate": 2.4934868979405667e-06,
      "loss": 0.5231,
      "step": 10688
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.111757755279541,
      "learning_rate": 2.492599187083134e-06,
      "loss": 0.5722,
      "step": 10689
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9807526469230652,
      "learning_rate": 2.491711581803625e-06,
      "loss": 0.4776,
      "step": 10690
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9862124919891357,
      "learning_rate": 2.490824082139415e-06,
      "loss": 0.4711,
      "step": 10691
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.117592215538025,
      "learning_rate": 2.489936688127875e-06,
      "loss": 0.5298,
      "step": 10692
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0601587295532227,
      "learning_rate": 2.4890493998063685e-06,
      "loss": 0.5039,
      "step": 10693
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.090985894203186,
      "learning_rate": 2.4881622172122595e-06,
      "loss": 0.5655,
      "step": 10694
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.056349515914917,
      "learning_rate": 2.4872751403828986e-06,
      "loss": 0.5423,
      "step": 10695
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.168630838394165,
      "learning_rate": 2.4863881693556393e-06,
      "loss": 0.5241,
      "step": 10696
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.084328055381775,
      "learning_rate": 2.4855013041678335e-06,
      "loss": 0.4798,
      "step": 10697
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.046592116355896,
      "learning_rate": 2.484614544856819e-06,
      "loss": 0.5097,
      "step": 10698
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0084978342056274,
      "learning_rate": 2.483727891459935e-06,
      "loss": 0.5117,
      "step": 10699
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.095803141593933,
      "learning_rate": 2.482841344014516e-06,
      "loss": 0.4997,
      "step": 10700
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0160983800888062,
      "learning_rate": 2.4819549025578917e-06,
      "loss": 0.5153,
      "step": 10701
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0845803022384644,
      "learning_rate": 2.481068567127389e-06,
      "loss": 0.5097,
      "step": 10702
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0083107948303223,
      "learning_rate": 2.4801823377603236e-06,
      "loss": 0.4816,
      "step": 10703
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1475107669830322,
      "learning_rate": 2.4792962144940148e-06,
      "loss": 0.5146,
      "step": 10704
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1041582822799683,
      "learning_rate": 2.4784101973657724e-06,
      "loss": 0.5572,
      "step": 10705
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9976695775985718,
      "learning_rate": 2.4775242864129055e-06,
      "loss": 0.4891,
      "step": 10706
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0840741395950317,
      "learning_rate": 2.4766384816727164e-06,
      "loss": 0.5289,
      "step": 10707
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0682698488235474,
      "learning_rate": 2.475752783182504e-06,
      "loss": 0.5124,
      "step": 10708
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0782908201217651,
      "learning_rate": 2.4748671909795568e-06,
      "loss": 0.5061,
      "step": 10709
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.015122652053833,
      "learning_rate": 2.4739817051011717e-06,
      "loss": 0.4699,
      "step": 10710
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0786633491516113,
      "learning_rate": 2.473096325584628e-06,
      "loss": 0.515,
      "step": 10711
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0838096141815186,
      "learning_rate": 2.4722110524672074e-06,
      "loss": 0.5407,
      "step": 10712
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1118197441101074,
      "learning_rate": 2.4713258857861856e-06,
      "loss": 0.5248,
      "step": 10713
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.047063946723938,
      "learning_rate": 2.4704408255788342e-06,
      "loss": 0.5378,
      "step": 10714
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9615069031715393,
      "learning_rate": 2.4695558718824204e-06,
      "loss": 0.4926,
      "step": 10715
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9735535383224487,
      "learning_rate": 2.468671024734208e-06,
      "loss": 0.4964,
      "step": 10716
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0757182836532593,
      "learning_rate": 2.4677862841714485e-06,
      "loss": 0.4802,
      "step": 10717
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9946287870407104,
      "learning_rate": 2.4669016502314038e-06,
      "loss": 0.5065,
      "step": 10718
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1031183004379272,
      "learning_rate": 2.4660171229513165e-06,
      "loss": 0.5147,
      "step": 10719
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0394502878189087,
      "learning_rate": 2.465132702368433e-06,
      "loss": 0.4895,
      "step": 10720
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9271803498268127,
      "learning_rate": 2.4642483885199938e-06,
      "loss": 0.4821,
      "step": 10721
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0470598936080933,
      "learning_rate": 2.463364181443233e-06,
      "loss": 0.5455,
      "step": 10722
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0677751302719116,
      "learning_rate": 2.4624800811753826e-06,
      "loss": 0.5372,
      "step": 10723
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.032324194908142,
      "learning_rate": 2.4615960877536706e-06,
      "loss": 0.4579,
      "step": 10724
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9706496596336365,
      "learning_rate": 2.4607122012153146e-06,
      "loss": 0.482,
      "step": 10725
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0008245706558228,
      "learning_rate": 2.459828421597534e-06,
      "loss": 0.4669,
      "step": 10726
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1143993139266968,
      "learning_rate": 2.458944748937543e-06,
      "loss": 0.5348,
      "step": 10727
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0865949392318726,
      "learning_rate": 2.4580611832725482e-06,
      "loss": 0.4915,
      "step": 10728
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0539220571517944,
      "learning_rate": 2.4571777246397543e-06,
      "loss": 0.5598,
      "step": 10729
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0416845083236694,
      "learning_rate": 2.456294373076361e-06,
      "loss": 0.497,
      "step": 10730
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0223973989486694,
      "learning_rate": 2.455411128619562e-06,
      "loss": 0.4578,
      "step": 10731
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1366970539093018,
      "learning_rate": 2.4545279913065513e-06,
      "loss": 0.5305,
      "step": 10732
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1712008714675903,
      "learning_rate": 2.4536449611745087e-06,
      "loss": 0.5753,
      "step": 10733
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0615066289901733,
      "learning_rate": 2.45276203826062e-06,
      "loss": 0.5046,
      "step": 10734
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9939412474632263,
      "learning_rate": 2.451879222602059e-06,
      "loss": 0.4732,
      "step": 10735
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2695436477661133,
      "learning_rate": 2.4509965142360013e-06,
      "loss": 0.5399,
      "step": 10736
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1129322052001953,
      "learning_rate": 2.4501139131996122e-06,
      "loss": 0.5113,
      "step": 10737
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.116741418838501,
      "learning_rate": 2.4492314195300583e-06,
      "loss": 0.5192,
      "step": 10738
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0863484144210815,
      "learning_rate": 2.4483490332644918e-06,
      "loss": 0.5285,
      "step": 10739
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0622340440750122,
      "learning_rate": 2.4474667544400744e-06,
      "loss": 0.4946,
      "step": 10740
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0966776609420776,
      "learning_rate": 2.4465845830939504e-06,
      "loss": 0.553,
      "step": 10741
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9429232478141785,
      "learning_rate": 2.4457025192632672e-06,
      "loss": 0.452,
      "step": 10742
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9729829430580139,
      "learning_rate": 2.444820562985165e-06,
      "loss": 0.4634,
      "step": 10743
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.085930585861206,
      "learning_rate": 2.443938714296781e-06,
      "loss": 0.5135,
      "step": 10744
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9769576787948608,
      "learning_rate": 2.4430569732352444e-06,
      "loss": 0.5007,
      "step": 10745
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0735533237457275,
      "learning_rate": 2.4421753398376865e-06,
      "loss": 0.5416,
      "step": 10746
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0380295515060425,
      "learning_rate": 2.441293814141223e-06,
      "loss": 0.5186,
      "step": 10747
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0052216053009033,
      "learning_rate": 2.4404123961829795e-06,
      "loss": 0.4998,
      "step": 10748
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9670013189315796,
      "learning_rate": 2.4395310860000644e-06,
      "loss": 0.4764,
      "step": 10749
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0874783992767334,
      "learning_rate": 2.438649883629588e-06,
      "loss": 0.5141,
      "step": 10750
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0350924730300903,
      "learning_rate": 2.437768789108656e-06,
      "loss": 0.5431,
      "step": 10751
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0951682329177856,
      "learning_rate": 2.4368878024743638e-06,
      "loss": 0.4804,
      "step": 10752
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0911171436309814,
      "learning_rate": 2.4360069237638114e-06,
      "loss": 0.522,
      "step": 10753
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0631780624389648,
      "learning_rate": 2.43512615301409e-06,
      "loss": 0.4894,
      "step": 10754
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0774295330047607,
      "learning_rate": 2.434245490262282e-06,
      "loss": 0.5196,
      "step": 10755
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0458868741989136,
      "learning_rate": 2.4333649355454704e-06,
      "loss": 0.528,
      "step": 10756
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0356422662734985,
      "learning_rate": 2.4324844889007328e-06,
      "loss": 0.5465,
      "step": 10757
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0376536846160889,
      "learning_rate": 2.4316041503651417e-06,
      "loss": 0.4611,
      "step": 10758
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1252613067626953,
      "learning_rate": 2.430723919975767e-06,
      "loss": 0.5091,
      "step": 10759
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0624945163726807,
      "learning_rate": 2.4298437977696658e-06,
      "loss": 0.5268,
      "step": 10760
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0497515201568604,
      "learning_rate": 2.428963783783904e-06,
      "loss": 0.4852,
      "step": 10761
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0248337984085083,
      "learning_rate": 2.4280838780555347e-06,
      "loss": 0.5007,
      "step": 10762
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0751545429229736,
      "learning_rate": 2.427204080621605e-06,
      "loss": 0.5021,
      "step": 10763
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0481481552124023,
      "learning_rate": 2.426324391519161e-06,
      "loss": 0.4454,
      "step": 10764
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0749013423919678,
      "learning_rate": 2.4254448107852434e-06,
      "loss": 0.5357,
      "step": 10765
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.035150170326233,
      "learning_rate": 2.424565338456889e-06,
      "loss": 0.5128,
      "step": 10766
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0497207641601562,
      "learning_rate": 2.4236859745711305e-06,
      "loss": 0.5407,
      "step": 10767
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0844335556030273,
      "learning_rate": 2.4228067191649917e-06,
      "loss": 0.5075,
      "step": 10768
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9757227301597595,
      "learning_rate": 2.421927572275494e-06,
      "loss": 0.5017,
      "step": 10769
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0580272674560547,
      "learning_rate": 2.4210485339396627e-06,
      "loss": 0.5002,
      "step": 10770
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.091774821281433,
      "learning_rate": 2.4201696041945033e-06,
      "loss": 0.5408,
      "step": 10771
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0345991849899292,
      "learning_rate": 2.419290783077028e-06,
      "loss": 0.4692,
      "step": 10772
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.04410982131958,
      "learning_rate": 2.41841207062424e-06,
      "loss": 0.4801,
      "step": 10773
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0724915266036987,
      "learning_rate": 2.4175334668731383e-06,
      "loss": 0.539,
      "step": 10774
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9957897067070007,
      "learning_rate": 2.416654971860721e-06,
      "loss": 0.5044,
      "step": 10775
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0342212915420532,
      "learning_rate": 2.415776585623974e-06,
      "loss": 0.5185,
      "step": 10776
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1434073448181152,
      "learning_rate": 2.4148983081998834e-06,
      "loss": 0.5179,
      "step": 10777
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.029054045677185,
      "learning_rate": 2.414020139625436e-06,
      "loss": 0.4609,
      "step": 10778
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0691587924957275,
      "learning_rate": 2.413142079937602e-06,
      "loss": 0.5596,
      "step": 10779
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.087937593460083,
      "learning_rate": 2.4122641291733567e-06,
      "loss": 0.5135,
      "step": 10780
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.101758599281311,
      "learning_rate": 2.4113862873696687e-06,
      "loss": 0.5433,
      "step": 10781
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0143336057662964,
      "learning_rate": 2.410508554563495e-06,
      "loss": 0.5028,
      "step": 10782
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.081833004951477,
      "learning_rate": 2.4096309307918013e-06,
      "loss": 0.5302,
      "step": 10783
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1051571369171143,
      "learning_rate": 2.4087534160915364e-06,
      "loss": 0.5296,
      "step": 10784
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9907381534576416,
      "learning_rate": 2.407876010499651e-06,
      "loss": 0.5312,
      "step": 10785
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1027705669403076,
      "learning_rate": 2.4069987140530893e-06,
      "loss": 0.5136,
      "step": 10786
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.004507064819336,
      "learning_rate": 2.4061215267887915e-06,
      "loss": 0.5009,
      "step": 10787
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0545748472213745,
      "learning_rate": 2.4052444487436925e-06,
      "loss": 0.4824,
      "step": 10788
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1453453302383423,
      "learning_rate": 2.4043674799547252e-06,
      "loss": 0.484,
      "step": 10789
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0942673683166504,
      "learning_rate": 2.4034906204588104e-06,
      "loss": 0.5205,
      "step": 10790
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0242911577224731,
      "learning_rate": 2.4026138702928763e-06,
      "loss": 0.4761,
      "step": 10791
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0240156650543213,
      "learning_rate": 2.4017372294938347e-06,
      "loss": 0.5277,
      "step": 10792
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0593849420547485,
      "learning_rate": 2.4008606980985994e-06,
      "loss": 0.5373,
      "step": 10793
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.114747166633606,
      "learning_rate": 2.399984276144079e-06,
      "loss": 0.528,
      "step": 10794
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0083765983581543,
      "learning_rate": 2.3991079636671755e-06,
      "loss": 0.528,
      "step": 10795
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0297280550003052,
      "learning_rate": 2.398231760704788e-06,
      "loss": 0.5163,
      "step": 10796
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0804991722106934,
      "learning_rate": 2.397355667293812e-06,
      "loss": 0.4925,
      "step": 10797
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0716068744659424,
      "learning_rate": 2.396479683471133e-06,
      "loss": 0.533,
      "step": 10798
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0013996362686157,
      "learning_rate": 2.395603809273635e-06,
      "loss": 0.5476,
      "step": 10799
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.211030125617981,
      "learning_rate": 2.3947280447382055e-06,
      "loss": 0.5368,
      "step": 10800
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0420416593551636,
      "learning_rate": 2.3938523899017124e-06,
      "loss": 0.4871,
      "step": 10801
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0371475219726562,
      "learning_rate": 2.392976844801029e-06,
      "loss": 0.5074,
      "step": 10802
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9832603335380554,
      "learning_rate": 2.3921014094730216e-06,
      "loss": 0.5094,
      "step": 10803
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.002054214477539,
      "learning_rate": 2.3912260839545514e-06,
      "loss": 0.5261,
      "step": 10804
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0066535472869873,
      "learning_rate": 2.390350868282478e-06,
      "loss": 0.5592,
      "step": 10805
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1311171054840088,
      "learning_rate": 2.389475762493649e-06,
      "loss": 0.5044,
      "step": 10806
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9754476547241211,
      "learning_rate": 2.3886007666249124e-06,
      "loss": 0.4875,
      "step": 10807
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0370256900787354,
      "learning_rate": 2.387725880713117e-06,
      "loss": 0.5078,
      "step": 10808
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1595977544784546,
      "learning_rate": 2.3868511047950955e-06,
      "loss": 0.5417,
      "step": 10809
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.078356146812439,
      "learning_rate": 2.3859764389076834e-06,
      "loss": 0.4764,
      "step": 10810
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0195378065109253,
      "learning_rate": 2.3851018830877115e-06,
      "loss": 0.4784,
      "step": 10811
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.075260877609253,
      "learning_rate": 2.3842274373719994e-06,
      "loss": 0.5229,
      "step": 10812
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0607411861419678,
      "learning_rate": 2.383353101797374e-06,
      "loss": 0.4833,
      "step": 10813
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.00974440574646,
      "learning_rate": 2.3824788764006446e-06,
      "loss": 0.526,
      "step": 10814
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9402571320533752,
      "learning_rate": 2.3816047612186243e-06,
      "loss": 0.4657,
      "step": 10815
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0360361337661743,
      "learning_rate": 2.3807307562881188e-06,
      "loss": 0.5157,
      "step": 10816
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0566238164901733,
      "learning_rate": 2.3798568616459295e-06,
      "loss": 0.5376,
      "step": 10817
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0949710607528687,
      "learning_rate": 2.378983077328853e-06,
      "loss": 0.4918,
      "step": 10818
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9928608536720276,
      "learning_rate": 2.378109403373683e-06,
      "loss": 0.4995,
      "step": 10819
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0890331268310547,
      "learning_rate": 2.3772358398172013e-06,
      "loss": 0.5446,
      "step": 10820
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9923766851425171,
      "learning_rate": 2.3763623866961984e-06,
      "loss": 0.4591,
      "step": 10821
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0734655857086182,
      "learning_rate": 2.375489044047446e-06,
      "loss": 0.4931,
      "step": 10822
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0973488092422485,
      "learning_rate": 2.37461581190772e-06,
      "loss": 0.549,
      "step": 10823
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0098018646240234,
      "learning_rate": 2.37374269031379e-06,
      "loss": 0.4981,
      "step": 10824
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9769476652145386,
      "learning_rate": 2.3728696793024187e-06,
      "loss": 0.478,
      "step": 10825
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1181920766830444,
      "learning_rate": 2.371996778910366e-06,
      "loss": 0.528,
      "step": 10826
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.112295389175415,
      "learning_rate": 2.3711239891743886e-06,
      "loss": 0.5127,
      "step": 10827
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9636074900627136,
      "learning_rate": 2.370251310131233e-06,
      "loss": 0.4271,
      "step": 10828
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0492388010025024,
      "learning_rate": 2.369378741817647e-06,
      "loss": 0.5006,
      "step": 10829
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.027235984802246,
      "learning_rate": 2.3685062842703697e-06,
      "loss": 0.5129,
      "step": 10830
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1333270072937012,
      "learning_rate": 2.3676339375261394e-06,
      "loss": 0.5233,
      "step": 10831
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0313310623168945,
      "learning_rate": 2.3667617016216885e-06,
      "loss": 0.4789,
      "step": 10832
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0129145383834839,
      "learning_rate": 2.365889576593738e-06,
      "loss": 0.4626,
      "step": 10833
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0947465896606445,
      "learning_rate": 2.365017562479016e-06,
      "loss": 0.5032,
      "step": 10834
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0129739046096802,
      "learning_rate": 2.36414565931424e-06,
      "loss": 0.4713,
      "step": 10835
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.03473699092865,
      "learning_rate": 2.3632738671361187e-06,
      "loss": 0.4923,
      "step": 10836
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.103641152381897,
      "learning_rate": 2.362402185981363e-06,
      "loss": 0.4954,
      "step": 10837
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.080424189567566,
      "learning_rate": 2.3615306158866745e-06,
      "loss": 0.5467,
      "step": 10838
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1474509239196777,
      "learning_rate": 2.360659156888754e-06,
      "loss": 0.5577,
      "step": 10839
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9592921137809753,
      "learning_rate": 2.359787809024297e-06,
      "loss": 0.4802,
      "step": 10840
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0543572902679443,
      "learning_rate": 2.358916572329986e-06,
      "loss": 0.4925,
      "step": 10841
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9967743754386902,
      "learning_rate": 2.3580454468425136e-06,
      "loss": 0.4902,
      "step": 10842
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.053407907485962,
      "learning_rate": 2.357174432598558e-06,
      "loss": 0.5122,
      "step": 10843
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0734866857528687,
      "learning_rate": 2.356303529634791e-06,
      "loss": 0.5312,
      "step": 10844
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0064641237258911,
      "learning_rate": 2.355432737987886e-06,
      "loss": 0.538,
      "step": 10845
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.090053915977478,
      "learning_rate": 2.3545620576945088e-06,
      "loss": 0.5243,
      "step": 10846
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0361047983169556,
      "learning_rate": 2.3536914887913203e-06,
      "loss": 0.5288,
      "step": 10847
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.083053708076477,
      "learning_rate": 2.3528210313149793e-06,
      "loss": 0.4931,
      "step": 10848
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.089536428451538,
      "learning_rate": 2.351950685302134e-06,
      "loss": 0.5345,
      "step": 10849
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.165738582611084,
      "learning_rate": 2.351080450789431e-06,
      "loss": 0.5085,
      "step": 10850
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0671491622924805,
      "learning_rate": 2.3502103278135203e-06,
      "loss": 0.4966,
      "step": 10851
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0162256956100464,
      "learning_rate": 2.349340316411032e-06,
      "loss": 0.4889,
      "step": 10852
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0168367624282837,
      "learning_rate": 2.3484704166186024e-06,
      "loss": 0.4913,
      "step": 10853
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0524020195007324,
      "learning_rate": 2.347600628472859e-06,
      "loss": 0.4952,
      "step": 10854
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.012343406677246,
      "learning_rate": 2.3467309520104265e-06,
      "loss": 0.4806,
      "step": 10855
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1498016119003296,
      "learning_rate": 2.3458613872679255e-06,
      "loss": 0.5259,
      "step": 10856
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0634065866470337,
      "learning_rate": 2.344991934281966e-06,
      "loss": 0.4772,
      "step": 10857
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.207221269607544,
      "learning_rate": 2.344122593089161e-06,
      "loss": 0.5228,
      "step": 10858
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9218490719795227,
      "learning_rate": 2.3432533637261135e-06,
      "loss": 0.446,
      "step": 10859
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.06690514087677,
      "learning_rate": 2.3423842462294257e-06,
      "loss": 0.5001,
      "step": 10860
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1247037649154663,
      "learning_rate": 2.341515240635691e-06,
      "loss": 0.4739,
      "step": 10861
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0756078958511353,
      "learning_rate": 2.340646346981504e-06,
      "loss": 0.5022,
      "step": 10862
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.073510766029358,
      "learning_rate": 2.339777565303444e-06,
      "loss": 0.5338,
      "step": 10863
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0145232677459717,
      "learning_rate": 2.3389088956380982e-06,
      "loss": 0.4355,
      "step": 10864
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.994920015335083,
      "learning_rate": 2.338040338022044e-06,
      "loss": 0.4767,
      "step": 10865
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1112754344940186,
      "learning_rate": 2.3371718924918487e-06,
      "loss": 0.4915,
      "step": 10866
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0785366296768188,
      "learning_rate": 2.3363035590840814e-06,
      "loss": 0.4844,
      "step": 10867
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0066707134246826,
      "learning_rate": 2.3354353378353056e-06,
      "loss": 0.5253,
      "step": 10868
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9772728681564331,
      "learning_rate": 2.334567228782078e-06,
      "loss": 0.4908,
      "step": 10869
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1539030075073242,
      "learning_rate": 2.3336992319609534e-06,
      "loss": 0.5205,
      "step": 10870
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0286020040512085,
      "learning_rate": 2.3328313474084755e-06,
      "loss": 0.5318,
      "step": 10871
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0363796949386597,
      "learning_rate": 2.3319635751611937e-06,
      "loss": 0.5115,
      "step": 10872
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0309985876083374,
      "learning_rate": 2.3310959152556453e-06,
      "loss": 0.55,
      "step": 10873
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0882515907287598,
      "learning_rate": 2.3302283677283618e-06,
      "loss": 0.5182,
      "step": 10874
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0382540225982666,
      "learning_rate": 2.3293609326158745e-06,
      "loss": 0.4654,
      "step": 10875
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0700305700302124,
      "learning_rate": 2.328493609954707e-06,
      "loss": 0.5194,
      "step": 10876
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0921525955200195,
      "learning_rate": 2.3276263997813812e-06,
      "loss": 0.4882,
      "step": 10877
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1156283617019653,
      "learning_rate": 2.3267593021324127e-06,
      "loss": 0.5642,
      "step": 10878
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1570100784301758,
      "learning_rate": 2.3258923170443087e-06,
      "loss": 0.5761,
      "step": 10879
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9395455121994019,
      "learning_rate": 2.3250254445535743e-06,
      "loss": 0.4777,
      "step": 10880
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0273380279541016,
      "learning_rate": 2.324158684696717e-06,
      "loss": 0.4392,
      "step": 10881
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.960543692111969,
      "learning_rate": 2.323292037510227e-06,
      "loss": 0.4258,
      "step": 10882
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1122788190841675,
      "learning_rate": 2.3224255030305977e-06,
      "loss": 0.5128,
      "step": 10883
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9551585912704468,
      "learning_rate": 2.321559081294316e-06,
      "loss": 0.4742,
      "step": 10884
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0875403881072998,
      "learning_rate": 2.3206927723378638e-06,
      "loss": 0.5332,
      "step": 10885
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0719062089920044,
      "learning_rate": 2.3198265761977196e-06,
      "loss": 0.5218,
      "step": 10886
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0475523471832275,
      "learning_rate": 2.3189604929103533e-06,
      "loss": 0.4952,
      "step": 10887
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0663193464279175,
      "learning_rate": 2.318094522512232e-06,
      "loss": 0.5228,
      "step": 10888
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.953425407409668,
      "learning_rate": 2.3172286650398247e-06,
      "loss": 0.4936,
      "step": 10889
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0176100730895996,
      "learning_rate": 2.3163629205295833e-06,
      "loss": 0.4967,
      "step": 10890
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0192761421203613,
      "learning_rate": 2.3154972890179638e-06,
      "loss": 0.4981,
      "step": 10891
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.119706153869629,
      "learning_rate": 2.3146317705414168e-06,
      "loss": 0.4856,
      "step": 10892
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0019359588623047,
      "learning_rate": 2.31376636513638e-06,
      "loss": 0.5136,
      "step": 10893
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0564658641815186,
      "learning_rate": 2.3129010728393012e-06,
      "loss": 0.5581,
      "step": 10894
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0997692346572876,
      "learning_rate": 2.3120358936866084e-06,
      "loss": 0.542,
      "step": 10895
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0283379554748535,
      "learning_rate": 2.3111708277147333e-06,
      "loss": 0.474,
      "step": 10896
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.082440972328186,
      "learning_rate": 2.310305874960101e-06,
      "loss": 0.5198,
      "step": 10897
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0326625108718872,
      "learning_rate": 2.3094410354591314e-06,
      "loss": 0.5633,
      "step": 10898
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0904386043548584,
      "learning_rate": 2.30857630924824e-06,
      "loss": 0.5055,
      "step": 10899
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9982689619064331,
      "learning_rate": 2.3077116963638396e-06,
      "loss": 0.5241,
      "step": 10900
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.984889030456543,
      "learning_rate": 2.3068471968423296e-06,
      "loss": 0.5254,
      "step": 10901
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.110054612159729,
      "learning_rate": 2.305982810720119e-06,
      "loss": 0.5533,
      "step": 10902
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.027411937713623,
      "learning_rate": 2.3051185380335995e-06,
      "loss": 0.488,
      "step": 10903
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9612981081008911,
      "learning_rate": 2.304254378819163e-06,
      "loss": 0.4934,
      "step": 10904
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0585256814956665,
      "learning_rate": 2.3033903331131986e-06,
      "loss": 0.5339,
      "step": 10905
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0645774602890015,
      "learning_rate": 2.3025264009520833e-06,
      "loss": 0.5331,
      "step": 10906
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0150748491287231,
      "learning_rate": 2.3016625823721985e-06,
      "loss": 0.5182,
      "step": 10907
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.169596552848816,
      "learning_rate": 2.300798877409918e-06,
      "loss": 0.5295,
      "step": 10908
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.067642331123352,
      "learning_rate": 2.2999352861016042e-06,
      "loss": 0.5006,
      "step": 10909
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0091102123260498,
      "learning_rate": 2.299071808483623e-06,
      "loss": 0.4797,
      "step": 10910
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0823348760604858,
      "learning_rate": 2.2982084445923327e-06,
      "loss": 0.5481,
      "step": 10911
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9775456190109253,
      "learning_rate": 2.297345194464086e-06,
      "loss": 0.5006,
      "step": 10912
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0186797380447388,
      "learning_rate": 2.2964820581352325e-06,
      "loss": 0.5257,
      "step": 10913
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9842431545257568,
      "learning_rate": 2.295619035642111e-06,
      "loss": 0.4726,
      "step": 10914
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.073970079421997,
      "learning_rate": 2.294756127021066e-06,
      "loss": 0.502,
      "step": 10915
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1141387224197388,
      "learning_rate": 2.2938933323084315e-06,
      "loss": 0.5829,
      "step": 10916
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0280675888061523,
      "learning_rate": 2.293030651540534e-06,
      "loss": 0.5358,
      "step": 10917
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0070101022720337,
      "learning_rate": 2.2921680847536976e-06,
      "loss": 0.4939,
      "step": 10918
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1027600765228271,
      "learning_rate": 2.2913056319842436e-06,
      "loss": 0.5214,
      "step": 10919
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9915065765380859,
      "learning_rate": 2.2904432932684865e-06,
      "loss": 0.5011,
      "step": 10920
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.006392002105713,
      "learning_rate": 2.289581068642737e-06,
      "loss": 0.4891,
      "step": 10921
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1884076595306396,
      "learning_rate": 2.2887189581433016e-06,
      "loss": 0.4954,
      "step": 10922
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1301230192184448,
      "learning_rate": 2.287856961806475e-06,
      "loss": 0.5255,
      "step": 10923
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.010464072227478,
      "learning_rate": 2.286995079668561e-06,
      "loss": 0.4895,
      "step": 10924
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0081626176834106,
      "learning_rate": 2.2861333117658442e-06,
      "loss": 0.4892,
      "step": 10925
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9857581853866577,
      "learning_rate": 2.2852716581346124e-06,
      "loss": 0.5362,
      "step": 10926
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.931063175201416,
      "learning_rate": 2.2844101188111477e-06,
      "loss": 0.5071,
      "step": 10927
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.031646728515625,
      "learning_rate": 2.283548693831726e-06,
      "loss": 0.4879,
      "step": 10928
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.063567042350769,
      "learning_rate": 2.2826873832326192e-06,
      "loss": 0.5253,
      "step": 10929
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0719285011291504,
      "learning_rate": 2.2818261870500954e-06,
      "loss": 0.4729,
      "step": 10930
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0247489213943481,
      "learning_rate": 2.280965105320411e-06,
      "loss": 0.4991,
      "step": 10931
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.008853793144226,
      "learning_rate": 2.280104138079831e-06,
      "loss": 0.5526,
      "step": 10932
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9935298562049866,
      "learning_rate": 2.2792432853646023e-06,
      "loss": 0.5016,
      "step": 10933
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0684596300125122,
      "learning_rate": 2.2783825472109743e-06,
      "loss": 0.4832,
      "step": 10934
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.020822525024414,
      "learning_rate": 2.277521923655189e-06,
      "loss": 0.5226,
      "step": 10935
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.047221302986145,
      "learning_rate": 2.276661414733485e-06,
      "loss": 0.51,
      "step": 10936
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0854012966156006,
      "learning_rate": 2.2758010204820945e-06,
      "loss": 0.5334,
      "step": 10937
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0282018184661865,
      "learning_rate": 2.2749407409372487e-06,
      "loss": 0.5203,
      "step": 10938
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0249234437942505,
      "learning_rate": 2.2740805761351664e-06,
      "loss": 0.4742,
      "step": 10939
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.078487515449524,
      "learning_rate": 2.273220526112068e-06,
      "loss": 0.5067,
      "step": 10940
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.024756669998169,
      "learning_rate": 2.272360590904168e-06,
      "loss": 0.5095,
      "step": 10941
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0135009288787842,
      "learning_rate": 2.2715007705476744e-06,
      "loss": 0.5056,
      "step": 10942
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0788309574127197,
      "learning_rate": 2.2706410650787937e-06,
      "loss": 0.5284,
      "step": 10943
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9614673852920532,
      "learning_rate": 2.2697814745337186e-06,
      "loss": 0.4832,
      "step": 10944
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9808701872825623,
      "learning_rate": 2.2689219989486506e-06,
      "loss": 0.5214,
      "step": 10945
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.163304328918457,
      "learning_rate": 2.2680626383597782e-06,
      "loss": 0.5577,
      "step": 10946
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9920132756233215,
      "learning_rate": 2.267203392803282e-06,
      "loss": 0.477,
      "step": 10947
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0092310905456543,
      "learning_rate": 2.266344262315345e-06,
      "loss": 0.5237,
      "step": 10948
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.155785083770752,
      "learning_rate": 2.2654852469321405e-06,
      "loss": 0.5016,
      "step": 10949
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1730620861053467,
      "learning_rate": 2.26462634668984e-06,
      "loss": 0.5936,
      "step": 10950
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9889048337936401,
      "learning_rate": 2.2637675616246103e-06,
      "loss": 0.5242,
      "step": 10951
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0237679481506348,
      "learning_rate": 2.262908891772608e-06,
      "loss": 0.5823,
      "step": 10952
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0033601522445679,
      "learning_rate": 2.2620503371699886e-06,
      "loss": 0.5212,
      "step": 10953
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0933045148849487,
      "learning_rate": 2.261191897852909e-06,
      "loss": 0.4784,
      "step": 10954
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0246044397354126,
      "learning_rate": 2.260333573857509e-06,
      "loss": 0.5272,
      "step": 10955
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0688954591751099,
      "learning_rate": 2.2594753652199313e-06,
      "loss": 0.5285,
      "step": 10956
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0384154319763184,
      "learning_rate": 2.2586172719763126e-06,
      "loss": 0.487,
      "step": 10957
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0645639896392822,
      "learning_rate": 2.2577592941627842e-06,
      "loss": 0.5067,
      "step": 10958
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0228403806686401,
      "learning_rate": 2.2569014318154735e-06,
      "loss": 0.498,
      "step": 10959
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2611366510391235,
      "learning_rate": 2.2560436849704996e-06,
      "loss": 0.5444,
      "step": 10960
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0925045013427734,
      "learning_rate": 2.255186053663979e-06,
      "loss": 0.5394,
      "step": 10961
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9899227023124695,
      "learning_rate": 2.2543285379320283e-06,
      "loss": 0.4803,
      "step": 10962
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0411585569381714,
      "learning_rate": 2.2534711378107498e-06,
      "loss": 0.5004,
      "step": 10963
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0692851543426514,
      "learning_rate": 2.2526138533362475e-06,
      "loss": 0.4789,
      "step": 10964
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.170039415359497,
      "learning_rate": 2.2517566845446182e-06,
      "loss": 0.5275,
      "step": 10965
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0667608976364136,
      "learning_rate": 2.2508996314719544e-06,
      "loss": 0.4947,
      "step": 10966
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0293246507644653,
      "learning_rate": 2.250042694154345e-06,
      "loss": 0.4829,
      "step": 10967
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0149614810943604,
      "learning_rate": 2.2491858726278704e-06,
      "loss": 0.4615,
      "step": 10968
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.046410322189331,
      "learning_rate": 2.248329166928609e-06,
      "loss": 0.5281,
      "step": 10969
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2291126251220703,
      "learning_rate": 2.2474725770926337e-06,
      "loss": 0.4785,
      "step": 10970
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1178964376449585,
      "learning_rate": 2.2466161031560136e-06,
      "loss": 0.5346,
      "step": 10971
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0019038915634155,
      "learning_rate": 2.2457597451548102e-06,
      "loss": 0.4555,
      "step": 10972
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0246751308441162,
      "learning_rate": 2.2449035031250847e-06,
      "loss": 0.5029,
      "step": 10973
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0310834646224976,
      "learning_rate": 2.2440473771028855e-06,
      "loss": 0.5217,
      "step": 10974
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0542534589767456,
      "learning_rate": 2.2431913671242666e-06,
      "loss": 0.5378,
      "step": 10975
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0208415985107422,
      "learning_rate": 2.242335473225268e-06,
      "loss": 0.4924,
      "step": 10976
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0091476440429688,
      "learning_rate": 2.2414796954419286e-06,
      "loss": 0.49,
      "step": 10977
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9918937683105469,
      "learning_rate": 2.2406240338102836e-06,
      "loss": 0.4849,
      "step": 10978
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0851179361343384,
      "learning_rate": 2.239768488366361e-06,
      "loss": 0.5236,
      "step": 10979
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0405689477920532,
      "learning_rate": 2.2389130591461855e-06,
      "loss": 0.5159,
      "step": 10980
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0542181730270386,
      "learning_rate": 2.2380577461857777e-06,
      "loss": 0.5225,
      "step": 10981
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0488357543945312,
      "learning_rate": 2.2372025495211465e-06,
      "loss": 0.5331,
      "step": 10982
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0630464553833008,
      "learning_rate": 2.236347469188308e-06,
      "loss": 0.5343,
      "step": 10983
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0251017808914185,
      "learning_rate": 2.2354925052232625e-06,
      "loss": 0.4954,
      "step": 10984
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.037653923034668,
      "learning_rate": 2.2346376576620103e-06,
      "loss": 0.4754,
      "step": 10985
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1310943365097046,
      "learning_rate": 2.2337829265405466e-06,
      "loss": 0.5137,
      "step": 10986
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0841907262802124,
      "learning_rate": 2.2329283118948604e-06,
      "loss": 0.5063,
      "step": 10987
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9623340964317322,
      "learning_rate": 2.232073813760937e-06,
      "loss": 0.5351,
      "step": 10988
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.03236722946167,
      "learning_rate": 2.2312194321747582e-06,
      "loss": 0.4978,
      "step": 10989
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.069541096687317,
      "learning_rate": 2.230365167172296e-06,
      "loss": 0.5143,
      "step": 10990
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1878217458724976,
      "learning_rate": 2.2295110187895215e-06,
      "loss": 0.4877,
      "step": 10991
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.984520435333252,
      "learning_rate": 2.2286569870624e-06,
      "loss": 0.5278,
      "step": 10992
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0635064840316772,
      "learning_rate": 2.227803072026892e-06,
      "loss": 0.4926,
      "step": 10993
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.092267632484436,
      "learning_rate": 2.226949273718953e-06,
      "loss": 0.5379,
      "step": 10994
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0395183563232422,
      "learning_rate": 2.226095592174533e-06,
      "loss": 0.5277,
      "step": 10995
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9301552176475525,
      "learning_rate": 2.2252420274295782e-06,
      "loss": 0.4542,
      "step": 10996
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9266456961631775,
      "learning_rate": 2.224388579520031e-06,
      "loss": 0.4951,
      "step": 10997
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0810495615005493,
      "learning_rate": 2.2235352484818228e-06,
      "loss": 0.5257,
      "step": 10998
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0020378828048706,
      "learning_rate": 2.222682034350887e-06,
      "loss": 0.5075,
      "step": 10999
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0670814514160156,
      "learning_rate": 2.221828937163149e-06,
      "loss": 0.5337,
      "step": 11000
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0369603633880615,
      "learning_rate": 2.22097595695453e-06,
      "loss": 0.4782,
      "step": 11001
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.049034833908081,
      "learning_rate": 2.220123093760946e-06,
      "loss": 0.5295,
      "step": 11002
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1661826372146606,
      "learning_rate": 2.2192703476183093e-06,
      "loss": 0.4902,
      "step": 11003
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1276367902755737,
      "learning_rate": 2.2184177185625217e-06,
      "loss": 0.5505,
      "step": 11004
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0560840368270874,
      "learning_rate": 2.217565206629491e-06,
      "loss": 0.498,
      "step": 11005
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0290144681930542,
      "learning_rate": 2.2167128118551084e-06,
      "loss": 0.5105,
      "step": 11006
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0535063743591309,
      "learning_rate": 2.2158605342752667e-06,
      "loss": 0.5385,
      "step": 11007
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0331565141677856,
      "learning_rate": 2.2150083739258525e-06,
      "loss": 0.4536,
      "step": 11008
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9949620962142944,
      "learning_rate": 2.214156330842748e-06,
      "loss": 0.4917,
      "step": 11009
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9716442227363586,
      "learning_rate": 2.2133044050618286e-06,
      "loss": 0.5163,
      "step": 11010
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0355647802352905,
      "learning_rate": 2.2124525966189685e-06,
      "loss": 0.4759,
      "step": 11011
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0520069599151611,
      "learning_rate": 2.211600905550029e-06,
      "loss": 0.4828,
      "step": 11012
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0424782037734985,
      "learning_rate": 2.2107493318908785e-06,
      "loss": 0.5187,
      "step": 11013
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9450308084487915,
      "learning_rate": 2.2098978756773687e-06,
      "loss": 0.4999,
      "step": 11014
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1031875610351562,
      "learning_rate": 2.2090465369453533e-06,
      "loss": 0.5441,
      "step": 11015
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9856180548667908,
      "learning_rate": 2.208195315730681e-06,
      "loss": 0.5022,
      "step": 11016
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1115809679031372,
      "learning_rate": 2.207344212069189e-06,
      "loss": 0.5499,
      "step": 11017
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0844727754592896,
      "learning_rate": 2.2064932259967188e-06,
      "loss": 0.5283,
      "step": 11018
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0206760168075562,
      "learning_rate": 2.2056423575491026e-06,
      "loss": 0.538,
      "step": 11019
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0844743251800537,
      "learning_rate": 2.204791606762164e-06,
      "loss": 0.5307,
      "step": 11020
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.080522894859314,
      "learning_rate": 2.2039409736717273e-06,
      "loss": 0.5078,
      "step": 11021
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9728801250457764,
      "learning_rate": 2.2030904583136085e-06,
      "loss": 0.5093,
      "step": 11022
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0371694564819336,
      "learning_rate": 2.2022400607236214e-06,
      "loss": 0.5241,
      "step": 11023
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.995800793170929,
      "learning_rate": 2.2013897809375753e-06,
      "loss": 0.5391,
      "step": 11024
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.137416124343872,
      "learning_rate": 2.2005396189912647e-06,
      "loss": 0.4828,
      "step": 11025
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9752306938171387,
      "learning_rate": 2.199689574920495e-06,
      "loss": 0.4923,
      "step": 11026
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0050386190414429,
      "learning_rate": 2.198839648761057e-06,
      "loss": 0.5086,
      "step": 11027
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9973408579826355,
      "learning_rate": 2.1979898405487354e-06,
      "loss": 0.4993,
      "step": 11028
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0028589963912964,
      "learning_rate": 2.197140150319314e-06,
      "loss": 0.4895,
      "step": 11029
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0483782291412354,
      "learning_rate": 2.19629057810857e-06,
      "loss": 0.5351,
      "step": 11030
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0445411205291748,
      "learning_rate": 2.195441123952277e-06,
      "loss": 0.5307,
      "step": 11031
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0868538618087769,
      "learning_rate": 2.1945917878862037e-06,
      "loss": 0.4981,
      "step": 11032
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.014582872390747,
      "learning_rate": 2.193742569946109e-06,
      "loss": 0.5276,
      "step": 11033
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9631657004356384,
      "learning_rate": 2.1928934701677507e-06,
      "loss": 0.4964,
      "step": 11034
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0068409442901611,
      "learning_rate": 2.1920444885868862e-06,
      "loss": 0.4928,
      "step": 11035
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0212610960006714,
      "learning_rate": 2.1911956252392593e-06,
      "loss": 0.4969,
      "step": 11036
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0570365190505981,
      "learning_rate": 2.1903468801606125e-06,
      "loss": 0.5042,
      "step": 11037
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0584505796432495,
      "learning_rate": 2.1894982533866852e-06,
      "loss": 0.5255,
      "step": 11038
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0139025449752808,
      "learning_rate": 2.188649744953209e-06,
      "loss": 0.5084,
      "step": 11039
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0854088068008423,
      "learning_rate": 2.1878013548959145e-06,
      "loss": 0.5195,
      "step": 11040
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0749729871749878,
      "learning_rate": 2.186953083250519e-06,
      "loss": 0.5583,
      "step": 11041
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0553749799728394,
      "learning_rate": 2.1861049300527426e-06,
      "loss": 0.5038,
      "step": 11042
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1561241149902344,
      "learning_rate": 2.1852568953383025e-06,
      "loss": 0.5258,
      "step": 11043
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0235668420791626,
      "learning_rate": 2.1844089791429002e-06,
      "loss": 0.4742,
      "step": 11044
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0335506200790405,
      "learning_rate": 2.1835611815022412e-06,
      "loss": 0.5198,
      "step": 11045
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0722906589508057,
      "learning_rate": 2.182713502452025e-06,
      "loss": 0.4756,
      "step": 11046
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0140724182128906,
      "learning_rate": 2.181865942027939e-06,
      "loss": 0.524,
      "step": 11047
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1023540496826172,
      "learning_rate": 2.181018500265679e-06,
      "loss": 0.5404,
      "step": 11048
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9999150633811951,
      "learning_rate": 2.1801711772009203e-06,
      "loss": 0.4834,
      "step": 11049
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0206165313720703,
      "learning_rate": 2.179323972869345e-06,
      "loss": 0.5255,
      "step": 11050
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0228021144866943,
      "learning_rate": 2.1784768873066243e-06,
      "loss": 0.5006,
      "step": 11051
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.073330283164978,
      "learning_rate": 2.1776299205484265e-06,
      "loss": 0.481,
      "step": 11052
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0602368116378784,
      "learning_rate": 2.176783072630414e-06,
      "loss": 0.4962,
      "step": 11053
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0899503231048584,
      "learning_rate": 2.1759363435882475e-06,
      "loss": 0.5091,
      "step": 11054
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9894519448280334,
      "learning_rate": 2.1750897334575736e-06,
      "loss": 0.5073,
      "step": 11055
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.984174370765686,
      "learning_rate": 2.174243242274047e-06,
      "loss": 0.4952,
      "step": 11056
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1872106790542603,
      "learning_rate": 2.1733968700733066e-06,
      "loss": 0.5355,
      "step": 11057
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1067372560501099,
      "learning_rate": 2.1725506168909903e-06,
      "loss": 0.5262,
      "step": 11058
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0191781520843506,
      "learning_rate": 2.1717044827627314e-06,
      "loss": 0.4672,
      "step": 11059
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.087689995765686,
      "learning_rate": 2.1708584677241586e-06,
      "loss": 0.5683,
      "step": 11060
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0050077438354492,
      "learning_rate": 2.170012571810893e-06,
      "loss": 0.4704,
      "step": 11061
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0539138317108154,
      "learning_rate": 2.1691667950585552e-06,
      "loss": 0.4936,
      "step": 11062
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0413488149642944,
      "learning_rate": 2.1683211375027543e-06,
      "loss": 0.5392,
      "step": 11063
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0355830192565918,
      "learning_rate": 2.1674755991790976e-06,
      "loss": 0.5305,
      "step": 11064
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0300675630569458,
      "learning_rate": 2.1666301801231937e-06,
      "loss": 0.5073,
      "step": 11065
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0976272821426392,
      "learning_rate": 2.1657848803706344e-06,
      "loss": 0.5136,
      "step": 11066
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9987136125564575,
      "learning_rate": 2.1649396999570137e-06,
      "loss": 0.4915,
      "step": 11067
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9640436172485352,
      "learning_rate": 2.1640946389179207e-06,
      "loss": 0.4741,
      "step": 11068
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.044983148574829,
      "learning_rate": 2.1632496972889366e-06,
      "loss": 0.4902,
      "step": 11069
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0915316343307495,
      "learning_rate": 2.162404875105641e-06,
      "loss": 0.5721,
      "step": 11070
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0868504047393799,
      "learning_rate": 2.1615601724036033e-06,
      "loss": 0.5099,
      "step": 11071
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0221885442733765,
      "learning_rate": 2.1607155892183905e-06,
      "loss": 0.4969,
      "step": 11072
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9462901949882507,
      "learning_rate": 2.1598711255855713e-06,
      "loss": 0.4853,
      "step": 11073
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0767602920532227,
      "learning_rate": 2.1590267815406968e-06,
      "loss": 0.5303,
      "step": 11074
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1433262825012207,
      "learning_rate": 2.1581825571193216e-06,
      "loss": 0.5026,
      "step": 11075
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9912518858909607,
      "learning_rate": 2.1573384523569945e-06,
      "loss": 0.5184,
      "step": 11076
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1095296144485474,
      "learning_rate": 2.1564944672892524e-06,
      "loss": 0.4929,
      "step": 11077
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0347744226455688,
      "learning_rate": 2.1556506019516405e-06,
      "loss": 0.4954,
      "step": 11078
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9864851236343384,
      "learning_rate": 2.1548068563796855e-06,
      "loss": 0.4837,
      "step": 11079
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9941489100456238,
      "learning_rate": 2.1539632306089153e-06,
      "loss": 0.4784,
      "step": 11080
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0469934940338135,
      "learning_rate": 2.153119724674853e-06,
      "loss": 0.5082,
      "step": 11081
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0488927364349365,
      "learning_rate": 2.1522763386130156e-06,
      "loss": 0.5143,
      "step": 11082
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0068544149398804,
      "learning_rate": 2.1514330724589156e-06,
      "loss": 0.5018,
      "step": 11083
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.047257423400879,
      "learning_rate": 2.1505899262480607e-06,
      "loss": 0.5603,
      "step": 11084
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1192617416381836,
      "learning_rate": 2.149746900015948e-06,
      "loss": 0.5434,
      "step": 11085
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0133657455444336,
      "learning_rate": 2.148903993798082e-06,
      "loss": 0.4893,
      "step": 11086
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0449867248535156,
      "learning_rate": 2.148061207629949e-06,
      "loss": 0.5281,
      "step": 11087
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0773366689682007,
      "learning_rate": 2.1472185415470365e-06,
      "loss": 0.4871,
      "step": 11088
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1084142923355103,
      "learning_rate": 2.1463759955848277e-06,
      "loss": 0.4858,
      "step": 11089
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.022783875465393,
      "learning_rate": 2.1455335697787987e-06,
      "loss": 0.5446,
      "step": 11090
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0690456628799438,
      "learning_rate": 2.1446912641644206e-06,
      "loss": 0.5205,
      "step": 11091
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0429967641830444,
      "learning_rate": 2.1438490787771634e-06,
      "loss": 0.5059,
      "step": 11092
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.024294137954712,
      "learning_rate": 2.1430070136524826e-06,
      "loss": 0.4773,
      "step": 11093
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.056626796722412,
      "learning_rate": 2.1421650688258384e-06,
      "loss": 0.4814,
      "step": 11094
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.05207097530365,
      "learning_rate": 2.1413232443326813e-06,
      "loss": 0.5463,
      "step": 11095
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.137169599533081,
      "learning_rate": 2.140481540208458e-06,
      "loss": 0.5285,
      "step": 11096
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9908720254898071,
      "learning_rate": 2.1396399564886113e-06,
      "loss": 0.4938,
      "step": 11097
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0105817317962646,
      "learning_rate": 2.1387984932085714e-06,
      "loss": 0.4875,
      "step": 11098
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1183953285217285,
      "learning_rate": 2.1379571504037754e-06,
      "loss": 0.5309,
      "step": 11099
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.161294937133789,
      "learning_rate": 2.1371159281096497e-06,
      "loss": 0.4929,
      "step": 11100
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.05838143825531,
      "learning_rate": 2.1362748263616112e-06,
      "loss": 0.5582,
      "step": 11101
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0622502565383911,
      "learning_rate": 2.1354338451950774e-06,
      "loss": 0.4505,
      "step": 11102
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0257731676101685,
      "learning_rate": 2.1345929846454593e-06,
      "loss": 0.5022,
      "step": 11103
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1054282188415527,
      "learning_rate": 2.133752244748163e-06,
      "loss": 0.5233,
      "step": 11104
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0054208040237427,
      "learning_rate": 2.1329116255385902e-06,
      "loss": 0.4959,
      "step": 11105
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0392478704452515,
      "learning_rate": 2.132071127052131e-06,
      "loss": 0.5656,
      "step": 11106
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0151344537734985,
      "learning_rate": 2.1312307493241825e-06,
      "loss": 0.484,
      "step": 11107
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.995629608631134,
      "learning_rate": 2.1303904923901288e-06,
      "loss": 0.5096,
      "step": 11108
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0044054985046387,
      "learning_rate": 2.1295503562853466e-06,
      "loss": 0.5092,
      "step": 11109
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0939685106277466,
      "learning_rate": 2.1287103410452135e-06,
      "loss": 0.5067,
      "step": 11110
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.074385166168213,
      "learning_rate": 2.1278704467050996e-06,
      "loss": 0.5428,
      "step": 11111
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9988450407981873,
      "learning_rate": 2.1270306733003697e-06,
      "loss": 0.4826,
      "step": 11112
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.10425865650177,
      "learning_rate": 2.126191020866386e-06,
      "loss": 0.5378,
      "step": 11113
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0218243598937988,
      "learning_rate": 2.125351489438499e-06,
      "loss": 0.502,
      "step": 11114
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.021587610244751,
      "learning_rate": 2.124512079052059e-06,
      "loss": 0.4711,
      "step": 11115
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9821772575378418,
      "learning_rate": 2.123672789742416e-06,
      "loss": 0.513,
      "step": 11116
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0624462366104126,
      "learning_rate": 2.1228336215449036e-06,
      "loss": 0.5217,
      "step": 11117
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0033706426620483,
      "learning_rate": 2.1219945744948584e-06,
      "loss": 0.5026,
      "step": 11118
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0093096494674683,
      "learning_rate": 2.12115564862761e-06,
      "loss": 0.4732,
      "step": 11119
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0732694864273071,
      "learning_rate": 2.1203168439784828e-06,
      "loss": 0.5009,
      "step": 11120
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.037276029586792,
      "learning_rate": 2.119478160582797e-06,
      "loss": 0.5289,
      "step": 11121
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.017081379890442,
      "learning_rate": 2.1186395984758633e-06,
      "loss": 0.4603,
      "step": 11122
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0212736129760742,
      "learning_rate": 2.117801157692993e-06,
      "loss": 0.5097,
      "step": 11123
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.090965986251831,
      "learning_rate": 2.1169628382694894e-06,
      "loss": 0.5066,
      "step": 11124
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0188883543014526,
      "learning_rate": 2.1161246402406518e-06,
      "loss": 0.4921,
      "step": 11125
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0649949312210083,
      "learning_rate": 2.1152865636417723e-06,
      "loss": 0.4927,
      "step": 11126
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0409399271011353,
      "learning_rate": 2.114448608508143e-06,
      "loss": 0.4848,
      "step": 11127
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0882501602172852,
      "learning_rate": 2.113610774875041e-06,
      "loss": 0.4795,
      "step": 11128
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0318166017532349,
      "learning_rate": 2.1127730627777497e-06,
      "loss": 0.4837,
      "step": 11129
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.11487877368927,
      "learning_rate": 2.111935472251543e-06,
      "loss": 0.5097,
      "step": 11130
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.081142544746399,
      "learning_rate": 2.1110980033316846e-06,
      "loss": 0.505,
      "step": 11131
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.092267632484436,
      "learning_rate": 2.1102606560534393e-06,
      "loss": 0.5276,
      "step": 11132
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.138821005821228,
      "learning_rate": 2.1094234304520655e-06,
      "loss": 0.5229,
      "step": 11133
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0012593269348145,
      "learning_rate": 2.108586326562816e-06,
      "loss": 0.5231,
      "step": 11134
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0540804862976074,
      "learning_rate": 2.1077493444209385e-06,
      "loss": 0.543,
      "step": 11135
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1856242418289185,
      "learning_rate": 2.1069124840616717e-06,
      "loss": 0.5298,
      "step": 11136
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0341116189956665,
      "learning_rate": 2.1060757455202574e-06,
      "loss": 0.495,
      "step": 11137
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.091784954071045,
      "learning_rate": 2.1052391288319285e-06,
      "loss": 0.5274,
      "step": 11138
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.112533450126648,
      "learning_rate": 2.1044026340319075e-06,
      "loss": 0.4736,
      "step": 11139
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0528075695037842,
      "learning_rate": 2.1035662611554187e-06,
      "loss": 0.5098,
      "step": 11140
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0677040815353394,
      "learning_rate": 2.1027300102376787e-06,
      "loss": 0.5021,
      "step": 11141
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.055909514427185,
      "learning_rate": 2.101893881313899e-06,
      "loss": 0.5107,
      "step": 11142
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1071017980575562,
      "learning_rate": 2.1010578744192885e-06,
      "loss": 0.5404,
      "step": 11143
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.079486608505249,
      "learning_rate": 2.1002219895890435e-06,
      "loss": 0.5516,
      "step": 11144
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0691206455230713,
      "learning_rate": 2.099386226858362e-06,
      "loss": 0.518,
      "step": 11145
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9732725620269775,
      "learning_rate": 2.098550586262439e-06,
      "loss": 0.5213,
      "step": 11146
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.065765380859375,
      "learning_rate": 2.097715067836456e-06,
      "loss": 0.4892,
      "step": 11147
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1149744987487793,
      "learning_rate": 2.096879671615595e-06,
      "loss": 0.5408,
      "step": 11148
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0927714109420776,
      "learning_rate": 2.0960443976350315e-06,
      "loss": 0.5461,
      "step": 11149
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0623589754104614,
      "learning_rate": 2.0952092459299366e-06,
      "loss": 0.5559,
      "step": 11150
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0234901905059814,
      "learning_rate": 2.0943742165354776e-06,
      "loss": 0.484,
      "step": 11151
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0182840824127197,
      "learning_rate": 2.0935393094868094e-06,
      "loss": 0.4995,
      "step": 11152
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0194854736328125,
      "learning_rate": 2.092704524819089e-06,
      "loss": 0.466,
      "step": 11153
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0558594465255737,
      "learning_rate": 2.091869862567471e-06,
      "loss": 0.5392,
      "step": 11154
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.046743392944336,
      "learning_rate": 2.091035322767095e-06,
      "loss": 0.5264,
      "step": 11155
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.071263074874878,
      "learning_rate": 2.0902009054531013e-06,
      "loss": 0.5353,
      "step": 11156
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.16365385055542,
      "learning_rate": 2.089366610660627e-06,
      "loss": 0.5367,
      "step": 11157
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0887624025344849,
      "learning_rate": 2.0885324384247956e-06,
      "loss": 0.5095,
      "step": 11158
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9924121499061584,
      "learning_rate": 2.087698388780739e-06,
      "loss": 0.4924,
      "step": 11159
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0082294940948486,
      "learning_rate": 2.0868644617635697e-06,
      "loss": 0.4927,
      "step": 11160
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0977580547332764,
      "learning_rate": 2.0860306574084043e-06,
      "loss": 0.4932,
      "step": 11161
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0345255136489868,
      "learning_rate": 2.085196975750351e-06,
      "loss": 0.5063,
      "step": 11162
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0377495288848877,
      "learning_rate": 2.084363416824513e-06,
      "loss": 0.4963,
      "step": 11163
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.968866765499115,
      "learning_rate": 2.0835299806659885e-06,
      "loss": 0.4895,
      "step": 11164
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0815399885177612,
      "learning_rate": 2.0826966673098737e-06,
      "loss": 0.5468,
      "step": 11165
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0565381050109863,
      "learning_rate": 2.0818634767912495e-06,
      "loss": 0.5364,
      "step": 11166
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0638211965560913,
      "learning_rate": 2.081030409145206e-06,
      "loss": 0.5149,
      "step": 11167
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0905492305755615,
      "learning_rate": 2.080197464406816e-06,
      "loss": 0.525,
      "step": 11168
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9871543645858765,
      "learning_rate": 2.0793646426111536e-06,
      "loss": 0.4742,
      "step": 11169
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0492466688156128,
      "learning_rate": 2.078531943793288e-06,
      "loss": 0.5319,
      "step": 11170
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1168328523635864,
      "learning_rate": 2.0776993679882752e-06,
      "loss": 0.4988,
      "step": 11171
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9689909815788269,
      "learning_rate": 2.076866915231178e-06,
      "loss": 0.5064,
      "step": 11172
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1330360174179077,
      "learning_rate": 2.076034585557048e-06,
      "loss": 0.5113,
      "step": 11173
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1285103559494019,
      "learning_rate": 2.075202379000928e-06,
      "loss": 0.4873,
      "step": 11174
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0235905647277832,
      "learning_rate": 2.074370295597861e-06,
      "loss": 0.5143,
      "step": 11175
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1421451568603516,
      "learning_rate": 2.0735383353828843e-06,
      "loss": 0.5372,
      "step": 11176
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0220662355422974,
      "learning_rate": 2.0727064983910266e-06,
      "loss": 0.5209,
      "step": 11177
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1236740350723267,
      "learning_rate": 2.071874784657318e-06,
      "loss": 0.5237,
      "step": 11178
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0522505044937134,
      "learning_rate": 2.0710431942167713e-06,
      "loss": 0.4875,
      "step": 11179
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1153757572174072,
      "learning_rate": 2.070211727104409e-06,
      "loss": 0.4592,
      "step": 11180
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1207882165908813,
      "learning_rate": 2.0693803833552407e-06,
      "loss": 0.5077,
      "step": 11181
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9922836422920227,
      "learning_rate": 2.0685491630042677e-06,
      "loss": 0.527,
      "step": 11182
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0508869886398315,
      "learning_rate": 2.0677180660864916e-06,
      "loss": 0.5101,
      "step": 11183
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1612192392349243,
      "learning_rate": 2.0668870926369068e-06,
      "loss": 0.496,
      "step": 11184
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0142918825149536,
      "learning_rate": 2.066056242690503e-06,
      "loss": 0.5285,
      "step": 11185
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0472638607025146,
      "learning_rate": 2.0652255162822665e-06,
      "loss": 0.4657,
      "step": 11186
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0182528495788574,
      "learning_rate": 2.0643949134471726e-06,
      "loss": 0.4788,
      "step": 11187
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0504941940307617,
      "learning_rate": 2.0635644342201942e-06,
      "loss": 0.528,
      "step": 11188
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9788329005241394,
      "learning_rate": 2.0627340786363063e-06,
      "loss": 0.5245,
      "step": 11189
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.095033049583435,
      "learning_rate": 2.0619038467304663e-06,
      "loss": 0.517,
      "step": 11190
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.038672685623169,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.4951,
      "step": 11191
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0306553840637207,
      "learning_rate": 2.0602437540927644e-06,
      "loss": 0.4946,
      "step": 11192
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1046713590621948,
      "learning_rate": 2.0594138934308027e-06,
      "loss": 0.5486,
      "step": 11193
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0907617807388306,
      "learning_rate": 2.058584156586692e-06,
      "loss": 0.5483,
      "step": 11194
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0609095096588135,
      "learning_rate": 2.0577545435953727e-06,
      "loss": 0.4487,
      "step": 11195
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0488719940185547,
      "learning_rate": 2.05692505449177e-06,
      "loss": 0.5099,
      "step": 11196
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1803630590438843,
      "learning_rate": 2.0560956893108188e-06,
      "loss": 0.5679,
      "step": 11197
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.075703501701355,
      "learning_rate": 2.0552664480874353e-06,
      "loss": 0.4943,
      "step": 11198
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1885625123977661,
      "learning_rate": 2.0544373308565374e-06,
      "loss": 0.5386,
      "step": 11199
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1075416803359985,
      "learning_rate": 2.0536083376530368e-06,
      "loss": 0.4967,
      "step": 11200
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.011660099029541,
      "learning_rate": 2.0527794685118397e-06,
      "loss": 0.5201,
      "step": 11201
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0100550651550293,
      "learning_rate": 2.0519507234678464e-06,
      "loss": 0.539,
      "step": 11202
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0797109603881836,
      "learning_rate": 2.051122102555954e-06,
      "loss": 0.5122,
      "step": 11203
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1042869091033936,
      "learning_rate": 2.0502936058110502e-06,
      "loss": 0.5564,
      "step": 11204
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0236577987670898,
      "learning_rate": 2.049465233268021e-06,
      "loss": 0.5112,
      "step": 11205
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.011128544807434,
      "learning_rate": 2.0486369849617467e-06,
      "loss": 0.4928,
      "step": 11206
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1878892183303833,
      "learning_rate": 2.0478088609271018e-06,
      "loss": 0.4953,
      "step": 11207
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1174811124801636,
      "learning_rate": 2.0469808611989583e-06,
      "loss": 0.5365,
      "step": 11208
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9994081854820251,
      "learning_rate": 2.0461529858121737e-06,
      "loss": 0.5011,
      "step": 11209
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0205951929092407,
      "learning_rate": 2.0453252348016133e-06,
      "loss": 0.5281,
      "step": 11210
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1167793273925781,
      "learning_rate": 2.04449760820213e-06,
      "loss": 0.5057,
      "step": 11211
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0231558084487915,
      "learning_rate": 2.04367010604857e-06,
      "loss": 0.5226,
      "step": 11212
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9758703112602234,
      "learning_rate": 2.042842728375777e-06,
      "loss": 0.5179,
      "step": 11213
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9797309637069702,
      "learning_rate": 2.0420154752185896e-06,
      "loss": 0.4918,
      "step": 11214
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0304148197174072,
      "learning_rate": 2.0411883466118406e-06,
      "loss": 0.5326,
      "step": 11215
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0446081161499023,
      "learning_rate": 2.0403613425903584e-06,
      "loss": 0.5099,
      "step": 11216
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0075976848602295,
      "learning_rate": 2.0395344631889636e-06,
      "loss": 0.4837,
      "step": 11217
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0283238887786865,
      "learning_rate": 2.038707708442471e-06,
      "loss": 0.5477,
      "step": 11218
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0798712968826294,
      "learning_rate": 2.0378810783856996e-06,
      "loss": 0.5146,
      "step": 11219
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9746232032775879,
      "learning_rate": 2.0370545730534493e-06,
      "loss": 0.478,
      "step": 11220
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0954176187515259,
      "learning_rate": 2.0362281924805238e-06,
      "loss": 0.5121,
      "step": 11221
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0521270036697388,
      "learning_rate": 2.035401936701719e-06,
      "loss": 0.5004,
      "step": 11222
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0404530763626099,
      "learning_rate": 2.034575805751825e-06,
      "loss": 0.5091,
      "step": 11223
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9950894713401794,
      "learning_rate": 2.0337497996656303e-06,
      "loss": 0.5174,
      "step": 11224
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9830288887023926,
      "learning_rate": 2.03292391847791e-06,
      "loss": 0.4834,
      "step": 11225
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1587971448898315,
      "learning_rate": 2.032098162223441e-06,
      "loss": 0.5193,
      "step": 11226
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9999991059303284,
      "learning_rate": 2.031272530936997e-06,
      "loss": 0.4845,
      "step": 11227
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9857888221740723,
      "learning_rate": 2.0304470246533377e-06,
      "loss": 0.5142,
      "step": 11228
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0121004581451416,
      "learning_rate": 2.0296216434072237e-06,
      "loss": 0.4667,
      "step": 11229
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0974059104919434,
      "learning_rate": 2.0287963872334093e-06,
      "loss": 0.5211,
      "step": 11230
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9777387976646423,
      "learning_rate": 2.0279712561666425e-06,
      "loss": 0.4779,
      "step": 11231
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0181130170822144,
      "learning_rate": 2.0271462502416694e-06,
      "loss": 0.5109,
      "step": 11232
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.071170687675476,
      "learning_rate": 2.0263213694932238e-06,
      "loss": 0.4536,
      "step": 11233
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.005286455154419,
      "learning_rate": 2.0254966139560404e-06,
      "loss": 0.4688,
      "step": 11234
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1063686609268188,
      "learning_rate": 2.0246719836648476e-06,
      "loss": 0.5128,
      "step": 11235
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.131303310394287,
      "learning_rate": 2.0238474786543673e-06,
      "loss": 0.4972,
      "step": 11236
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.00669264793396,
      "learning_rate": 2.0230230989593157e-06,
      "loss": 0.4993,
      "step": 11237
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0949076414108276,
      "learning_rate": 2.0221988446144076e-06,
      "loss": 0.5441,
      "step": 11238
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1112231016159058,
      "learning_rate": 2.0213747156543432e-06,
      "loss": 0.4917,
      "step": 11239
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0323106050491333,
      "learning_rate": 2.0205507121138316e-06,
      "loss": 0.4912,
      "step": 11240
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1358355283737183,
      "learning_rate": 2.019726834027563e-06,
      "loss": 0.511,
      "step": 11241
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0783838033676147,
      "learning_rate": 2.0189030814302295e-06,
      "loss": 0.4974,
      "step": 11242
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1246992349624634,
      "learning_rate": 2.018079454356517e-06,
      "loss": 0.4554,
      "step": 11243
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1323680877685547,
      "learning_rate": 2.017255952841105e-06,
      "loss": 0.5229,
      "step": 11244
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.051601767539978,
      "learning_rate": 2.016432576918669e-06,
      "loss": 0.4711,
      "step": 11245
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0374492406845093,
      "learning_rate": 2.0156093266238795e-06,
      "loss": 0.4495,
      "step": 11246
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.093229055404663,
      "learning_rate": 2.014786201991396e-06,
      "loss": 0.5367,
      "step": 11247
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1299701929092407,
      "learning_rate": 2.0139632030558844e-06,
      "loss": 0.5289,
      "step": 11248
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.997376561164856,
      "learning_rate": 2.0131403298519927e-06,
      "loss": 0.4649,
      "step": 11249
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.056484580039978,
      "learning_rate": 2.012317582414371e-06,
      "loss": 0.5172,
      "step": 11250
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0173227787017822,
      "learning_rate": 2.011494960777663e-06,
      "loss": 0.4651,
      "step": 11251
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0372679233551025,
      "learning_rate": 2.0106724649765055e-06,
      "loss": 0.5308,
      "step": 11252
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1156564950942993,
      "learning_rate": 2.0098500950455313e-06,
      "loss": 0.5787,
      "step": 11253
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0682792663574219,
      "learning_rate": 2.00902785101937e-06,
      "loss": 0.4912,
      "step": 11254
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9660002589225769,
      "learning_rate": 2.008205732932639e-06,
      "loss": 0.4581,
      "step": 11255
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0296785831451416,
      "learning_rate": 2.0073837408199566e-06,
      "loss": 0.4989,
      "step": 11256
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.090598225593567,
      "learning_rate": 2.0065618747159342e-06,
      "loss": 0.4808,
      "step": 11257
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.943744957447052,
      "learning_rate": 2.0057401346551785e-06,
      "loss": 0.4732,
      "step": 11258
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0518137216567993,
      "learning_rate": 2.004918520672289e-06,
      "loss": 0.5272,
      "step": 11259
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.031441569328308,
      "learning_rate": 2.0040970328018618e-06,
      "loss": 0.535,
      "step": 11260
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0439598560333252,
      "learning_rate": 2.0032756710784864e-06,
      "loss": 0.5144,
      "step": 11261
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0440160036087036,
      "learning_rate": 2.0024544355367494e-06,
      "loss": 0.5096,
      "step": 11262
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.067297339439392,
      "learning_rate": 2.001633326211227e-06,
      "loss": 0.4597,
      "step": 11263
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.032602310180664,
      "learning_rate": 2.000812343136494e-06,
      "loss": 0.5235,
      "step": 11264
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0836875438690186,
      "learning_rate": 1.99999148634712e-06,
      "loss": 0.5136,
      "step": 11265
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0357921123504639,
      "learning_rate": 1.9991707558776686e-06,
      "loss": 0.5102,
      "step": 11266
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1127569675445557,
      "learning_rate": 1.9983501517626976e-06,
      "loss": 0.5721,
      "step": 11267
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9863804578781128,
      "learning_rate": 1.997529674036761e-06,
      "loss": 0.4779,
      "step": 11268
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0605449676513672,
      "learning_rate": 1.9967093227344013e-06,
      "loss": 0.5381,
      "step": 11269
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0444042682647705,
      "learning_rate": 1.9958890978901685e-06,
      "loss": 0.5028,
      "step": 11270
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1172131299972534,
      "learning_rate": 1.9950689995385936e-06,
      "loss": 0.4988,
      "step": 11271
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.117167592048645,
      "learning_rate": 1.994249027714209e-06,
      "loss": 0.5496,
      "step": 11272
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0403753519058228,
      "learning_rate": 1.9934291824515423e-06,
      "loss": 0.493,
      "step": 11273
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0917158126831055,
      "learning_rate": 1.9926094637851135e-06,
      "loss": 0.5256,
      "step": 11274
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0141263008117676,
      "learning_rate": 1.9917898717494377e-06,
      "loss": 0.5241,
      "step": 11275
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0066581964492798,
      "learning_rate": 1.990970406379028e-06,
      "loss": 0.4802,
      "step": 11276
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.133183479309082,
      "learning_rate": 1.990151067708383e-06,
      "loss": 0.5512,
      "step": 11277
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0141957998275757,
      "learning_rate": 1.9893318557720093e-06,
      "loss": 0.507,
      "step": 11278
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0804271697998047,
      "learning_rate": 1.9885127706043966e-06,
      "loss": 0.5047,
      "step": 11279
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.102764368057251,
      "learning_rate": 1.9876938122400348e-06,
      "loss": 0.5152,
      "step": 11280
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1434690952301025,
      "learning_rate": 1.9868749807134087e-06,
      "loss": 0.5075,
      "step": 11281
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0453921556472778,
      "learning_rate": 1.9860562760589926e-06,
      "loss": 0.4667,
      "step": 11282
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1235970258712769,
      "learning_rate": 1.9852376983112632e-06,
      "loss": 0.5656,
      "step": 11283
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0331393480300903,
      "learning_rate": 1.9844192475046885e-06,
      "loss": 0.5191,
      "step": 11284
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0233731269836426,
      "learning_rate": 1.983600923673727e-06,
      "loss": 0.4515,
      "step": 11285
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.056684136390686,
      "learning_rate": 1.9827827268528378e-06,
      "loss": 0.471,
      "step": 11286
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.06417977809906,
      "learning_rate": 1.9819646570764712e-06,
      "loss": 0.5335,
      "step": 11287
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.177718997001648,
      "learning_rate": 1.981146714379074e-06,
      "loss": 0.5429,
      "step": 11288
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0227022171020508,
      "learning_rate": 1.980328898795089e-06,
      "loss": 0.4898,
      "step": 11289
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1076995134353638,
      "learning_rate": 1.979511210358946e-06,
      "loss": 0.5011,
      "step": 11290
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0327277183532715,
      "learning_rate": 1.9786936491050803e-06,
      "loss": 0.4993,
      "step": 11291
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0501066446304321,
      "learning_rate": 1.9778762150679155e-06,
      "loss": 0.4762,
      "step": 11292
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.088322639465332,
      "learning_rate": 1.9770589082818694e-06,
      "loss": 0.5064,
      "step": 11293
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0388498306274414,
      "learning_rate": 1.9762417287813557e-06,
      "loss": 0.4958,
      "step": 11294
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1175765991210938,
      "learning_rate": 1.9754246766007847e-06,
      "loss": 0.5141,
      "step": 11295
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0213274955749512,
      "learning_rate": 1.9746077517745582e-06,
      "loss": 0.4621,
      "step": 11296
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9821017980575562,
      "learning_rate": 1.9737909543370764e-06,
      "loss": 0.4978,
      "step": 11297
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0173606872558594,
      "learning_rate": 1.972974284322729e-06,
      "loss": 0.4404,
      "step": 11298
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9861236214637756,
      "learning_rate": 1.9721577417659023e-06,
      "loss": 0.4939,
      "step": 11299
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0011417865753174,
      "learning_rate": 1.9713413267009827e-06,
      "loss": 0.4903,
      "step": 11300
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.010124921798706,
      "learning_rate": 1.970525039162343e-06,
      "loss": 0.5351,
      "step": 11301
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1067034006118774,
      "learning_rate": 1.969708879184355e-06,
      "loss": 0.5409,
      "step": 11302
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0343722105026245,
      "learning_rate": 1.9688928468013846e-06,
      "loss": 0.4735,
      "step": 11303
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.037208914756775,
      "learning_rate": 1.968076942047791e-06,
      "loss": 0.4547,
      "step": 11304
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0555593967437744,
      "learning_rate": 1.9672611649579332e-06,
      "loss": 0.5243,
      "step": 11305
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0606285333633423,
      "learning_rate": 1.966445515566155e-06,
      "loss": 0.4775,
      "step": 11306
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1268073320388794,
      "learning_rate": 1.965629993906802e-06,
      "loss": 0.538,
      "step": 11307
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.985319972038269,
      "learning_rate": 1.9648146000142173e-06,
      "loss": 0.4964,
      "step": 11308
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.029909610748291,
      "learning_rate": 1.963999333922729e-06,
      "loss": 0.4562,
      "step": 11309
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.000618815422058,
      "learning_rate": 1.963184195666668e-06,
      "loss": 0.4975,
      "step": 11310
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0195996761322021,
      "learning_rate": 1.9623691852803577e-06,
      "loss": 0.4938,
      "step": 11311
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0395841598510742,
      "learning_rate": 1.9615543027981105e-06,
      "loss": 0.5104,
      "step": 11312
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0242948532104492,
      "learning_rate": 1.9607395482542446e-06,
      "loss": 0.4656,
      "step": 11313
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.117276906967163,
      "learning_rate": 1.9599249216830624e-06,
      "loss": 0.5337,
      "step": 11314
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0707858800888062,
      "learning_rate": 1.9591104231188656e-06,
      "loss": 0.5017,
      "step": 11315
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9831403493881226,
      "learning_rate": 1.958296052595951e-06,
      "loss": 0.4853,
      "step": 11316
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0604276657104492,
      "learning_rate": 1.9574818101486075e-06,
      "loss": 0.4831,
      "step": 11317
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0770059823989868,
      "learning_rate": 1.9566676958111214e-06,
      "loss": 0.5233,
      "step": 11318
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0600957870483398,
      "learning_rate": 1.955853709617773e-06,
      "loss": 0.517,
      "step": 11319
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.092265009880066,
      "learning_rate": 1.955039851602832e-06,
      "loss": 0.5532,
      "step": 11320
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0496869087219238,
      "learning_rate": 1.9542261218005737e-06,
      "loss": 0.5111,
      "step": 11321
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0955640077590942,
      "learning_rate": 1.9534125202452557e-06,
      "loss": 0.5324,
      "step": 11322
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0555497407913208,
      "learning_rate": 1.952599046971139e-06,
      "loss": 0.5649,
      "step": 11323
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0537457466125488,
      "learning_rate": 1.951785702012475e-06,
      "loss": 0.5116,
      "step": 11324
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.101248025894165,
      "learning_rate": 1.9509724854035105e-06,
      "loss": 0.5101,
      "step": 11325
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.058995246887207,
      "learning_rate": 1.950159397178488e-06,
      "loss": 0.5711,
      "step": 11326
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1053820848464966,
      "learning_rate": 1.9493464373716458e-06,
      "loss": 0.5189,
      "step": 11327
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.038176417350769,
      "learning_rate": 1.9485336060172106e-06,
      "loss": 0.5208,
      "step": 11328
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.162173867225647,
      "learning_rate": 1.9477209031494104e-06,
      "loss": 0.5208,
      "step": 11329
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8813678026199341,
      "learning_rate": 1.9469083288024647e-06,
      "loss": 0.4458,
      "step": 11330
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0723689794540405,
      "learning_rate": 1.9460958830105882e-06,
      "loss": 0.5337,
      "step": 11331
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0425341129302979,
      "learning_rate": 1.9452835658079905e-06,
      "loss": 0.4759,
      "step": 11332
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.118994951248169,
      "learning_rate": 1.9444713772288747e-06,
      "loss": 0.5553,
      "step": 11333
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1561713218688965,
      "learning_rate": 1.94365931730744e-06,
      "loss": 0.5353,
      "step": 11334
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0623376369476318,
      "learning_rate": 1.9428473860778817e-06,
      "loss": 0.5039,
      "step": 11335
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9964343309402466,
      "learning_rate": 1.9420355835743826e-06,
      "loss": 0.5395,
      "step": 11336
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0136643648147583,
      "learning_rate": 1.941223909831125e-06,
      "loss": 0.5167,
      "step": 11337
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1962140798568726,
      "learning_rate": 1.9404123648822924e-06,
      "loss": 0.5547,
      "step": 11338
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.083602786064148,
      "learning_rate": 1.9396009487620494e-06,
      "loss": 0.4856,
      "step": 11339
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0315816402435303,
      "learning_rate": 1.9387896615045636e-06,
      "loss": 0.5043,
      "step": 11340
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1157749891281128,
      "learning_rate": 1.9379785031439985e-06,
      "loss": 0.4993,
      "step": 11341
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0104255676269531,
      "learning_rate": 1.9371674737145023e-06,
      "loss": 0.489,
      "step": 11342
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9678119421005249,
      "learning_rate": 1.936356573250233e-06,
      "loss": 0.5093,
      "step": 11343
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0498720407485962,
      "learning_rate": 1.935545801785329e-06,
      "loss": 0.4843,
      "step": 11344
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.045360803604126,
      "learning_rate": 1.934735159353931e-06,
      "loss": 0.4978,
      "step": 11345
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9735890030860901,
      "learning_rate": 1.9339246459901715e-06,
      "loss": 0.5155,
      "step": 11346
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0178331136703491,
      "learning_rate": 1.93311426172818e-06,
      "loss": 0.5073,
      "step": 11347
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0024511814117432,
      "learning_rate": 1.9323040066020774e-06,
      "loss": 0.5313,
      "step": 11348
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.004668116569519,
      "learning_rate": 1.931493880645983e-06,
      "loss": 0.5297,
      "step": 11349
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0038278102874756,
      "learning_rate": 1.9306838838940035e-06,
      "loss": 0.481,
      "step": 11350
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0347565412521362,
      "learning_rate": 1.9298740163802523e-06,
      "loss": 0.5263,
      "step": 11351
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9967230558395386,
      "learning_rate": 1.929064278138823e-06,
      "loss": 0.5257,
      "step": 11352
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.077918529510498,
      "learning_rate": 1.928254669203815e-06,
      "loss": 0.5305,
      "step": 11353
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.112382173538208,
      "learning_rate": 1.9274451896093164e-06,
      "loss": 0.5486,
      "step": 11354
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0560412406921387,
      "learning_rate": 1.926635839389413e-06,
      "loss": 0.462,
      "step": 11355
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9729539155960083,
      "learning_rate": 1.925826618578182e-06,
      "loss": 0.4788,
      "step": 11356
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0809910297393799,
      "learning_rate": 1.9250175272097003e-06,
      "loss": 0.4713,
      "step": 11357
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.012824535369873,
      "learning_rate": 1.9242085653180314e-06,
      "loss": 0.4867,
      "step": 11358
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1029964685440063,
      "learning_rate": 1.9233997329372402e-06,
      "loss": 0.5303,
      "step": 11359
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1370534896850586,
      "learning_rate": 1.9225910301013834e-06,
      "loss": 0.4887,
      "step": 11360
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.029061198234558,
      "learning_rate": 1.9217824568445125e-06,
      "loss": 0.5302,
      "step": 11361
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.021522879600525,
      "learning_rate": 1.920974013200676e-06,
      "loss": 0.5076,
      "step": 11362
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.06879460811615,
      "learning_rate": 1.9201656992039092e-06,
      "loss": 0.4848,
      "step": 11363
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2080460786819458,
      "learning_rate": 1.9193575148882526e-06,
      "loss": 0.5572,
      "step": 11364
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1446185111999512,
      "learning_rate": 1.918549460287736e-06,
      "loss": 0.5437,
      "step": 11365
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9499037861824036,
      "learning_rate": 1.9177415354363802e-06,
      "loss": 0.4596,
      "step": 11366
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0208501815795898,
      "learning_rate": 1.916933740368206e-06,
      "loss": 0.4868,
      "step": 11367
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9540845155715942,
      "learning_rate": 1.916126075117227e-06,
      "loss": 0.4696,
      "step": 11368
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9916105270385742,
      "learning_rate": 1.9153185397174506e-06,
      "loss": 0.5258,
      "step": 11369
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9990249872207642,
      "learning_rate": 1.9145111342028817e-06,
      "loss": 0.5047,
      "step": 11370
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.140777587890625,
      "learning_rate": 1.9137038586075117e-06,
      "loss": 0.5219,
      "step": 11371
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0075373649597168,
      "learning_rate": 1.9128967129653375e-06,
      "loss": 0.4989,
      "step": 11372
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0705087184906006,
      "learning_rate": 1.9120896973103453e-06,
      "loss": 0.5279,
      "step": 11373
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1543906927108765,
      "learning_rate": 1.911282811676512e-06,
      "loss": 0.5446,
      "step": 11374
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9973766207695007,
      "learning_rate": 1.9104760560978147e-06,
      "loss": 0.4964,
      "step": 11375
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.022469401359558,
      "learning_rate": 1.909669430608223e-06,
      "loss": 0.4754,
      "step": 11376
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0960901975631714,
      "learning_rate": 1.908862935241701e-06,
      "loss": 0.5173,
      "step": 11377
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.974963903427124,
      "learning_rate": 1.9080565700322095e-06,
      "loss": 0.4718,
      "step": 11378
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1384514570236206,
      "learning_rate": 1.9072503350136979e-06,
      "loss": 0.4665,
      "step": 11379
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.045607328414917,
      "learning_rate": 1.9064442302201136e-06,
      "loss": 0.4791,
      "step": 11380
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.064322829246521,
      "learning_rate": 1.9056382556854053e-06,
      "loss": 0.5211,
      "step": 11381
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9941070675849915,
      "learning_rate": 1.9048324114435036e-06,
      "loss": 0.5104,
      "step": 11382
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.000701665878296,
      "learning_rate": 1.9040266975283417e-06,
      "loss": 0.4908,
      "step": 11383
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0231435298919678,
      "learning_rate": 1.9032211139738455e-06,
      "loss": 0.497,
      "step": 11384
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9945744276046753,
      "learning_rate": 1.902415660813935e-06,
      "loss": 0.5399,
      "step": 11385
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.140770673751831,
      "learning_rate": 1.9016103380825274e-06,
      "loss": 0.4905,
      "step": 11386
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0725229978561401,
      "learning_rate": 1.900805145813528e-06,
      "loss": 0.5188,
      "step": 11387
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0397700071334839,
      "learning_rate": 1.9000000840408421e-06,
      "loss": 0.5006,
      "step": 11388
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0895681381225586,
      "learning_rate": 1.8991951527983694e-06,
      "loss": 0.5217,
      "step": 11389
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0540904998779297,
      "learning_rate": 1.8983903521200015e-06,
      "loss": 0.5251,
      "step": 11390
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0280147790908813,
      "learning_rate": 1.8975856820396265e-06,
      "loss": 0.4993,
      "step": 11391
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9986823797225952,
      "learning_rate": 1.8967811425911275e-06,
      "loss": 0.5326,
      "step": 11392
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0397193431854248,
      "learning_rate": 1.8959767338083758e-06,
      "loss": 0.501,
      "step": 11393
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0313090085983276,
      "learning_rate": 1.8951724557252472e-06,
      "loss": 0.4645,
      "step": 11394
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9666309952735901,
      "learning_rate": 1.8943683083756075e-06,
      "loss": 0.5031,
      "step": 11395
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.138631820678711,
      "learning_rate": 1.8935642917933128e-06,
      "loss": 0.5772,
      "step": 11396
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0611635446548462,
      "learning_rate": 1.8927604060122196e-06,
      "loss": 0.5068,
      "step": 11397
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1099668741226196,
      "learning_rate": 1.8919566510661758e-06,
      "loss": 0.5055,
      "step": 11398
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0566827058792114,
      "learning_rate": 1.891153026989026e-06,
      "loss": 0.5213,
      "step": 11399
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0656869411468506,
      "learning_rate": 1.8903495338146089e-06,
      "loss": 0.5202,
      "step": 11400
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1294125318527222,
      "learning_rate": 1.8895461715767517e-06,
      "loss": 0.4552,
      "step": 11401
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9855488538742065,
      "learning_rate": 1.888742940309286e-06,
      "loss": 0.4594,
      "step": 11402
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1352120637893677,
      "learning_rate": 1.8879398400460342e-06,
      "loss": 0.5326,
      "step": 11403
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9844202995300293,
      "learning_rate": 1.8871368708208076e-06,
      "loss": 0.4825,
      "step": 11404
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0212651491165161,
      "learning_rate": 1.8863340326674184e-06,
      "loss": 0.5259,
      "step": 11405
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.008789300918579,
      "learning_rate": 1.8855313256196722e-06,
      "loss": 0.5367,
      "step": 11406
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.014675498008728,
      "learning_rate": 1.8847287497113664e-06,
      "loss": 0.4533,
      "step": 11407
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0439823865890503,
      "learning_rate": 1.883926304976298e-06,
      "loss": 0.4689,
      "step": 11408
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0333361625671387,
      "learning_rate": 1.8831239914482512e-06,
      "loss": 0.4874,
      "step": 11409
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0649745464324951,
      "learning_rate": 1.8823218091610085e-06,
      "loss": 0.5201,
      "step": 11410
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0688813924789429,
      "learning_rate": 1.8815197581483523e-06,
      "loss": 0.4849,
      "step": 11411
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0986980199813843,
      "learning_rate": 1.880717838444049e-06,
      "loss": 0.516,
      "step": 11412
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0068950653076172,
      "learning_rate": 1.879916050081866e-06,
      "loss": 0.474,
      "step": 11413
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0525602102279663,
      "learning_rate": 1.8791143930955641e-06,
      "loss": 0.4844,
      "step": 11414
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0503764152526855,
      "learning_rate": 1.8783128675188988e-06,
      "loss": 0.5612,
      "step": 11415
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.080776572227478,
      "learning_rate": 1.8775114733856203e-06,
      "loss": 0.492,
      "step": 11416
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9848031997680664,
      "learning_rate": 1.87671021072947e-06,
      "loss": 0.5013,
      "step": 11417
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.121417760848999,
      "learning_rate": 1.8759090795841856e-06,
      "loss": 0.5254,
      "step": 11418
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0587754249572754,
      "learning_rate": 1.8751080799835059e-06,
      "loss": 0.4595,
      "step": 11419
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.024379849433899,
      "learning_rate": 1.8743072119611522e-06,
      "loss": 0.4652,
      "step": 11420
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0574456453323364,
      "learning_rate": 1.873506475550848e-06,
      "loss": 0.5539,
      "step": 11421
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9183557629585266,
      "learning_rate": 1.8727058707863121e-06,
      "loss": 0.4387,
      "step": 11422
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0336495637893677,
      "learning_rate": 1.871905397701249e-06,
      "loss": 0.5565,
      "step": 11423
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1050636768341064,
      "learning_rate": 1.8711050563293714e-06,
      "loss": 0.5102,
      "step": 11424
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.00590980052948,
      "learning_rate": 1.8703048467043732e-06,
      "loss": 0.4588,
      "step": 11425
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1239380836486816,
      "learning_rate": 1.869504768859951e-06,
      "loss": 0.4816,
      "step": 11426
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1312364339828491,
      "learning_rate": 1.8687048228297928e-06,
      "loss": 0.539,
      "step": 11427
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0960451364517212,
      "learning_rate": 1.8679050086475814e-06,
      "loss": 0.4997,
      "step": 11428
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0225590467453003,
      "learning_rate": 1.867105326346994e-06,
      "loss": 0.5485,
      "step": 11429
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0000320672988892,
      "learning_rate": 1.8663057759617048e-06,
      "loss": 0.5167,
      "step": 11430
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1537526845932007,
      "learning_rate": 1.8655063575253746e-06,
      "loss": 0.5222,
      "step": 11431
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1456413269042969,
      "learning_rate": 1.8647070710716709e-06,
      "loss": 0.5506,
      "step": 11432
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9794883131980896,
      "learning_rate": 1.8639079166342438e-06,
      "loss": 0.4625,
      "step": 11433
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0213878154754639,
      "learning_rate": 1.8631088942467452e-06,
      "loss": 0.5083,
      "step": 11434
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0443166494369507,
      "learning_rate": 1.8623100039428194e-06,
      "loss": 0.4777,
      "step": 11435
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.068365454673767,
      "learning_rate": 1.8615112457561013e-06,
      "loss": 0.5047,
      "step": 11436
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0105688571929932,
      "learning_rate": 1.860712619720228e-06,
      "loss": 0.5454,
      "step": 11437
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.080657958984375,
      "learning_rate": 1.8599141258688274e-06,
      "loss": 0.5188,
      "step": 11438
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0406492948532104,
      "learning_rate": 1.8591157642355179e-06,
      "loss": 0.4535,
      "step": 11439
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0377037525177002,
      "learning_rate": 1.8583175348539173e-06,
      "loss": 0.4785,
      "step": 11440
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0017094612121582,
      "learning_rate": 1.8575194377576355e-06,
      "loss": 0.445,
      "step": 11441
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.05403470993042,
      "learning_rate": 1.856721472980279e-06,
      "loss": 0.4839,
      "step": 11442
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.117053508758545,
      "learning_rate": 1.855923640555448e-06,
      "loss": 0.4894,
      "step": 11443
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.067420482635498,
      "learning_rate": 1.8551259405167315e-06,
      "loss": 0.4831,
      "step": 11444
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.935054361820221,
      "learning_rate": 1.8543283728977234e-06,
      "loss": 0.4642,
      "step": 11445
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9502724409103394,
      "learning_rate": 1.8535309377320059e-06,
      "loss": 0.5063,
      "step": 11446
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.96114581823349,
      "learning_rate": 1.8527336350531532e-06,
      "loss": 0.4877,
      "step": 11447
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.013749361038208,
      "learning_rate": 1.851936464894739e-06,
      "loss": 0.5046,
      "step": 11448
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0789744853973389,
      "learning_rate": 1.8511394272903287e-06,
      "loss": 0.5355,
      "step": 11449
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0386903285980225,
      "learning_rate": 1.8503425222734834e-06,
      "loss": 0.4728,
      "step": 11450
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0988272428512573,
      "learning_rate": 1.8495457498777585e-06,
      "loss": 0.5491,
      "step": 11451
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.028246283531189,
      "learning_rate": 1.8487491101367016e-06,
      "loss": 0.4953,
      "step": 11452
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1107392311096191,
      "learning_rate": 1.8479526030838552e-06,
      "loss": 0.5548,
      "step": 11453
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0843489170074463,
      "learning_rate": 1.8471562287527627e-06,
      "loss": 0.5282,
      "step": 11454
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0672394037246704,
      "learning_rate": 1.8463599871769516e-06,
      "loss": 0.5588,
      "step": 11455
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1077324151992798,
      "learning_rate": 1.8455638783899515e-06,
      "loss": 0.525,
      "step": 11456
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9883722066879272,
      "learning_rate": 1.8447679024252825e-06,
      "loss": 0.5146,
      "step": 11457
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.081155776977539,
      "learning_rate": 1.8439720593164606e-06,
      "loss": 0.5191,
      "step": 11458
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9750945568084717,
      "learning_rate": 1.8431763490969968e-06,
      "loss": 0.5055,
      "step": 11459
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.024051547050476,
      "learning_rate": 1.8423807718003967e-06,
      "loss": 0.477,
      "step": 11460
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0928895473480225,
      "learning_rate": 1.8415853274601541e-06,
      "loss": 0.476,
      "step": 11461
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1447224617004395,
      "learning_rate": 1.8407900161097698e-06,
      "loss": 0.5256,
      "step": 11462
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0564632415771484,
      "learning_rate": 1.839994837782726e-06,
      "loss": 0.5082,
      "step": 11463
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0120878219604492,
      "learning_rate": 1.8391997925125066e-06,
      "loss": 0.5329,
      "step": 11464
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0647292137145996,
      "learning_rate": 1.8384048803325887e-06,
      "loss": 0.505,
      "step": 11465
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.062643051147461,
      "learning_rate": 1.8376101012764424e-06,
      "loss": 0.4714,
      "step": 11466
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0453153848648071,
      "learning_rate": 1.8368154553775342e-06,
      "loss": 0.4973,
      "step": 11467
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0845417976379395,
      "learning_rate": 1.8360209426693242e-06,
      "loss": 0.4738,
      "step": 11468
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.014893651008606,
      "learning_rate": 1.8352265631852645e-06,
      "loss": 0.4753,
      "step": 11469
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.152695655822754,
      "learning_rate": 1.8344323169588045e-06,
      "loss": 0.5137,
      "step": 11470
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.087654948234558,
      "learning_rate": 1.8336382040233874e-06,
      "loss": 0.5012,
      "step": 11471
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.026802659034729,
      "learning_rate": 1.8328442244124506e-06,
      "loss": 0.4385,
      "step": 11472
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0484585762023926,
      "learning_rate": 1.8320503781594273e-06,
      "loss": 0.5454,
      "step": 11473
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.069604754447937,
      "learning_rate": 1.8312566652977393e-06,
      "loss": 0.5103,
      "step": 11474
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.077880620956421,
      "learning_rate": 1.8304630858608107e-06,
      "loss": 0.5361,
      "step": 11475
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0621845722198486,
      "learning_rate": 1.8296696398820579e-06,
      "loss": 0.5359,
      "step": 11476
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2201637029647827,
      "learning_rate": 1.828876327394886e-06,
      "loss": 0.5066,
      "step": 11477
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.049737572669983,
      "learning_rate": 1.8280831484327006e-06,
      "loss": 0.5025,
      "step": 11478
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0424972772598267,
      "learning_rate": 1.8272901030288991e-06,
      "loss": 0.5339,
      "step": 11479
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9816557765007019,
      "learning_rate": 1.8264971912168744e-06,
      "loss": 0.4873,
      "step": 11480
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0051323175430298,
      "learning_rate": 1.825704413030015e-06,
      "loss": 0.517,
      "step": 11481
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2196458578109741,
      "learning_rate": 1.8249117685016983e-06,
      "loss": 0.5198,
      "step": 11482
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1237990856170654,
      "learning_rate": 1.8241192576653e-06,
      "loss": 0.5429,
      "step": 11483
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.209991216659546,
      "learning_rate": 1.8233268805541953e-06,
      "loss": 0.451,
      "step": 11484
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0044230222702026,
      "learning_rate": 1.8225346372017432e-06,
      "loss": 0.492,
      "step": 11485
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0598922967910767,
      "learning_rate": 1.8217425276413037e-06,
      "loss": 0.5012,
      "step": 11486
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0145241022109985,
      "learning_rate": 1.8209505519062299e-06,
      "loss": 0.48,
      "step": 11487
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.007987141609192,
      "learning_rate": 1.8201587100298694e-06,
      "loss": 0.4696,
      "step": 11488
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0017046928405762,
      "learning_rate": 1.8193670020455656e-06,
      "loss": 0.4911,
      "step": 11489
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1037015914916992,
      "learning_rate": 1.8185754279866508e-06,
      "loss": 0.4793,
      "step": 11490
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0826473236083984,
      "learning_rate": 1.8177839878864562e-06,
      "loss": 0.4515,
      "step": 11491
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0383855104446411,
      "learning_rate": 1.8169926817783106e-06,
      "loss": 0.513,
      "step": 11492
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1208995580673218,
      "learning_rate": 1.8162015096955288e-06,
      "loss": 0.5273,
      "step": 11493
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0882205963134766,
      "learning_rate": 1.8154104716714254e-06,
      "loss": 0.5168,
      "step": 11494
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0574021339416504,
      "learning_rate": 1.814619567739309e-06,
      "loss": 0.4873,
      "step": 11495
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.085392713546753,
      "learning_rate": 1.8138287979324815e-06,
      "loss": 0.496,
      "step": 11496
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9917359352111816,
      "learning_rate": 1.8130381622842414e-06,
      "loss": 0.4118,
      "step": 11497
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0514519214630127,
      "learning_rate": 1.8122476608278755e-06,
      "loss": 0.5219,
      "step": 11498
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9837106466293335,
      "learning_rate": 1.8114572935966713e-06,
      "loss": 0.4996,
      "step": 11499
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1096519231796265,
      "learning_rate": 1.8106670606239086e-06,
      "loss": 0.4988,
      "step": 11500
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9787556529045105,
      "learning_rate": 1.8098769619428607e-06,
      "loss": 0.4731,
      "step": 11501
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.083164930343628,
      "learning_rate": 1.8090869975867964e-06,
      "loss": 0.4633,
      "step": 11502
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0905555486679077,
      "learning_rate": 1.8082971675889798e-06,
      "loss": 0.5448,
      "step": 11503
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0619641542434692,
      "learning_rate": 1.8075074719826636e-06,
      "loss": 0.4995,
      "step": 11504
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9724273085594177,
      "learning_rate": 1.8067179108011047e-06,
      "loss": 0.4842,
      "step": 11505
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0353304147720337,
      "learning_rate": 1.8059284840775443e-06,
      "loss": 0.4954,
      "step": 11506
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9989253878593445,
      "learning_rate": 1.8051391918452244e-06,
      "loss": 0.5239,
      "step": 11507
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.065077543258667,
      "learning_rate": 1.8043500341373788e-06,
      "loss": 0.4882,
      "step": 11508
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9911226630210876,
      "learning_rate": 1.8035610109872364e-06,
      "loss": 0.5167,
      "step": 11509
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0643283128738403,
      "learning_rate": 1.8027721224280204e-06,
      "loss": 0.5408,
      "step": 11510
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0329303741455078,
      "learning_rate": 1.8019833684929493e-06,
      "loss": 0.5091,
      "step": 11511
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0223429203033447,
      "learning_rate": 1.8011947492152303e-06,
      "loss": 0.5148,
      "step": 11512
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0493983030319214,
      "learning_rate": 1.8004062646280762e-06,
      "loss": 0.4806,
      "step": 11513
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0583444833755493,
      "learning_rate": 1.799617914764682e-06,
      "loss": 0.4884,
      "step": 11514
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1490446329116821,
      "learning_rate": 1.7988296996582438e-06,
      "loss": 0.5446,
      "step": 11515
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.089064359664917,
      "learning_rate": 1.7980416193419509e-06,
      "loss": 0.4835,
      "step": 11516
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0678657293319702,
      "learning_rate": 1.7972536738489865e-06,
      "loss": 0.5207,
      "step": 11517
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.07902991771698,
      "learning_rate": 1.7964658632125286e-06,
      "loss": 0.5375,
      "step": 11518
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9613833427429199,
      "learning_rate": 1.7956781874657508e-06,
      "loss": 0.4995,
      "step": 11519
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0828293561935425,
      "learning_rate": 1.7948906466418154e-06,
      "loss": 0.5334,
      "step": 11520
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9815015196800232,
      "learning_rate": 1.7941032407738857e-06,
      "loss": 0.457,
      "step": 11521
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0951595306396484,
      "learning_rate": 1.7933159698951153e-06,
      "loss": 0.5049,
      "step": 11522
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.978574275970459,
      "learning_rate": 1.7925288340386543e-06,
      "loss": 0.5281,
      "step": 11523
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9065704345703125,
      "learning_rate": 1.7917418332376463e-06,
      "loss": 0.4371,
      "step": 11524
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0113238096237183,
      "learning_rate": 1.7909549675252291e-06,
      "loss": 0.5222,
      "step": 11525
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0330299139022827,
      "learning_rate": 1.7901682369345346e-06,
      "loss": 0.4832,
      "step": 11526
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0719724893569946,
      "learning_rate": 1.7893816414986915e-06,
      "loss": 0.4994,
      "step": 11527
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0883289575576782,
      "learning_rate": 1.7885951812508163e-06,
      "loss": 0.5058,
      "step": 11528
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1325196027755737,
      "learning_rate": 1.787808856224027e-06,
      "loss": 0.5706,
      "step": 11529
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0494288206100464,
      "learning_rate": 1.7870226664514318e-06,
      "loss": 0.4965,
      "step": 11530
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1580727100372314,
      "learning_rate": 1.786236611966135e-06,
      "loss": 0.5332,
      "step": 11531
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1171400547027588,
      "learning_rate": 1.7854506928012349e-06,
      "loss": 0.4641,
      "step": 11532
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0406197309494019,
      "learning_rate": 1.784664908989825e-06,
      "loss": 0.4893,
      "step": 11533
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0887912511825562,
      "learning_rate": 1.7838792605649874e-06,
      "loss": 0.5642,
      "step": 11534
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0635976791381836,
      "learning_rate": 1.7830937475598092e-06,
      "loss": 0.4902,
      "step": 11535
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1203384399414062,
      "learning_rate": 1.7823083700073607e-06,
      "loss": 0.5024,
      "step": 11536
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9832225441932678,
      "learning_rate": 1.781523127940713e-06,
      "loss": 0.49,
      "step": 11537
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9907360076904297,
      "learning_rate": 1.7807380213929304e-06,
      "loss": 0.4781,
      "step": 11538
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.090733289718628,
      "learning_rate": 1.7799530503970707e-06,
      "loss": 0.5082,
      "step": 11539
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0755808353424072,
      "learning_rate": 1.7791682149861866e-06,
      "loss": 0.5073,
      "step": 11540
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.15406334400177,
      "learning_rate": 1.778383515193326e-06,
      "loss": 0.5331,
      "step": 11541
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0282824039459229,
      "learning_rate": 1.777598951051525e-06,
      "loss": 0.4369,
      "step": 11542
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0402719974517822,
      "learning_rate": 1.7768145225938254e-06,
      "loss": 0.4479,
      "step": 11543
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.085084080696106,
      "learning_rate": 1.7760302298532522e-06,
      "loss": 0.4458,
      "step": 11544
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9971480369567871,
      "learning_rate": 1.7752460728628308e-06,
      "loss": 0.4439,
      "step": 11545
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0229544639587402,
      "learning_rate": 1.7744620516555804e-06,
      "loss": 0.5041,
      "step": 11546
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2048213481903076,
      "learning_rate": 1.7736781662645092e-06,
      "loss": 0.5566,
      "step": 11547
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1265819072723389,
      "learning_rate": 1.7728944167226287e-06,
      "loss": 0.5144,
      "step": 11548
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.080599069595337,
      "learning_rate": 1.772110803062939e-06,
      "loss": 0.5197,
      "step": 11549
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0882823467254639,
      "learning_rate": 1.7713273253184331e-06,
      "loss": 0.4827,
      "step": 11550
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0465631484985352,
      "learning_rate": 1.7705439835221022e-06,
      "loss": 0.5496,
      "step": 11551
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0857951641082764,
      "learning_rate": 1.7697607777069291e-06,
      "loss": 0.511,
      "step": 11552
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0986204147338867,
      "learning_rate": 1.7689777079058929e-06,
      "loss": 0.4755,
      "step": 11553
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0771892070770264,
      "learning_rate": 1.7681947741519668e-06,
      "loss": 0.4833,
      "step": 11554
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.156043291091919,
      "learning_rate": 1.7674119764781129e-06,
      "loss": 0.5587,
      "step": 11555
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0552773475646973,
      "learning_rate": 1.7666293149172969e-06,
      "loss": 0.5368,
      "step": 11556
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0510469675064087,
      "learning_rate": 1.7658467895024744e-06,
      "loss": 0.5071,
      "step": 11557
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.121157169342041,
      "learning_rate": 1.7650644002665906e-06,
      "loss": 0.5401,
      "step": 11558
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1302895545959473,
      "learning_rate": 1.7642821472425918e-06,
      "loss": 0.5189,
      "step": 11559
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1765938997268677,
      "learning_rate": 1.7635000304634154e-06,
      "loss": 0.4987,
      "step": 11560
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9780496954917908,
      "learning_rate": 1.762718049961994e-06,
      "loss": 0.4999,
      "step": 11561
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.053433895111084,
      "learning_rate": 1.7619362057712552e-06,
      "loss": 0.4768,
      "step": 11562
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0024325847625732,
      "learning_rate": 1.761154497924117e-06,
      "loss": 0.4808,
      "step": 11563
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0675638914108276,
      "learning_rate": 1.7603729264534936e-06,
      "loss": 0.4671,
      "step": 11564
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.020726203918457,
      "learning_rate": 1.7595914913923001e-06,
      "loss": 0.5356,
      "step": 11565
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.047310471534729,
      "learning_rate": 1.7588101927734346e-06,
      "loss": 0.5275,
      "step": 11566
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9428762197494507,
      "learning_rate": 1.7580290306297965e-06,
      "loss": 0.5259,
      "step": 11567
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.005489706993103,
      "learning_rate": 1.7572480049942781e-06,
      "loss": 0.535,
      "step": 11568
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0362120866775513,
      "learning_rate": 1.7564671158997653e-06,
      "loss": 0.4989,
      "step": 11569
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0559964179992676,
      "learning_rate": 1.755686363379141e-06,
      "loss": 0.5146,
      "step": 11570
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1007598638534546,
      "learning_rate": 1.7549057474652753e-06,
      "loss": 0.5642,
      "step": 11571
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0522035360336304,
      "learning_rate": 1.7541252681910386e-06,
      "loss": 0.5137,
      "step": 11572
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.049686312675476,
      "learning_rate": 1.7533449255892986e-06,
      "loss": 0.4666,
      "step": 11573
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1081347465515137,
      "learning_rate": 1.7525647196929079e-06,
      "loss": 0.5185,
      "step": 11574
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.145727515220642,
      "learning_rate": 1.7517846505347197e-06,
      "loss": 0.4984,
      "step": 11575
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0616527795791626,
      "learning_rate": 1.751004718147582e-06,
      "loss": 0.4923,
      "step": 11576
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0432676076889038,
      "learning_rate": 1.7502249225643291e-06,
      "loss": 0.4889,
      "step": 11577
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0427736043930054,
      "learning_rate": 1.7494452638178039e-06,
      "loss": 0.4669,
      "step": 11578
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.017297625541687,
      "learning_rate": 1.7486657419408287e-06,
      "loss": 0.4975,
      "step": 11579
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0261050462722778,
      "learning_rate": 1.7478863569662286e-06,
      "loss": 0.4811,
      "step": 11580
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.107507348060608,
      "learning_rate": 1.7471071089268204e-06,
      "loss": 0.5162,
      "step": 11581
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0576726198196411,
      "learning_rate": 1.7463279978554166e-06,
      "loss": 0.5206,
      "step": 11582
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9989815354347229,
      "learning_rate": 1.745549023784821e-06,
      "loss": 0.4535,
      "step": 11583
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1163592338562012,
      "learning_rate": 1.7447701867478372e-06,
      "loss": 0.4898,
      "step": 11584
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1261216402053833,
      "learning_rate": 1.7439914867772529e-06,
      "loss": 0.4587,
      "step": 11585
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0687371492385864,
      "learning_rate": 1.7432129239058637e-06,
      "loss": 0.4969,
      "step": 11586
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0555487871170044,
      "learning_rate": 1.7424344981664475e-06,
      "loss": 0.541,
      "step": 11587
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1022698879241943,
      "learning_rate": 1.7416562095917822e-06,
      "loss": 0.4831,
      "step": 11588
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9749317169189453,
      "learning_rate": 1.7408780582146383e-06,
      "loss": 0.52,
      "step": 11589
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0140492916107178,
      "learning_rate": 1.7401000440677824e-06,
      "loss": 0.4737,
      "step": 11590
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0857648849487305,
      "learning_rate": 1.7393221671839727e-06,
      "loss": 0.4885,
      "step": 11591
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1199688911437988,
      "learning_rate": 1.7385444275959657e-06,
      "loss": 0.4537,
      "step": 11592
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0685760974884033,
      "learning_rate": 1.7377668253365054e-06,
      "loss": 0.5195,
      "step": 11593
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0911962985992432,
      "learning_rate": 1.7369893604383353e-06,
      "loss": 0.5048,
      "step": 11594
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0440233945846558,
      "learning_rate": 1.736212032934192e-06,
      "loss": 0.4949,
      "step": 11595
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0071343183517456,
      "learning_rate": 1.7354348428568063e-06,
      "loss": 0.4748,
      "step": 11596
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.161497950553894,
      "learning_rate": 1.7346577902389028e-06,
      "loss": 0.5158,
      "step": 11597
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.056105613708496,
      "learning_rate": 1.7338808751132002e-06,
      "loss": 0.5007,
      "step": 11598
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0153801441192627,
      "learning_rate": 1.7331040975124125e-06,
      "loss": 0.4892,
      "step": 11599
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0123332738876343,
      "learning_rate": 1.7323274574692479e-06,
      "loss": 0.4661,
      "step": 11600
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1221553087234497,
      "learning_rate": 1.7315509550164044e-06,
      "loss": 0.4927,
      "step": 11601
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0159032344818115,
      "learning_rate": 1.730774590186579e-06,
      "loss": 0.5677,
      "step": 11602
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0440219640731812,
      "learning_rate": 1.7299983630124663e-06,
      "loss": 0.5028,
      "step": 11603
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0416673421859741,
      "learning_rate": 1.729222273526745e-06,
      "loss": 0.5041,
      "step": 11604
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0251445770263672,
      "learning_rate": 1.7284463217620955e-06,
      "loss": 0.4876,
      "step": 11605
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.074592113494873,
      "learning_rate": 1.727670507751193e-06,
      "loss": 0.4929,
      "step": 11606
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1091501712799072,
      "learning_rate": 1.7268948315266975e-06,
      "loss": 0.5498,
      "step": 11607
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0287474393844604,
      "learning_rate": 1.7261192931212783e-06,
      "loss": 0.532,
      "step": 11608
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.012856125831604,
      "learning_rate": 1.7253438925675847e-06,
      "loss": 0.4625,
      "step": 11609
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1315114498138428,
      "learning_rate": 1.7245686298982678e-06,
      "loss": 0.5651,
      "step": 11610
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0674339532852173,
      "learning_rate": 1.723793505145972e-06,
      "loss": 0.5238,
      "step": 11611
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0249159336090088,
      "learning_rate": 1.7230185183433345e-06,
      "loss": 0.4844,
      "step": 11612
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9884284734725952,
      "learning_rate": 1.722243669522987e-06,
      "loss": 0.4926,
      "step": 11613
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0254402160644531,
      "learning_rate": 1.7214689587175582e-06,
      "loss": 0.5323,
      "step": 11614
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9610978364944458,
      "learning_rate": 1.720694385959663e-06,
      "loss": 0.4544,
      "step": 11615
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9989670515060425,
      "learning_rate": 1.7199199512819225e-06,
      "loss": 0.4831,
      "step": 11616
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0955671072006226,
      "learning_rate": 1.7191456547169405e-06,
      "loss": 0.5301,
      "step": 11617
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0697169303894043,
      "learning_rate": 1.718371496297322e-06,
      "loss": 0.5327,
      "step": 11618
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.096488356590271,
      "learning_rate": 1.717597476055664e-06,
      "loss": 0.5321,
      "step": 11619
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0395913124084473,
      "learning_rate": 1.716823594024557e-06,
      "loss": 0.5056,
      "step": 11620
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0414578914642334,
      "learning_rate": 1.716049850236588e-06,
      "loss": 0.5046,
      "step": 11621
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.090161919593811,
      "learning_rate": 1.7152762447243365e-06,
      "loss": 0.5255,
      "step": 11622
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0255171060562134,
      "learning_rate": 1.7145027775203748e-06,
      "loss": 0.4791,
      "step": 11623
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0892038345336914,
      "learning_rate": 1.7137294486572714e-06,
      "loss": 0.5079,
      "step": 11624
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0260745286941528,
      "learning_rate": 1.7129562581675885e-06,
      "loss": 0.5071,
      "step": 11625
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9881934523582458,
      "learning_rate": 1.7121832060838833e-06,
      "loss": 0.488,
      "step": 11626
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.098183035850525,
      "learning_rate": 1.711410292438707e-06,
      "loss": 0.5461,
      "step": 11627
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0220118761062622,
      "learning_rate": 1.7106375172646e-06,
      "loss": 0.4528,
      "step": 11628
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0334930419921875,
      "learning_rate": 1.709864880594106e-06,
      "loss": 0.4964,
      "step": 11629
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.043442726135254,
      "learning_rate": 1.7090923824597578e-06,
      "loss": 0.5603,
      "step": 11630
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0600742101669312,
      "learning_rate": 1.70832002289408e-06,
      "loss": 0.5463,
      "step": 11631
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0279513597488403,
      "learning_rate": 1.7075478019295943e-06,
      "loss": 0.5061,
      "step": 11632
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.015267252922058,
      "learning_rate": 1.7067757195988178e-06,
      "loss": 0.4569,
      "step": 11633
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0748802423477173,
      "learning_rate": 1.706003775934259e-06,
      "loss": 0.5272,
      "step": 11634
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0735557079315186,
      "learning_rate": 1.705231970968424e-06,
      "loss": 0.5236,
      "step": 11635
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.057826280593872,
      "learning_rate": 1.704460304733806e-06,
      "loss": 0.5284,
      "step": 11636
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.074376106262207,
      "learning_rate": 1.7036887772629012e-06,
      "loss": 0.5452,
      "step": 11637
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.018315315246582,
      "learning_rate": 1.7029173885881973e-06,
      "loss": 0.5,
      "step": 11638
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.068433403968811,
      "learning_rate": 1.7021461387421705e-06,
      "loss": 0.4995,
      "step": 11639
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0389368534088135,
      "learning_rate": 1.7013750277572977e-06,
      "loss": 0.5143,
      "step": 11640
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.063865303993225,
      "learning_rate": 1.7006040556660468e-06,
      "loss": 0.522,
      "step": 11641
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.966158390045166,
      "learning_rate": 1.6998332225008817e-06,
      "loss": 0.4891,
      "step": 11642
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0541883707046509,
      "learning_rate": 1.6990625282942607e-06,
      "loss": 0.506,
      "step": 11643
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1151005029678345,
      "learning_rate": 1.6982919730786323e-06,
      "loss": 0.494,
      "step": 11644
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1117256879806519,
      "learning_rate": 1.697521556886441e-06,
      "loss": 0.534,
      "step": 11645
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.973153829574585,
      "learning_rate": 1.6967512797501317e-06,
      "loss": 0.498,
      "step": 11646
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0751382112503052,
      "learning_rate": 1.6959811417021338e-06,
      "loss": 0.4757,
      "step": 11647
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.03907310962677,
      "learning_rate": 1.6952111427748758e-06,
      "loss": 0.4955,
      "step": 11648
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0551830530166626,
      "learning_rate": 1.69444128300078e-06,
      "loss": 0.466,
      "step": 11649
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9973853230476379,
      "learning_rate": 1.6936715624122623e-06,
      "loss": 0.4897,
      "step": 11650
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0001341104507446,
      "learning_rate": 1.6929019810417352e-06,
      "loss": 0.4778,
      "step": 11651
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0439047813415527,
      "learning_rate": 1.6921325389215993e-06,
      "loss": 0.5034,
      "step": 11652
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0582904815673828,
      "learning_rate": 1.6913632360842553e-06,
      "loss": 0.5121,
      "step": 11653
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.020135760307312,
      "learning_rate": 1.6905940725620951e-06,
      "loss": 0.5069,
      "step": 11654
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0762420892715454,
      "learning_rate": 1.6898250483875063e-06,
      "loss": 0.5376,
      "step": 11655
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9812977313995361,
      "learning_rate": 1.6890561635928692e-06,
      "loss": 0.4486,
      "step": 11656
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0338646173477173,
      "learning_rate": 1.6882874182105613e-06,
      "loss": 0.482,
      "step": 11657
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.021016240119934,
      "learning_rate": 1.6875188122729458e-06,
      "loss": 0.4814,
      "step": 11658
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9645542502403259,
      "learning_rate": 1.6867503458123913e-06,
      "loss": 0.4765,
      "step": 11659
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0475287437438965,
      "learning_rate": 1.6859820188612557e-06,
      "loss": 0.5145,
      "step": 11660
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0525175333023071,
      "learning_rate": 1.6852138314518873e-06,
      "loss": 0.465,
      "step": 11661
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0847764015197754,
      "learning_rate": 1.6844457836166329e-06,
      "loss": 0.4694,
      "step": 11662
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1165729761123657,
      "learning_rate": 1.6836778753878324e-06,
      "loss": 0.5256,
      "step": 11663
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1158928871154785,
      "learning_rate": 1.68291010679782e-06,
      "loss": 0.4778,
      "step": 11664
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9745147228240967,
      "learning_rate": 1.6821424778789252e-06,
      "loss": 0.4828,
      "step": 11665
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0973376035690308,
      "learning_rate": 1.6813749886634657e-06,
      "loss": 0.5097,
      "step": 11666
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0163177251815796,
      "learning_rate": 1.6806076391837622e-06,
      "loss": 0.4506,
      "step": 11667
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0826618671417236,
      "learning_rate": 1.6798404294721254e-06,
      "loss": 0.4805,
      "step": 11668
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0028828382492065,
      "learning_rate": 1.6790733595608567e-06,
      "loss": 0.5133,
      "step": 11669
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1223706007003784,
      "learning_rate": 1.6783064294822559e-06,
      "loss": 0.4689,
      "step": 11670
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0388615131378174,
      "learning_rate": 1.677539639268616e-06,
      "loss": 0.5267,
      "step": 11671
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.052419900894165,
      "learning_rate": 1.6767729889522239e-06,
      "loss": 0.5219,
      "step": 11672
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0997850894927979,
      "learning_rate": 1.6760064785653624e-06,
      "loss": 0.5127,
      "step": 11673
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0978507995605469,
      "learning_rate": 1.675240108140303e-06,
      "loss": 0.5084,
      "step": 11674
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0522551536560059,
      "learning_rate": 1.674473877709315e-06,
      "loss": 0.4859,
      "step": 11675
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1092435121536255,
      "learning_rate": 1.6737077873046669e-06,
      "loss": 0.5307,
      "step": 11676
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.059883713722229,
      "learning_rate": 1.672941836958611e-06,
      "loss": 0.5186,
      "step": 11677
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0982457399368286,
      "learning_rate": 1.6721760267033998e-06,
      "loss": 0.5106,
      "step": 11678
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0341575145721436,
      "learning_rate": 1.6714103565712798e-06,
      "loss": 0.4907,
      "step": 11679
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0636166334152222,
      "learning_rate": 1.6706448265944902e-06,
      "loss": 0.5007,
      "step": 11680
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2040977478027344,
      "learning_rate": 1.6698794368052669e-06,
      "loss": 0.4697,
      "step": 11681
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0305360555648804,
      "learning_rate": 1.6691141872358336e-06,
      "loss": 0.5235,
      "step": 11682
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0078924894332886,
      "learning_rate": 1.668349077918413e-06,
      "loss": 0.4649,
      "step": 11683
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0714657306671143,
      "learning_rate": 1.6675841088852268e-06,
      "loss": 0.4572,
      "step": 11684
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0063605308532715,
      "learning_rate": 1.666819280168479e-06,
      "loss": 0.4524,
      "step": 11685
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.017889142036438,
      "learning_rate": 1.6660545918003762e-06,
      "loss": 0.4554,
      "step": 11686
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1632343530654907,
      "learning_rate": 1.6652900438131181e-06,
      "loss": 0.5496,
      "step": 11687
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.17890202999115,
      "learning_rate": 1.6645256362388922e-06,
      "loss": 0.4955,
      "step": 11688
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1335285902023315,
      "learning_rate": 1.663761369109892e-06,
      "loss": 0.5317,
      "step": 11689
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.197165608406067,
      "learning_rate": 1.662997242458293e-06,
      "loss": 0.5028,
      "step": 11690
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.125783920288086,
      "learning_rate": 1.6622332563162714e-06,
      "loss": 0.5791,
      "step": 11691
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0092953443527222,
      "learning_rate": 1.6614694107159962e-06,
      "loss": 0.4893,
      "step": 11692
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.010783314704895,
      "learning_rate": 1.6607057056896304e-06,
      "loss": 0.4617,
      "step": 11693
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0468823909759521,
      "learning_rate": 1.6599421412693307e-06,
      "loss": 0.5048,
      "step": 11694
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1088182926177979,
      "learning_rate": 1.65917871748725e-06,
      "loss": 0.5079,
      "step": 11695
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0559271574020386,
      "learning_rate": 1.6584154343755276e-06,
      "loss": 0.4817,
      "step": 11696
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0422968864440918,
      "learning_rate": 1.6576522919663107e-06,
      "loss": 0.4628,
      "step": 11697
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0454442501068115,
      "learning_rate": 1.6568892902917267e-06,
      "loss": 0.4989,
      "step": 11698
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0210545063018799,
      "learning_rate": 1.6561264293839051e-06,
      "loss": 0.473,
      "step": 11699
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1097790002822876,
      "learning_rate": 1.6553637092749685e-06,
      "loss": 0.4984,
      "step": 11700
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.015505075454712,
      "learning_rate": 1.6546011299970276e-06,
      "loss": 0.4598,
      "step": 11701
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0487347841262817,
      "learning_rate": 1.6538386915821975e-06,
      "loss": 0.4664,
      "step": 11702
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0431348085403442,
      "learning_rate": 1.6530763940625805e-06,
      "loss": 0.4781,
      "step": 11703
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1704440116882324,
      "learning_rate": 1.6523142374702722e-06,
      "loss": 0.534,
      "step": 11704
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0386048555374146,
      "learning_rate": 1.6515522218373658e-06,
      "loss": 0.4654,
      "step": 11705
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0807963609695435,
      "learning_rate": 1.6507903471959468e-06,
      "loss": 0.488,
      "step": 11706
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.065360426902771,
      "learning_rate": 1.6500286135780951e-06,
      "loss": 0.5316,
      "step": 11707
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0242459774017334,
      "learning_rate": 1.6492670210158863e-06,
      "loss": 0.5152,
      "step": 11708
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.048977255821228,
      "learning_rate": 1.6485055695413838e-06,
      "loss": 0.5036,
      "step": 11709
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0696985721588135,
      "learning_rate": 1.6477442591866544e-06,
      "loss": 0.4892,
      "step": 11710
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1575871706008911,
      "learning_rate": 1.6469830899837547e-06,
      "loss": 0.4745,
      "step": 11711
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0191328525543213,
      "learning_rate": 1.6462220619647306e-06,
      "loss": 0.5159,
      "step": 11712
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0086557865142822,
      "learning_rate": 1.6454611751616283e-06,
      "loss": 0.5013,
      "step": 11713
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0208491086959839,
      "learning_rate": 1.6447004296064867e-06,
      "loss": 0.4726,
      "step": 11714
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0429731607437134,
      "learning_rate": 1.6439398253313377e-06,
      "loss": 0.5131,
      "step": 11715
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9891948699951172,
      "learning_rate": 1.6431793623682096e-06,
      "loss": 0.4921,
      "step": 11716
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1223905086517334,
      "learning_rate": 1.642419040749119e-06,
      "loss": 0.5548,
      "step": 11717
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.060289978981018,
      "learning_rate": 1.6416588605060812e-06,
      "loss": 0.516,
      "step": 11718
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.227434515953064,
      "learning_rate": 1.6408988216711092e-06,
      "loss": 0.4774,
      "step": 11719
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9952710270881653,
      "learning_rate": 1.6401389242762006e-06,
      "loss": 0.4812,
      "step": 11720
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0805935859680176,
      "learning_rate": 1.639379168353354e-06,
      "loss": 0.4917,
      "step": 11721
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0161795616149902,
      "learning_rate": 1.6386195539345596e-06,
      "loss": 0.4974,
      "step": 11722
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9648051261901855,
      "learning_rate": 1.6378600810518026e-06,
      "loss": 0.5372,
      "step": 11723
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0348917245864868,
      "learning_rate": 1.6371007497370612e-06,
      "loss": 0.4883,
      "step": 11724
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.080922245979309,
      "learning_rate": 1.6363415600223103e-06,
      "loss": 0.5468,
      "step": 11725
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9747493267059326,
      "learning_rate": 1.6355825119395118e-06,
      "loss": 0.4795,
      "step": 11726
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0580079555511475,
      "learning_rate": 1.634823605520633e-06,
      "loss": 0.484,
      "step": 11727
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1747360229492188,
      "learning_rate": 1.634064840797624e-06,
      "loss": 0.4909,
      "step": 11728
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0317986011505127,
      "learning_rate": 1.6333062178024355e-06,
      "loss": 0.5099,
      "step": 11729
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9565672874450684,
      "learning_rate": 1.63254773656701e-06,
      "loss": 0.4824,
      "step": 11730
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0761483907699585,
      "learning_rate": 1.6317893971232852e-06,
      "loss": 0.4888,
      "step": 11731
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1051914691925049,
      "learning_rate": 1.6310311995031913e-06,
      "loss": 0.523,
      "step": 11732
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0961300134658813,
      "learning_rate": 1.6302731437386555e-06,
      "loss": 0.5789,
      "step": 11733
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1655371189117432,
      "learning_rate": 1.6295152298615936e-06,
      "loss": 0.4587,
      "step": 11734
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1272354125976562,
      "learning_rate": 1.62875745790392e-06,
      "loss": 0.5127,
      "step": 11735
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.126760482788086,
      "learning_rate": 1.6279998278975428e-06,
      "loss": 0.5392,
      "step": 11736
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0484238862991333,
      "learning_rate": 1.627242339874362e-06,
      "loss": 0.4885,
      "step": 11737
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0628280639648438,
      "learning_rate": 1.6264849938662753e-06,
      "loss": 0.5307,
      "step": 11738
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0625278949737549,
      "learning_rate": 1.6257277899051666e-06,
      "loss": 0.5333,
      "step": 11739
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0485646724700928,
      "learning_rate": 1.6249707280229237e-06,
      "loss": 0.5439,
      "step": 11740
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0634983777999878,
      "learning_rate": 1.6242138082514247e-06,
      "loss": 0.5308,
      "step": 11741
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0296636819839478,
      "learning_rate": 1.6234570306225366e-06,
      "loss": 0.5131,
      "step": 11742
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.01493239402771,
      "learning_rate": 1.6227003951681276e-06,
      "loss": 0.4532,
      "step": 11743
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.019639253616333,
      "learning_rate": 1.6219439019200557e-06,
      "loss": 0.4923,
      "step": 11744
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.104752779006958,
      "learning_rate": 1.6211875509101744e-06,
      "loss": 0.549,
      "step": 11745
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.04420804977417,
      "learning_rate": 1.6204313421703332e-06,
      "loss": 0.4891,
      "step": 11746
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0602904558181763,
      "learning_rate": 1.6196752757323698e-06,
      "loss": 0.5742,
      "step": 11747
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0057286024093628,
      "learning_rate": 1.61891935162812e-06,
      "loss": 0.4803,
      "step": 11748
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.003815770149231,
      "learning_rate": 1.6181635698894171e-06,
      "loss": 0.45,
      "step": 11749
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0611565113067627,
      "learning_rate": 1.61740793054808e-06,
      "loss": 0.5316,
      "step": 11750
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0178951025009155,
      "learning_rate": 1.6166524336359285e-06,
      "loss": 0.4699,
      "step": 11751
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1051855087280273,
      "learning_rate": 1.6158970791847728e-06,
      "loss": 0.5401,
      "step": 11752
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0515493154525757,
      "learning_rate": 1.6151418672264186e-06,
      "loss": 0.4843,
      "step": 11753
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0373625755310059,
      "learning_rate": 1.614386797792667e-06,
      "loss": 0.5156,
      "step": 11754
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9942763447761536,
      "learning_rate": 1.6136318709153075e-06,
      "loss": 0.4878,
      "step": 11755
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8941650986671448,
      "learning_rate": 1.612877086626129e-06,
      "loss": 0.538,
      "step": 11756
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0180784463882446,
      "learning_rate": 1.612122444956916e-06,
      "loss": 0.4957,
      "step": 11757
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.027891993522644,
      "learning_rate": 1.6113679459394398e-06,
      "loss": 0.52,
      "step": 11758
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0554982423782349,
      "learning_rate": 1.6106135896054714e-06,
      "loss": 0.5112,
      "step": 11759
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.08571195602417,
      "learning_rate": 1.6098593759867736e-06,
      "loss": 0.5015,
      "step": 11760
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.065076470375061,
      "learning_rate": 1.609105305115104e-06,
      "loss": 0.4743,
      "step": 11761
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0663999319076538,
      "learning_rate": 1.6083513770222158e-06,
      "loss": 0.5285,
      "step": 11762
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.05372154712677,
      "learning_rate": 1.6075975917398512e-06,
      "loss": 0.475,
      "step": 11763
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.063764214515686,
      "learning_rate": 1.60684394929975e-06,
      "loss": 0.5586,
      "step": 11764
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.027434229850769,
      "learning_rate": 1.6060904497336465e-06,
      "loss": 0.5199,
      "step": 11765
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0786575078964233,
      "learning_rate": 1.6053370930732676e-06,
      "loss": 0.4494,
      "step": 11766
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0548747777938843,
      "learning_rate": 1.6045838793503342e-06,
      "loss": 0.5112,
      "step": 11767
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0842057466506958,
      "learning_rate": 1.6038308085965642e-06,
      "loss": 0.5167,
      "step": 11768
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0827819108963013,
      "learning_rate": 1.6030778808436609e-06,
      "loss": 0.4692,
      "step": 11769
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9974976181983948,
      "learning_rate": 1.6023250961233338e-06,
      "loss": 0.5102,
      "step": 11770
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1199263334274292,
      "learning_rate": 1.6015724544672762e-06,
      "loss": 0.493,
      "step": 11771
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0420162677764893,
      "learning_rate": 1.6008199559071795e-06,
      "loss": 0.5241,
      "step": 11772
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0192173719406128,
      "learning_rate": 1.6000676004747306e-06,
      "loss": 0.4867,
      "step": 11773
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.05176842212677,
      "learning_rate": 1.5993153882016065e-06,
      "loss": 0.5473,
      "step": 11774
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0142534971237183,
      "learning_rate": 1.5985633191194821e-06,
      "loss": 0.5313,
      "step": 11775
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0849391222000122,
      "learning_rate": 1.5978113932600248e-06,
      "loss": 0.5187,
      "step": 11776
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.089290976524353,
      "learning_rate": 1.5970596106548913e-06,
      "loss": 0.5012,
      "step": 11777
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9700396060943604,
      "learning_rate": 1.5963079713357432e-06,
      "loss": 0.5173,
      "step": 11778
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0083379745483398,
      "learning_rate": 1.595556475334224e-06,
      "loss": 0.5048,
      "step": 11779
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0637929439544678,
      "learning_rate": 1.5948051226819783e-06,
      "loss": 0.4625,
      "step": 11780
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0219900608062744,
      "learning_rate": 1.5940539134106442e-06,
      "loss": 0.4917,
      "step": 11781
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.065132975578308,
      "learning_rate": 1.5933028475518486e-06,
      "loss": 0.4959,
      "step": 11782
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.069326639175415,
      "learning_rate": 1.5925519251372212e-06,
      "loss": 0.4303,
      "step": 11783
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9691741466522217,
      "learning_rate": 1.5918011461983796e-06,
      "loss": 0.4694,
      "step": 11784
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0848075151443481,
      "learning_rate": 1.5910505107669339e-06,
      "loss": 0.5011,
      "step": 11785
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0530424118041992,
      "learning_rate": 1.5903000188744922e-06,
      "loss": 0.4995,
      "step": 11786
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0618120431900024,
      "learning_rate": 1.589549670552656e-06,
      "loss": 0.521,
      "step": 11787
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9845284819602966,
      "learning_rate": 1.588799465833018e-06,
      "loss": 0.474,
      "step": 11788
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1666288375854492,
      "learning_rate": 1.5880494047471683e-06,
      "loss": 0.5411,
      "step": 11789
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9710365533828735,
      "learning_rate": 1.587299487326689e-06,
      "loss": 0.5239,
      "step": 11790
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0692170858383179,
      "learning_rate": 1.586549713603156e-06,
      "loss": 0.5279,
      "step": 11791
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0434986352920532,
      "learning_rate": 1.5858000836081422e-06,
      "loss": 0.4978,
      "step": 11792
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0952305793762207,
      "learning_rate": 1.5850505973732077e-06,
      "loss": 0.5001,
      "step": 11793
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0733871459960938,
      "learning_rate": 1.5843012549299131e-06,
      "loss": 0.4971,
      "step": 11794
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0665565729141235,
      "learning_rate": 1.58355205630981e-06,
      "loss": 0.5307,
      "step": 11795
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1024495363235474,
      "learning_rate": 1.5828030015444451e-06,
      "loss": 0.4628,
      "step": 11796
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.038615107536316,
      "learning_rate": 1.5820540906653581e-06,
      "loss": 0.5358,
      "step": 11797
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9799812436103821,
      "learning_rate": 1.5813053237040849e-06,
      "loss": 0.4921,
      "step": 11798
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.041947841644287,
      "learning_rate": 1.580556700692148e-06,
      "loss": 0.4849,
      "step": 11799
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0639641284942627,
      "learning_rate": 1.5798082216610766e-06,
      "loss": 0.4866,
      "step": 11800
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0486043691635132,
      "learning_rate": 1.5790598866423818e-06,
      "loss": 0.54,
      "step": 11801
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0393097400665283,
      "learning_rate": 1.5783116956675742e-06,
      "loss": 0.4939,
      "step": 11802
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0743504762649536,
      "learning_rate": 1.5775636487681579e-06,
      "loss": 0.4922,
      "step": 11803
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1121182441711426,
      "learning_rate": 1.5768157459756307e-06,
      "loss": 0.5391,
      "step": 11804
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0420786142349243,
      "learning_rate": 1.576067987321484e-06,
      "loss": 0.4904,
      "step": 11805
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9685152173042297,
      "learning_rate": 1.5753203728372052e-06,
      "loss": 0.4603,
      "step": 11806
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9745351672172546,
      "learning_rate": 1.5745729025542684e-06,
      "loss": 0.4562,
      "step": 11807
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1631194353103638,
      "learning_rate": 1.5738255765041537e-06,
      "loss": 0.5422,
      "step": 11808
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0341942310333252,
      "learning_rate": 1.5730783947183237e-06,
      "loss": 0.5079,
      "step": 11809
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0354013442993164,
      "learning_rate": 1.5723313572282412e-06,
      "loss": 0.448,
      "step": 11810
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0714869499206543,
      "learning_rate": 1.5715844640653627e-06,
      "loss": 0.5068,
      "step": 11811
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.095607876777649,
      "learning_rate": 1.5708377152611326e-06,
      "loss": 0.5168,
      "step": 11812
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0534653663635254,
      "learning_rate": 1.5700911108469986e-06,
      "loss": 0.4959,
      "step": 11813
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0473425388336182,
      "learning_rate": 1.569344650854398e-06,
      "loss": 0.4728,
      "step": 11814
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0482488870620728,
      "learning_rate": 1.5685983353147582e-06,
      "loss": 0.5058,
      "step": 11815
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0301032066345215,
      "learning_rate": 1.5678521642595052e-06,
      "loss": 0.4793,
      "step": 11816
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0333133935928345,
      "learning_rate": 1.567106137720058e-06,
      "loss": 0.4461,
      "step": 11817
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1130452156066895,
      "learning_rate": 1.5663602557278297e-06,
      "loss": 0.4937,
      "step": 11818
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.985506534576416,
      "learning_rate": 1.5656145183142274e-06,
      "loss": 0.4818,
      "step": 11819
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0618568658828735,
      "learning_rate": 1.5648689255106474e-06,
      "loss": 0.5234,
      "step": 11820
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1874841451644897,
      "learning_rate": 1.5641234773484887e-06,
      "loss": 0.5073,
      "step": 11821
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0233137607574463,
      "learning_rate": 1.5633781738591392e-06,
      "loss": 0.5044,
      "step": 11822
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1124851703643799,
      "learning_rate": 1.5626330150739776e-06,
      "loss": 0.5375,
      "step": 11823
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0393725633621216,
      "learning_rate": 1.5618880010243831e-06,
      "loss": 0.4793,
      "step": 11824
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0990042686462402,
      "learning_rate": 1.5611431317417235e-06,
      "loss": 0.5231,
      "step": 11825
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1121004819869995,
      "learning_rate": 1.5603984072573648e-06,
      "loss": 0.4983,
      "step": 11826
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0521689653396606,
      "learning_rate": 1.5596538276026641e-06,
      "loss": 0.462,
      "step": 11827
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0436952114105225,
      "learning_rate": 1.5589093928089715e-06,
      "loss": 0.5098,
      "step": 11828
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0428225994110107,
      "learning_rate": 1.5581651029076322e-06,
      "loss": 0.5065,
      "step": 11829
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.144033670425415,
      "learning_rate": 1.5574209579299903e-06,
      "loss": 0.5242,
      "step": 11830
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1208655834197998,
      "learning_rate": 1.5566769579073747e-06,
      "loss": 0.4954,
      "step": 11831
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1524823904037476,
      "learning_rate": 1.555933102871114e-06,
      "loss": 0.4988,
      "step": 11832
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.185436725616455,
      "learning_rate": 1.5551893928525285e-06,
      "loss": 0.5559,
      "step": 11833
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.037841558456421,
      "learning_rate": 1.5544458278829344e-06,
      "loss": 0.5449,
      "step": 11834
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.063841700553894,
      "learning_rate": 1.5537024079936425e-06,
      "loss": 0.478,
      "step": 11835
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0968282222747803,
      "learning_rate": 1.5529591332159511e-06,
      "loss": 0.5301,
      "step": 11836
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1035841703414917,
      "learning_rate": 1.5522160035811578e-06,
      "loss": 0.502,
      "step": 11837
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9873964190483093,
      "learning_rate": 1.551473019120558e-06,
      "loss": 0.5064,
      "step": 11838
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1052426099777222,
      "learning_rate": 1.5507301798654313e-06,
      "loss": 0.4939,
      "step": 11839
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0275089740753174,
      "learning_rate": 1.549987485847057e-06,
      "loss": 0.4861,
      "step": 11840
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9469333291053772,
      "learning_rate": 1.54924493709671e-06,
      "loss": 0.4617,
      "step": 11841
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0526026487350464,
      "learning_rate": 1.5485025336456511e-06,
      "loss": 0.5609,
      "step": 11842
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.032482624053955,
      "learning_rate": 1.547760275525147e-06,
      "loss": 0.4819,
      "step": 11843
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.04742431640625,
      "learning_rate": 1.547018162766446e-06,
      "loss": 0.546,
      "step": 11844
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1183147430419922,
      "learning_rate": 1.5462761954007987e-06,
      "loss": 0.5077,
      "step": 11845
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.044101357460022,
      "learning_rate": 1.5455343734594463e-06,
      "loss": 0.5132,
      "step": 11846
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1227136850357056,
      "learning_rate": 1.5447926969736237e-06,
      "loss": 0.5326,
      "step": 11847
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0950839519500732,
      "learning_rate": 1.5440511659745611e-06,
      "loss": 0.5109,
      "step": 11848
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1029894351959229,
      "learning_rate": 1.5433097804934833e-06,
      "loss": 0.4805,
      "step": 11849
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1721646785736084,
      "learning_rate": 1.5425685405616026e-06,
      "loss": 0.5316,
      "step": 11850
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.037771224975586,
      "learning_rate": 1.5418274462101358e-06,
      "loss": 0.4589,
      "step": 11851
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1142371892929077,
      "learning_rate": 1.541086497470284e-06,
      "loss": 0.5214,
      "step": 11852
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.152891993522644,
      "learning_rate": 1.540345694373247e-06,
      "loss": 0.4986,
      "step": 11853
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9909875988960266,
      "learning_rate": 1.5396050369502175e-06,
      "loss": 0.5031,
      "step": 11854
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2157238721847534,
      "learning_rate": 1.538864525232382e-06,
      "loss": 0.5067,
      "step": 11855
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2282987833023071,
      "learning_rate": 1.538124159250921e-06,
      "loss": 0.4681,
      "step": 11856
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0671184062957764,
      "learning_rate": 1.5373839390370098e-06,
      "loss": 0.5296,
      "step": 11857
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0432041883468628,
      "learning_rate": 1.5366438646218146e-06,
      "loss": 0.5212,
      "step": 11858
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0385440587997437,
      "learning_rate": 1.5359039360364975e-06,
      "loss": 0.484,
      "step": 11859
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9888316988945007,
      "learning_rate": 1.5351641533122153e-06,
      "loss": 0.4988,
      "step": 11860
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0522485971450806,
      "learning_rate": 1.5344245164801174e-06,
      "loss": 0.4948,
      "step": 11861
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0455278158187866,
      "learning_rate": 1.533685025571347e-06,
      "loss": 0.4481,
      "step": 11862
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0585129261016846,
      "learning_rate": 1.5329456806170418e-06,
      "loss": 0.5255,
      "step": 11863
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1429243087768555,
      "learning_rate": 1.5322064816483328e-06,
      "loss": 0.5375,
      "step": 11864
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0262020826339722,
      "learning_rate": 1.5314674286963471e-06,
      "loss": 0.5271,
      "step": 11865
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0965781211853027,
      "learning_rate": 1.5307285217922003e-06,
      "loss": 0.5291,
      "step": 11866
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.013327956199646,
      "learning_rate": 1.529989760967005e-06,
      "loss": 0.5029,
      "step": 11867
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.078499436378479,
      "learning_rate": 1.5292511462518728e-06,
      "loss": 0.4964,
      "step": 11868
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.977638304233551,
      "learning_rate": 1.528512677677899e-06,
      "loss": 0.4668,
      "step": 11869
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0083192586898804,
      "learning_rate": 1.5277743552761809e-06,
      "loss": 0.4853,
      "step": 11870
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0516363382339478,
      "learning_rate": 1.5270361790778065e-06,
      "loss": 0.5104,
      "step": 11871
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0042387247085571,
      "learning_rate": 1.526298149113854e-06,
      "loss": 0.5028,
      "step": 11872
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1472967863082886,
      "learning_rate": 1.5255602654154055e-06,
      "loss": 0.5059,
      "step": 11873
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1058728694915771,
      "learning_rate": 1.5248225280135258e-06,
      "loss": 0.4976,
      "step": 11874
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0069677829742432,
      "learning_rate": 1.5240849369392807e-06,
      "loss": 0.4857,
      "step": 11875
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1479973793029785,
      "learning_rate": 1.5233474922237268e-06,
      "loss": 0.5161,
      "step": 11876
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0414130687713623,
      "learning_rate": 1.5226101938979153e-06,
      "loss": 0.487,
      "step": 11877
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0197550058364868,
      "learning_rate": 1.5218730419928917e-06,
      "loss": 0.442,
      "step": 11878
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0681520700454712,
      "learning_rate": 1.5211360365396972e-06,
      "loss": 0.5254,
      "step": 11879
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.124697208404541,
      "learning_rate": 1.5203991775693577e-06,
      "loss": 0.5418,
      "step": 11880
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0970063209533691,
      "learning_rate": 1.5196624651129084e-06,
      "loss": 0.5125,
      "step": 11881
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1287860870361328,
      "learning_rate": 1.5189258992013635e-06,
      "loss": 0.5159,
      "step": 11882
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0907492637634277,
      "learning_rate": 1.5181894798657388e-06,
      "loss": 0.4972,
      "step": 11883
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0564488172531128,
      "learning_rate": 1.517453207137043e-06,
      "loss": 0.49,
      "step": 11884
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.044112205505371,
      "learning_rate": 1.5167170810462777e-06,
      "loss": 0.5233,
      "step": 11885
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0471843481063843,
      "learning_rate": 1.5159811016244392e-06,
      "loss": 0.5003,
      "step": 11886
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0996730327606201,
      "learning_rate": 1.5152452689025176e-06,
      "loss": 0.4893,
      "step": 11887
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9819523692131042,
      "learning_rate": 1.5145095829114937e-06,
      "loss": 0.4941,
      "step": 11888
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0815109014511108,
      "learning_rate": 1.5137740436823462e-06,
      "loss": 0.5005,
      "step": 11889
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.076339840888977,
      "learning_rate": 1.5130386512460454e-06,
      "loss": 0.541,
      "step": 11890
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0075737237930298,
      "learning_rate": 1.5123034056335572e-06,
      "loss": 0.5041,
      "step": 11891
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.055673360824585,
      "learning_rate": 1.5115683068758419e-06,
      "loss": 0.5106,
      "step": 11892
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0819106101989746,
      "learning_rate": 1.5108333550038461e-06,
      "loss": 0.5044,
      "step": 11893
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0794376134872437,
      "learning_rate": 1.510098550048521e-06,
      "loss": 0.5016,
      "step": 11894
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9727383852005005,
      "learning_rate": 1.5093638920408077e-06,
      "loss": 0.4704,
      "step": 11895
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0047531127929688,
      "learning_rate": 1.508629381011636e-06,
      "loss": 0.5153,
      "step": 11896
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0383678674697876,
      "learning_rate": 1.507895016991936e-06,
      "loss": 0.4804,
      "step": 11897
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.071780800819397,
      "learning_rate": 1.507160800012628e-06,
      "loss": 0.5098,
      "step": 11898
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.142799973487854,
      "learning_rate": 1.5064267301046281e-06,
      "loss": 0.5245,
      "step": 11899
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.138761043548584,
      "learning_rate": 1.5056928072988475e-06,
      "loss": 0.5021,
      "step": 11900
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0979869365692139,
      "learning_rate": 1.504959031626183e-06,
      "loss": 0.5337,
      "step": 11901
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1427834033966064,
      "learning_rate": 1.5042254031175373e-06,
      "loss": 0.5304,
      "step": 11902
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.061793565750122,
      "learning_rate": 1.5034919218038007e-06,
      "loss": 0.4581,
      "step": 11903
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0412780046463013,
      "learning_rate": 1.502758587715854e-06,
      "loss": 0.4771,
      "step": 11904
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.126717448234558,
      "learning_rate": 1.5020254008845775e-06,
      "loss": 0.5269,
      "step": 11905
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.048540711402893,
      "learning_rate": 1.501292361340842e-06,
      "loss": 0.523,
      "step": 11906
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9710584878921509,
      "learning_rate": 1.500559469115515e-06,
      "loss": 0.5073,
      "step": 11907
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0929603576660156,
      "learning_rate": 1.499826724239456e-06,
      "loss": 0.4943,
      "step": 11908
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.036075234413147,
      "learning_rate": 1.499094126743516e-06,
      "loss": 0.5468,
      "step": 11909
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9933010339736938,
      "learning_rate": 1.4983616766585423e-06,
      "loss": 0.463,
      "step": 11910
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1443263292312622,
      "learning_rate": 1.4976293740153803e-06,
      "loss": 0.5859,
      "step": 11911
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0146114826202393,
      "learning_rate": 1.4968972188448593e-06,
      "loss": 0.4801,
      "step": 11912
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0006181001663208,
      "learning_rate": 1.4961652111778103e-06,
      "loss": 0.4741,
      "step": 11913
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0206917524337769,
      "learning_rate": 1.4954333510450552e-06,
      "loss": 0.5021,
      "step": 11914
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9869930148124695,
      "learning_rate": 1.4947016384774105e-06,
      "loss": 0.4571,
      "step": 11915
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0013905763626099,
      "learning_rate": 1.4939700735056873e-06,
      "loss": 0.4808,
      "step": 11916
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0572274923324585,
      "learning_rate": 1.493238656160686e-06,
      "loss": 0.5408,
      "step": 11917
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9899517893791199,
      "learning_rate": 1.492507386473206e-06,
      "loss": 0.4837,
      "step": 11918
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0914475917816162,
      "learning_rate": 1.4917762644740381e-06,
      "loss": 0.5066,
      "step": 11919
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.005753755569458,
      "learning_rate": 1.4910452901939671e-06,
      "loss": 0.4579,
      "step": 11920
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0300661325454712,
      "learning_rate": 1.4903144636637723e-06,
      "loss": 0.4903,
      "step": 11921
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.06362783908844,
      "learning_rate": 1.489583784914228e-06,
      "loss": 0.4918,
      "step": 11922
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1783336400985718,
      "learning_rate": 1.4888532539760958e-06,
      "loss": 0.557,
      "step": 11923
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9961380362510681,
      "learning_rate": 1.4881228708801409e-06,
      "loss": 0.4597,
      "step": 11924
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.076972246170044,
      "learning_rate": 1.4873926356571144e-06,
      "loss": 0.5305,
      "step": 11925
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0707809925079346,
      "learning_rate": 1.486662548337764e-06,
      "loss": 0.5411,
      "step": 11926
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1027288436889648,
      "learning_rate": 1.485932608952832e-06,
      "loss": 0.4703,
      "step": 11927
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.047572374343872,
      "learning_rate": 1.485202817533053e-06,
      "loss": 0.4692,
      "step": 11928
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0753308534622192,
      "learning_rate": 1.4844731741091561e-06,
      "loss": 0.5282,
      "step": 11929
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0842903852462769,
      "learning_rate": 1.4837436787118665e-06,
      "loss": 0.5028,
      "step": 11930
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1072250604629517,
      "learning_rate": 1.4830143313718943e-06,
      "loss": 0.5216,
      "step": 11931
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0920616388320923,
      "learning_rate": 1.482285132119956e-06,
      "loss": 0.5112,
      "step": 11932
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9837995767593384,
      "learning_rate": 1.4815560809867551e-06,
      "loss": 0.5189,
      "step": 11933
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1012341976165771,
      "learning_rate": 1.4808271780029864e-06,
      "loss": 0.5073,
      "step": 11934
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0157415866851807,
      "learning_rate": 1.4800984231993432e-06,
      "loss": 0.4533,
      "step": 11935
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1307048797607422,
      "learning_rate": 1.47936981660651e-06,
      "loss": 0.5154,
      "step": 11936
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.076559066772461,
      "learning_rate": 1.4786413582551668e-06,
      "loss": 0.5612,
      "step": 11937
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.066542148590088,
      "learning_rate": 1.4779130481759874e-06,
      "loss": 0.5214,
      "step": 11938
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0553350448608398,
      "learning_rate": 1.4771848863996353e-06,
      "loss": 0.5669,
      "step": 11939
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.042103886604309,
      "learning_rate": 1.4764568729567714e-06,
      "loss": 0.4762,
      "step": 11940
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0295274257659912,
      "learning_rate": 1.4757290078780545e-06,
      "loss": 0.4596,
      "step": 11941
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1455708742141724,
      "learning_rate": 1.475001291194127e-06,
      "loss": 0.5457,
      "step": 11942
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.07152259349823,
      "learning_rate": 1.4742737229356324e-06,
      "loss": 0.5586,
      "step": 11943
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0826992988586426,
      "learning_rate": 1.473546303133207e-06,
      "loss": 0.5437,
      "step": 11944
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0262315273284912,
      "learning_rate": 1.4728190318174785e-06,
      "loss": 0.5024,
      "step": 11945
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0386667251586914,
      "learning_rate": 1.4720919090190723e-06,
      "loss": 0.4647,
      "step": 11946
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2081540822982788,
      "learning_rate": 1.471364934768601e-06,
      "loss": 0.5186,
      "step": 11947
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0901449918746948,
      "learning_rate": 1.470638109096676e-06,
      "loss": 0.4978,
      "step": 11948
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0483721494674683,
      "learning_rate": 1.469911432033906e-06,
      "loss": 0.4476,
      "step": 11949
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0376167297363281,
      "learning_rate": 1.469184903610883e-06,
      "loss": 0.4897,
      "step": 11950
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1253174543380737,
      "learning_rate": 1.468458523858201e-06,
      "loss": 0.4876,
      "step": 11951
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.062595009803772,
      "learning_rate": 1.467732292806447e-06,
      "loss": 0.5413,
      "step": 11952
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0641722679138184,
      "learning_rate": 1.4670062104861948e-06,
      "loss": 0.4689,
      "step": 11953
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0055220127105713,
      "learning_rate": 1.4662802769280244e-06,
      "loss": 0.5123,
      "step": 11954
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0673023462295532,
      "learning_rate": 1.4655544921624964e-06,
      "loss": 0.5036,
      "step": 11955
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0751420259475708,
      "learning_rate": 1.464828856220174e-06,
      "loss": 0.4862,
      "step": 11956
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0711745023727417,
      "learning_rate": 1.4641033691316104e-06,
      "loss": 0.5402,
      "step": 11957
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9954444766044617,
      "learning_rate": 1.4633780309273532e-06,
      "loss": 0.4982,
      "step": 11958
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.03201162815094,
      "learning_rate": 1.4626528416379438e-06,
      "loss": 0.526,
      "step": 11959
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0769315958023071,
      "learning_rate": 1.4619278012939197e-06,
      "loss": 0.5002,
      "step": 11960
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.043395757675171,
      "learning_rate": 1.4612029099258046e-06,
      "loss": 0.4672,
      "step": 11961
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1494218111038208,
      "learning_rate": 1.4604781675641273e-06,
      "loss": 0.4877,
      "step": 11962
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0968637466430664,
      "learning_rate": 1.4597535742393998e-06,
      "loss": 0.4801,
      "step": 11963
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0693919658660889,
      "learning_rate": 1.459029129982134e-06,
      "loss": 0.516,
      "step": 11964
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1250993013381958,
      "learning_rate": 1.4583048348228345e-06,
      "loss": 0.5083,
      "step": 11965
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0110074281692505,
      "learning_rate": 1.4575806887919951e-06,
      "loss": 0.5104,
      "step": 11966
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.988934338092804,
      "learning_rate": 1.456856691920111e-06,
      "loss": 0.5069,
      "step": 11967
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.084764003753662,
      "learning_rate": 1.4561328442376678e-06,
      "loss": 0.5018,
      "step": 11968
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9927456378936768,
      "learning_rate": 1.45540914577514e-06,
      "loss": 0.4934,
      "step": 11969
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.011318564414978,
      "learning_rate": 1.454685596563003e-06,
      "loss": 0.4814,
      "step": 11970
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0381759405136108,
      "learning_rate": 1.4539621966317219e-06,
      "loss": 0.4941,
      "step": 11971
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.016323208808899,
      "learning_rate": 1.4532389460117574e-06,
      "loss": 0.4807,
      "step": 11972
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1208360195159912,
      "learning_rate": 1.4525158447335635e-06,
      "loss": 0.4807,
      "step": 11973
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9977958798408508,
      "learning_rate": 1.4517928928275843e-06,
      "loss": 0.5039,
      "step": 11974
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0501761436462402,
      "learning_rate": 1.4510700903242642e-06,
      "loss": 0.4634,
      "step": 11975
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0842777490615845,
      "learning_rate": 1.4503474372540382e-06,
      "loss": 0.5476,
      "step": 11976
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0287162065505981,
      "learning_rate": 1.4496249336473318e-06,
      "loss": 0.4973,
      "step": 11977
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0873017311096191,
      "learning_rate": 1.4489025795345686e-06,
      "loss": 0.5014,
      "step": 11978
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0273690223693848,
      "learning_rate": 1.4481803749461643e-06,
      "loss": 0.5119,
      "step": 11979
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.057515025138855,
      "learning_rate": 1.4474583199125285e-06,
      "loss": 0.5005,
      "step": 11980
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9910272359848022,
      "learning_rate": 1.446736414464066e-06,
      "loss": 0.4732,
      "step": 11981
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0097696781158447,
      "learning_rate": 1.4460146586311713e-06,
      "loss": 0.5092,
      "step": 11982
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0665124654769897,
      "learning_rate": 1.4452930524442338e-06,
      "loss": 0.5283,
      "step": 11983
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.074441909790039,
      "learning_rate": 1.4445715959336432e-06,
      "loss": 0.4865,
      "step": 11984
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0354875326156616,
      "learning_rate": 1.4438502891297723e-06,
      "loss": 0.4793,
      "step": 11985
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0373681783676147,
      "learning_rate": 1.4431291320629953e-06,
      "loss": 0.4863,
      "step": 11986
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.163691520690918,
      "learning_rate": 1.4424081247636768e-06,
      "loss": 0.4999,
      "step": 11987
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0569572448730469,
      "learning_rate": 1.4416872672621762e-06,
      "loss": 0.5374,
      "step": 11988
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0788553953170776,
      "learning_rate": 1.440966559588846e-06,
      "loss": 0.5012,
      "step": 11989
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0290710926055908,
      "learning_rate": 1.4402460017740355e-06,
      "loss": 0.5061,
      "step": 11990
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0616074800491333,
      "learning_rate": 1.4395255938480785e-06,
      "loss": 0.4649,
      "step": 11991
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0956692695617676,
      "learning_rate": 1.4388053358413162e-06,
      "loss": 0.5222,
      "step": 11992
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9791175723075867,
      "learning_rate": 1.4380852277840712e-06,
      "loss": 0.4925,
      "step": 11993
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0590614080429077,
      "learning_rate": 1.437365269706666e-06,
      "loss": 0.5184,
      "step": 11994
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.033125877380371,
      "learning_rate": 1.436645461639416e-06,
      "loss": 0.4932,
      "step": 11995
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1277824640274048,
      "learning_rate": 1.4359258036126295e-06,
      "loss": 0.5071,
      "step": 11996
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0017791986465454,
      "learning_rate": 1.4352062956566088e-06,
      "loss": 0.4937,
      "step": 11997
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0337672233581543,
      "learning_rate": 1.4344869378016518e-06,
      "loss": 0.4753,
      "step": 11998
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0503156185150146,
      "learning_rate": 1.4337677300780445e-06,
      "loss": 0.529,
      "step": 11999
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.000065565109253,
      "learning_rate": 1.433048672516072e-06,
      "loss": 0.5215,
      "step": 12000
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0630111694335938,
      "learning_rate": 1.4323297651460117e-06,
      "loss": 0.5189,
      "step": 12001
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1379293203353882,
      "learning_rate": 1.4316110079981339e-06,
      "loss": 0.5171,
      "step": 12002
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.014651894569397,
      "learning_rate": 1.4308924011027042e-06,
      "loss": 0.5122,
      "step": 12003
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0537596940994263,
      "learning_rate": 1.430173944489977e-06,
      "loss": 0.4997,
      "step": 12004
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9623979330062866,
      "learning_rate": 1.4294556381902074e-06,
      "loss": 0.4396,
      "step": 12005
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0301682949066162,
      "learning_rate": 1.428737482233642e-06,
      "loss": 0.4786,
      "step": 12006
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0479462146759033,
      "learning_rate": 1.4280194766505156e-06,
      "loss": 0.507,
      "step": 12007
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.057776927947998,
      "learning_rate": 1.427301621471064e-06,
      "loss": 0.5253,
      "step": 12008
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0924488306045532,
      "learning_rate": 1.4265839167255114e-06,
      "loss": 0.5541,
      "step": 12009
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9652511477470398,
      "learning_rate": 1.42586636244408e-06,
      "loss": 0.4903,
      "step": 12010
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0759705305099487,
      "learning_rate": 1.4251489586569834e-06,
      "loss": 0.5713,
      "step": 12011
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0573580265045166,
      "learning_rate": 1.4244317053944268e-06,
      "loss": 0.5024,
      "step": 12012
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0048032999038696,
      "learning_rate": 1.423714602686611e-06,
      "loss": 0.4958,
      "step": 12013
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.005070447921753,
      "learning_rate": 1.4229976505637361e-06,
      "loss": 0.4997,
      "step": 12014
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.086395263671875,
      "learning_rate": 1.4222808490559842e-06,
      "loss": 0.5486,
      "step": 12015
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9975227117538452,
      "learning_rate": 1.4215641981935403e-06,
      "loss": 0.4314,
      "step": 12016
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1591862440109253,
      "learning_rate": 1.4208476980065794e-06,
      "loss": 0.5137,
      "step": 12017
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0774359703063965,
      "learning_rate": 1.420131348525271e-06,
      "loss": 0.5082,
      "step": 12018
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0310287475585938,
      "learning_rate": 1.4194151497797793e-06,
      "loss": 0.5299,
      "step": 12019
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.06497323513031,
      "learning_rate": 1.4186991018002582e-06,
      "loss": 0.4938,
      "step": 12020
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0487126111984253,
      "learning_rate": 1.4179832046168584e-06,
      "loss": 0.534,
      "step": 12021
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.084673523902893,
      "learning_rate": 1.417267458259728e-06,
      "loss": 0.509,
      "step": 12022
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0191333293914795,
      "learning_rate": 1.4165518627589991e-06,
      "loss": 0.4419,
      "step": 12023
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.062406063079834,
      "learning_rate": 1.4158364181448065e-06,
      "loss": 0.5042,
      "step": 12024
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1256499290466309,
      "learning_rate": 1.4151211244472734e-06,
      "loss": 0.5556,
      "step": 12025
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2402355670928955,
      "learning_rate": 1.414405981696519e-06,
      "loss": 0.5159,
      "step": 12026
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0793510675430298,
      "learning_rate": 1.4136909899226564e-06,
      "loss": 0.521,
      "step": 12027
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0173790454864502,
      "learning_rate": 1.412976149155789e-06,
      "loss": 0.4589,
      "step": 12028
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1383095979690552,
      "learning_rate": 1.412261459426018e-06,
      "loss": 0.4998,
      "step": 12029
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0682752132415771,
      "learning_rate": 1.4115469207634358e-06,
      "loss": 0.542,
      "step": 12030
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.086917519569397,
      "learning_rate": 1.4108325331981298e-06,
      "loss": 0.5032,
      "step": 12031
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9840065240859985,
      "learning_rate": 1.4101182967601796e-06,
      "loss": 0.535,
      "step": 12032
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.03810715675354,
      "learning_rate": 1.4094042114796613e-06,
      "loss": 0.4871,
      "step": 12033
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1853363513946533,
      "learning_rate": 1.4086902773866379e-06,
      "loss": 0.5153,
      "step": 12034
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0018763542175293,
      "learning_rate": 1.4079764945111767e-06,
      "loss": 0.5123,
      "step": 12035
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0280405282974243,
      "learning_rate": 1.407262862883328e-06,
      "loss": 0.4687,
      "step": 12036
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1003395318984985,
      "learning_rate": 1.4065493825331416e-06,
      "loss": 0.4533,
      "step": 12037
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0540413856506348,
      "learning_rate": 1.4058360534906607e-06,
      "loss": 0.5389,
      "step": 12038
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1108030080795288,
      "learning_rate": 1.4051228757859197e-06,
      "loss": 0.5359,
      "step": 12039
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0300624370574951,
      "learning_rate": 1.4044098494489494e-06,
      "loss": 0.4832,
      "step": 12040
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0136255025863647,
      "learning_rate": 1.4036969745097735e-06,
      "loss": 0.4871,
      "step": 12041
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0157047510147095,
      "learning_rate": 1.4029842509984043e-06,
      "loss": 0.4868,
      "step": 12042
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0380834341049194,
      "learning_rate": 1.4022716789448581e-06,
      "loss": 0.4896,
      "step": 12043
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0514111518859863,
      "learning_rate": 1.4015592583791343e-06,
      "loss": 0.4819,
      "step": 12044
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0694193840026855,
      "learning_rate": 1.4008469893312321e-06,
      "loss": 0.5212,
      "step": 12045
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0697076320648193,
      "learning_rate": 1.4001348718311446e-06,
      "loss": 0.4913,
      "step": 12046
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9682444930076599,
      "learning_rate": 1.399422905908851e-06,
      "loss": 0.4974,
      "step": 12047
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0870487689971924,
      "learning_rate": 1.3987110915943352e-06,
      "loss": 0.5247,
      "step": 12048
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.036061406135559,
      "learning_rate": 1.397999428917569e-06,
      "loss": 0.5026,
      "step": 12049
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2339755296707153,
      "learning_rate": 1.3972879179085147e-06,
      "loss": 0.5166,
      "step": 12050
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1082885265350342,
      "learning_rate": 1.396576558597133e-06,
      "loss": 0.5244,
      "step": 12051
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.032477855682373,
      "learning_rate": 1.3958653510133774e-06,
      "loss": 0.4679,
      "step": 12052
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1053731441497803,
      "learning_rate": 1.3951542951871938e-06,
      "loss": 0.5049,
      "step": 12053
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0226027965545654,
      "learning_rate": 1.3944433911485229e-06,
      "loss": 0.4656,
      "step": 12054
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.096422791481018,
      "learning_rate": 1.3937326389272977e-06,
      "loss": 0.5359,
      "step": 12055
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0210285186767578,
      "learning_rate": 1.3930220385534453e-06,
      "loss": 0.5203,
      "step": 12056
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0053527355194092,
      "learning_rate": 1.3923115900568896e-06,
      "loss": 0.5356,
      "step": 12057
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.033771276473999,
      "learning_rate": 1.3916012934675405e-06,
      "loss": 0.4959,
      "step": 12058
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0186415910720825,
      "learning_rate": 1.3908911488153081e-06,
      "loss": 0.5164,
      "step": 12059
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9885281324386597,
      "learning_rate": 1.3901811561300944e-06,
      "loss": 0.5108,
      "step": 12060
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0362237691879272,
      "learning_rate": 1.3894713154417944e-06,
      "loss": 0.4997,
      "step": 12061
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0876007080078125,
      "learning_rate": 1.3887616267802972e-06,
      "loss": 0.5109,
      "step": 12062
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0648605823516846,
      "learning_rate": 1.3880520901754874e-06,
      "loss": 0.511,
      "step": 12063
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1079448461532593,
      "learning_rate": 1.3873427056572354e-06,
      "loss": 0.5325,
      "step": 12064
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.069300651550293,
      "learning_rate": 1.386633473255418e-06,
      "loss": 0.5054,
      "step": 12065
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.086875557899475,
      "learning_rate": 1.3859243929998933e-06,
      "loss": 0.5346,
      "step": 12066
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.056308388710022,
      "learning_rate": 1.3852154649205201e-06,
      "loss": 0.4556,
      "step": 12067
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1065456867218018,
      "learning_rate": 1.3845066890471487e-06,
      "loss": 0.5237,
      "step": 12068
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0740257501602173,
      "learning_rate": 1.3837980654096229e-06,
      "loss": 0.5322,
      "step": 12069
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.046135425567627,
      "learning_rate": 1.383089594037781e-06,
      "loss": 0.4691,
      "step": 12070
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0684499740600586,
      "learning_rate": 1.3823812749614556e-06,
      "loss": 0.4921,
      "step": 12071
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0555635690689087,
      "learning_rate": 1.3816731082104668e-06,
      "loss": 0.5079,
      "step": 12072
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1421395540237427,
      "learning_rate": 1.3809650938146391e-06,
      "loss": 0.5579,
      "step": 12073
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0496187210083008,
      "learning_rate": 1.3802572318037804e-06,
      "loss": 0.479,
      "step": 12074
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0246468782424927,
      "learning_rate": 1.379549522207697e-06,
      "loss": 0.507,
      "step": 12075
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1108874082565308,
      "learning_rate": 1.3788419650561908e-06,
      "loss": 0.5469,
      "step": 12076
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.041033387184143,
      "learning_rate": 1.3781345603790485e-06,
      "loss": 0.4406,
      "step": 12077
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0171122550964355,
      "learning_rate": 1.3774273082060625e-06,
      "loss": 0.4816,
      "step": 12078
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0950605869293213,
      "learning_rate": 1.3767202085670118e-06,
      "loss": 0.5512,
      "step": 12079
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.026475429534912,
      "learning_rate": 1.3760132614916672e-06,
      "loss": 0.5266,
      "step": 12080
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1308759450912476,
      "learning_rate": 1.375306467009797e-06,
      "loss": 0.5612,
      "step": 12081
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1372259855270386,
      "learning_rate": 1.3745998251511622e-06,
      "loss": 0.5104,
      "step": 12082
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1877919435501099,
      "learning_rate": 1.373893335945517e-06,
      "loss": 0.4836,
      "step": 12083
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1031029224395752,
      "learning_rate": 1.373186999422611e-06,
      "loss": 0.5006,
      "step": 12084
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0803934335708618,
      "learning_rate": 1.3724808156121799e-06,
      "loss": 0.4791,
      "step": 12085
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9792132377624512,
      "learning_rate": 1.3717747845439645e-06,
      "loss": 0.4612,
      "step": 12086
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9646353125572205,
      "learning_rate": 1.371068906247693e-06,
      "loss": 0.4692,
      "step": 12087
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1279658079147339,
      "learning_rate": 1.3703631807530831e-06,
      "loss": 0.543,
      "step": 12088
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0468178987503052,
      "learning_rate": 1.3696576080898538e-06,
      "loss": 0.5678,
      "step": 12089
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.095420241355896,
      "learning_rate": 1.3689521882877137e-06,
      "loss": 0.489,
      "step": 12090
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.02595055103302,
      "learning_rate": 1.3682469213763655e-06,
      "loss": 0.5023,
      "step": 12091
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.052215337753296,
      "learning_rate": 1.367541807385507e-06,
      "loss": 0.4901,
      "step": 12092
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1547960042953491,
      "learning_rate": 1.3668368463448246e-06,
      "loss": 0.4822,
      "step": 12093
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0154019594192505,
      "learning_rate": 1.3661320382840026e-06,
      "loss": 0.4237,
      "step": 12094
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1761852502822876,
      "learning_rate": 1.3654273832327219e-06,
      "loss": 0.5079,
      "step": 12095
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1030158996582031,
      "learning_rate": 1.3647228812206493e-06,
      "loss": 0.5249,
      "step": 12096
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.056185007095337,
      "learning_rate": 1.3640185322774495e-06,
      "loss": 0.5108,
      "step": 12097
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0081676244735718,
      "learning_rate": 1.3633143364327812e-06,
      "loss": 0.461,
      "step": 12098
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9790708422660828,
      "learning_rate": 1.3626102937162943e-06,
      "loss": 0.4379,
      "step": 12099
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.081037163734436,
      "learning_rate": 1.3619064041576368e-06,
      "loss": 0.4806,
      "step": 12100
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.055122971534729,
      "learning_rate": 1.3612026677864426e-06,
      "loss": 0.5337,
      "step": 12101
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0966159105300903,
      "learning_rate": 1.360499084632344e-06,
      "loss": 0.5351,
      "step": 12102
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0062079429626465,
      "learning_rate": 1.3597956547249713e-06,
      "loss": 0.5121,
      "step": 12103
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0459173917770386,
      "learning_rate": 1.3590923780939386e-06,
      "loss": 0.5077,
      "step": 12104
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0308589935302734,
      "learning_rate": 1.3583892547688598e-06,
      "loss": 0.501,
      "step": 12105
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.011716604232788,
      "learning_rate": 1.357686284779343e-06,
      "loss": 0.4987,
      "step": 12106
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9909331202507019,
      "learning_rate": 1.3569834681549832e-06,
      "loss": 0.5138,
      "step": 12107
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0783926248550415,
      "learning_rate": 1.3562808049253795e-06,
      "loss": 0.5043,
      "step": 12108
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0912294387817383,
      "learning_rate": 1.3555782951201134e-06,
      "loss": 0.5251,
      "step": 12109
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.054231882095337,
      "learning_rate": 1.3548759387687683e-06,
      "loss": 0.5128,
      "step": 12110
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0848804712295532,
      "learning_rate": 1.3541737359009161e-06,
      "loss": 0.5027,
      "step": 12111
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.993877112865448,
      "learning_rate": 1.3534716865461256e-06,
      "loss": 0.4662,
      "step": 12112
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.960350751876831,
      "learning_rate": 1.3527697907339565e-06,
      "loss": 0.4709,
      "step": 12113
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9977060556411743,
      "learning_rate": 1.3520680484939651e-06,
      "loss": 0.491,
      "step": 12114
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1944913864135742,
      "learning_rate": 1.3513664598556952e-06,
      "loss": 0.4732,
      "step": 12115
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0275923013687134,
      "learning_rate": 1.3506650248486946e-06,
      "loss": 0.508,
      "step": 12116
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.158739447593689,
      "learning_rate": 1.3499637435024926e-06,
      "loss": 0.5093,
      "step": 12117
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1126782894134521,
      "learning_rate": 1.34926261584662e-06,
      "loss": 0.5013,
      "step": 12118
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9790872931480408,
      "learning_rate": 1.3485616419105985e-06,
      "loss": 0.5346,
      "step": 12119
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1475367546081543,
      "learning_rate": 1.3478608217239435e-06,
      "loss": 0.5139,
      "step": 12120
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.059121012687683,
      "learning_rate": 1.347160155316165e-06,
      "loss": 0.5058,
      "step": 12121
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0250189304351807,
      "learning_rate": 1.3464596427167663e-06,
      "loss": 0.4953,
      "step": 12122
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.021460771560669,
      "learning_rate": 1.3457592839552409e-06,
      "loss": 0.4608,
      "step": 12123
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0707124471664429,
      "learning_rate": 1.3450590790610795e-06,
      "loss": 0.5109,
      "step": 12124
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.124280333518982,
      "learning_rate": 1.3443590280637664e-06,
      "loss": 0.484,
      "step": 12125
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1179440021514893,
      "learning_rate": 1.3436591309927772e-06,
      "loss": 0.5157,
      "step": 12126
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0698933601379395,
      "learning_rate": 1.3429593878775825e-06,
      "loss": 0.4855,
      "step": 12127
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.113325834274292,
      "learning_rate": 1.342259798747646e-06,
      "loss": 0.4858,
      "step": 12128
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0449678897857666,
      "learning_rate": 1.3415603636324248e-06,
      "loss": 0.5135,
      "step": 12129
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0090608596801758,
      "learning_rate": 1.3408610825613722e-06,
      "loss": 0.4645,
      "step": 12130
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.122188925743103,
      "learning_rate": 1.340161955563928e-06,
      "loss": 0.4604,
      "step": 12131
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0805073976516724,
      "learning_rate": 1.339462982669531e-06,
      "loss": 0.4815,
      "step": 12132
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.982419490814209,
      "learning_rate": 1.3387641639076165e-06,
      "loss": 0.4846,
      "step": 12133
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0559958219528198,
      "learning_rate": 1.3380654993076054e-06,
      "loss": 0.4709,
      "step": 12134
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9687513709068298,
      "learning_rate": 1.3373669888989167e-06,
      "loss": 0.4518,
      "step": 12135
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9971938133239746,
      "learning_rate": 1.3366686327109645e-06,
      "loss": 0.5197,
      "step": 12136
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.090461254119873,
      "learning_rate": 1.3359704307731491e-06,
      "loss": 0.4876,
      "step": 12137
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0781184434890747,
      "learning_rate": 1.3352723831148761e-06,
      "loss": 0.4923,
      "step": 12138
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1592669486999512,
      "learning_rate": 1.3345744897655327e-06,
      "loss": 0.5068,
      "step": 12139
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0591822862625122,
      "learning_rate": 1.3338767507545064e-06,
      "loss": 0.4973,
      "step": 12140
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0624432563781738,
      "learning_rate": 1.3331791661111765e-06,
      "loss": 0.489,
      "step": 12141
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.091153621673584,
      "learning_rate": 1.3324817358649162e-06,
      "loss": 0.5274,
      "step": 12142
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0193672180175781,
      "learning_rate": 1.3317844600450912e-06,
      "loss": 0.4756,
      "step": 12143
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.069133996963501,
      "learning_rate": 1.3310873386810641e-06,
      "loss": 0.5254,
      "step": 12144
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0465642213821411,
      "learning_rate": 1.330390371802182e-06,
      "loss": 0.501,
      "step": 12145
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.132053017616272,
      "learning_rate": 1.3296935594377996e-06,
      "loss": 0.5368,
      "step": 12146
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0702149868011475,
      "learning_rate": 1.3289969016172515e-06,
      "loss": 0.5338,
      "step": 12147
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0705736875534058,
      "learning_rate": 1.3283003983698733e-06,
      "loss": 0.4942,
      "step": 12148
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1318503618240356,
      "learning_rate": 1.3276040497249926e-06,
      "loss": 0.4867,
      "step": 12149
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.042022705078125,
      "learning_rate": 1.3269078557119297e-06,
      "loss": 0.483,
      "step": 12150
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1017260551452637,
      "learning_rate": 1.3262118163599992e-06,
      "loss": 0.4648,
      "step": 12151
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0930343866348267,
      "learning_rate": 1.3255159316985105e-06,
      "loss": 0.4973,
      "step": 12152
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0908854007720947,
      "learning_rate": 1.3248202017567624e-06,
      "loss": 0.527,
      "step": 12153
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.092955470085144,
      "learning_rate": 1.32412462656405e-06,
      "loss": 0.5338,
      "step": 12154
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0554345846176147,
      "learning_rate": 1.3234292061496622e-06,
      "loss": 0.5016,
      "step": 12155
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0174444913864136,
      "learning_rate": 1.3227339405428807e-06,
      "loss": 0.5004,
      "step": 12156
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0531197786331177,
      "learning_rate": 1.3220388297729825e-06,
      "loss": 0.5208,
      "step": 12157
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.966866135597229,
      "learning_rate": 1.3213438738692313e-06,
      "loss": 0.4504,
      "step": 12158
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0735490322113037,
      "learning_rate": 1.320649072860894e-06,
      "loss": 0.5237,
      "step": 12159
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0257203578948975,
      "learning_rate": 1.3199544267772257e-06,
      "loss": 0.4605,
      "step": 12160
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0558432340621948,
      "learning_rate": 1.3192599356474733e-06,
      "loss": 0.5669,
      "step": 12161
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.050284504890442,
      "learning_rate": 1.318565599500881e-06,
      "loss": 0.545,
      "step": 12162
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0238032341003418,
      "learning_rate": 1.3178714183666846e-06,
      "loss": 0.5122,
      "step": 12163
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0815541744232178,
      "learning_rate": 1.3171773922741132e-06,
      "loss": 0.4996,
      "step": 12164
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.057782530784607,
      "learning_rate": 1.316483521252392e-06,
      "loss": 0.5329,
      "step": 12165
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9373903870582581,
      "learning_rate": 1.3157898053307322e-06,
      "loss": 0.4324,
      "step": 12166
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1545541286468506,
      "learning_rate": 1.3150962445383492e-06,
      "loss": 0.5078,
      "step": 12167
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0350265502929688,
      "learning_rate": 1.314402838904446e-06,
      "loss": 0.4906,
      "step": 12168
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0755099058151245,
      "learning_rate": 1.3137095884582163e-06,
      "loss": 0.556,
      "step": 12169
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9735145568847656,
      "learning_rate": 1.3130164932288524e-06,
      "loss": 0.4431,
      "step": 12170
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0998507738113403,
      "learning_rate": 1.3123235532455376e-06,
      "loss": 0.5312,
      "step": 12171
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0288559198379517,
      "learning_rate": 1.3116307685374497e-06,
      "loss": 0.4932,
      "step": 12172
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0293195247650146,
      "learning_rate": 1.3109381391337605e-06,
      "loss": 0.4914,
      "step": 12173
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1143139600753784,
      "learning_rate": 1.3102456650636314e-06,
      "loss": 0.5403,
      "step": 12174
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0851128101348877,
      "learning_rate": 1.3095533463562204e-06,
      "loss": 0.5424,
      "step": 12175
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1166521310806274,
      "learning_rate": 1.3088611830406828e-06,
      "loss": 0.4971,
      "step": 12176
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0827765464782715,
      "learning_rate": 1.3081691751461588e-06,
      "loss": 0.4561,
      "step": 12177
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1333348751068115,
      "learning_rate": 1.3074773227017878e-06,
      "loss": 0.5245,
      "step": 12178
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.097198247909546,
      "learning_rate": 1.3067856257367018e-06,
      "loss": 0.4648,
      "step": 12179
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0951207876205444,
      "learning_rate": 1.3060940842800247e-06,
      "loss": 0.5086,
      "step": 12180
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0213844776153564,
      "learning_rate": 1.3054026983608776e-06,
      "loss": 0.4805,
      "step": 12181
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0557488203048706,
      "learning_rate": 1.3047114680083683e-06,
      "loss": 0.4748,
      "step": 12182
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0473880767822266,
      "learning_rate": 1.3040203932516043e-06,
      "loss": 0.5136,
      "step": 12183
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0972284078598022,
      "learning_rate": 1.303329474119684e-06,
      "loss": 0.4723,
      "step": 12184
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1365244388580322,
      "learning_rate": 1.3026387106417e-06,
      "loss": 0.4823,
      "step": 12185
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.034995436668396,
      "learning_rate": 1.301948102846738e-06,
      "loss": 0.4946,
      "step": 12186
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0066113471984863,
      "learning_rate": 1.301257650763878e-06,
      "loss": 0.5085,
      "step": 12187
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1857848167419434,
      "learning_rate": 1.3005673544221882e-06,
      "loss": 0.5569,
      "step": 12188
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.049791693687439,
      "learning_rate": 1.299877213850741e-06,
      "loss": 0.4777,
      "step": 12189
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1054198741912842,
      "learning_rate": 1.2991872290785906e-06,
      "loss": 0.5321,
      "step": 12190
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.070559024810791,
      "learning_rate": 1.2984974001347922e-06,
      "loss": 0.4905,
      "step": 12191
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0235702991485596,
      "learning_rate": 1.2978077270483913e-06,
      "loss": 0.4836,
      "step": 12192
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0622280836105347,
      "learning_rate": 1.2971182098484286e-06,
      "loss": 0.4874,
      "step": 12193
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0791078805923462,
      "learning_rate": 1.2964288485639366e-06,
      "loss": 0.5058,
      "step": 12194
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.209069013595581,
      "learning_rate": 1.2957396432239427e-06,
      "loss": 0.5056,
      "step": 12195
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0712496042251587,
      "learning_rate": 1.2950505938574643e-06,
      "loss": 0.5171,
      "step": 12196
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.025773286819458,
      "learning_rate": 1.2943617004935176e-06,
      "loss": 0.4676,
      "step": 12197
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0884387493133545,
      "learning_rate": 1.2936729631611106e-06,
      "loss": 0.5218,
      "step": 12198
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1169272661209106,
      "learning_rate": 1.2929843818892401e-06,
      "loss": 0.4675,
      "step": 12199
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0363742113113403,
      "learning_rate": 1.2922959567069016e-06,
      "loss": 0.4974,
      "step": 12200
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0577375888824463,
      "learning_rate": 1.2916076876430821e-06,
      "loss": 0.5397,
      "step": 12201
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0551846027374268,
      "learning_rate": 1.2909195747267622e-06,
      "loss": 0.5299,
      "step": 12202
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.037851333618164,
      "learning_rate": 1.2902316179869179e-06,
      "loss": 0.5021,
      "step": 12203
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1661293506622314,
      "learning_rate": 1.2895438174525127e-06,
      "loss": 0.5248,
      "step": 12204
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0050801038742065,
      "learning_rate": 1.288856173152509e-06,
      "loss": 0.5077,
      "step": 12205
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.065712332725525,
      "learning_rate": 1.2881686851158642e-06,
      "loss": 0.556,
      "step": 12206
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.059660792350769,
      "learning_rate": 1.287481353371522e-06,
      "loss": 0.5284,
      "step": 12207
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0822430849075317,
      "learning_rate": 1.286794177948425e-06,
      "loss": 0.5514,
      "step": 12208
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1334432363510132,
      "learning_rate": 1.286107158875508e-06,
      "loss": 0.5393,
      "step": 12209
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9908674359321594,
      "learning_rate": 1.285420296181699e-06,
      "loss": 0.4934,
      "step": 12210
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1943225860595703,
      "learning_rate": 1.2847335898959207e-06,
      "loss": 0.4991,
      "step": 12211
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0061407089233398,
      "learning_rate": 1.284047040047085e-06,
      "loss": 0.5331,
      "step": 12212
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.099575161933899,
      "learning_rate": 1.2833606466641001e-06,
      "loss": 0.5291,
      "step": 12213
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.074028491973877,
      "learning_rate": 1.282674409775872e-06,
      "loss": 0.4862,
      "step": 12214
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9675447344779968,
      "learning_rate": 1.2819883294112918e-06,
      "loss": 0.4982,
      "step": 12215
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0054662227630615,
      "learning_rate": 1.2813024055992486e-06,
      "loss": 0.4806,
      "step": 12216
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0863933563232422,
      "learning_rate": 1.2806166383686258e-06,
      "loss": 0.5069,
      "step": 12217
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0222904682159424,
      "learning_rate": 1.2799310277482952e-06,
      "loss": 0.4901,
      "step": 12218
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1853152513504028,
      "learning_rate": 1.2792455737671306e-06,
      "loss": 0.5426,
      "step": 12219
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0142987966537476,
      "learning_rate": 1.27856027645399e-06,
      "loss": 0.4637,
      "step": 12220
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0342990159988403,
      "learning_rate": 1.27787513583773e-06,
      "loss": 0.447,
      "step": 12221
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0292028188705444,
      "learning_rate": 1.2771901519471997e-06,
      "loss": 0.5055,
      "step": 12222
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0016099214553833,
      "learning_rate": 1.2765053248112414e-06,
      "loss": 0.5186,
      "step": 12223
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1074825525283813,
      "learning_rate": 1.2758206544586909e-06,
      "loss": 0.5292,
      "step": 12224
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.079474687576294,
      "learning_rate": 1.2751361409183788e-06,
      "loss": 0.5449,
      "step": 12225
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.993061900138855,
      "learning_rate": 1.2744517842191228e-06,
      "loss": 0.5057,
      "step": 12226
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.108346700668335,
      "learning_rate": 1.2737675843897452e-06,
      "loss": 0.5171,
      "step": 12227
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0924155712127686,
      "learning_rate": 1.2730835414590498e-06,
      "loss": 0.5515,
      "step": 12228
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.986985445022583,
      "learning_rate": 1.272399655455842e-06,
      "loss": 0.4846,
      "step": 12229
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0078566074371338,
      "learning_rate": 1.2717159264089185e-06,
      "loss": 0.4736,
      "step": 12230
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1671011447906494,
      "learning_rate": 1.2710323543470648e-06,
      "loss": 0.4549,
      "step": 12231
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.081705093383789,
      "learning_rate": 1.2703489392990682e-06,
      "loss": 0.5101,
      "step": 12232
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.06071138381958,
      "learning_rate": 1.2696656812937047e-06,
      "loss": 0.5194,
      "step": 12233
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.013026237487793,
      "learning_rate": 1.268982580359741e-06,
      "loss": 0.4562,
      "step": 12234
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0899467468261719,
      "learning_rate": 1.2682996365259415e-06,
      "loss": 0.4819,
      "step": 12235
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0048573017120361,
      "learning_rate": 1.2676168498210623e-06,
      "loss": 0.5089,
      "step": 12236
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.022613286972046,
      "learning_rate": 1.2669342202738537e-06,
      "loss": 0.4961,
      "step": 12237
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0211726427078247,
      "learning_rate": 1.2662517479130605e-06,
      "loss": 0.4917,
      "step": 12238
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1397035121917725,
      "learning_rate": 1.2655694327674145e-06,
      "loss": 0.5077,
      "step": 12239
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0699193477630615,
      "learning_rate": 1.2648872748656498e-06,
      "loss": 0.4947,
      "step": 12240
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0356310606002808,
      "learning_rate": 1.2642052742364903e-06,
      "loss": 0.4952,
      "step": 12241
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0501240491867065,
      "learning_rate": 1.2635234309086486e-06,
      "loss": 0.5111,
      "step": 12242
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1141310930252075,
      "learning_rate": 1.2628417449108376e-06,
      "loss": 0.5439,
      "step": 12243
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1341676712036133,
      "learning_rate": 1.2621602162717594e-06,
      "loss": 0.4811,
      "step": 12244
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0090206861495972,
      "learning_rate": 1.261478845020112e-06,
      "loss": 0.4582,
      "step": 12245
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1057790517807007,
      "learning_rate": 1.2607976311845865e-06,
      "loss": 0.4488,
      "step": 12246
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0496375560760498,
      "learning_rate": 1.2601165747938638e-06,
      "loss": 0.4418,
      "step": 12247
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0294616222381592,
      "learning_rate": 1.2594356758766201e-06,
      "loss": 0.4762,
      "step": 12248
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0037492513656616,
      "learning_rate": 1.2587549344615308e-06,
      "loss": 0.4527,
      "step": 12249
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.045936107635498,
      "learning_rate": 1.2580743505772553e-06,
      "loss": 0.5087,
      "step": 12250
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0254948139190674,
      "learning_rate": 1.2573939242524508e-06,
      "loss": 0.4914,
      "step": 12251
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0283900499343872,
      "learning_rate": 1.2567136555157694e-06,
      "loss": 0.5073,
      "step": 12252
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0395734310150146,
      "learning_rate": 1.2560335443958533e-06,
      "loss": 0.4758,
      "step": 12253
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0320192575454712,
      "learning_rate": 1.2553535909213422e-06,
      "loss": 0.5084,
      "step": 12254
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0554685592651367,
      "learning_rate": 1.254673795120863e-06,
      "loss": 0.5101,
      "step": 12255
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.08400559425354,
      "learning_rate": 1.2539941570230402e-06,
      "loss": 0.477,
      "step": 12256
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9304841160774231,
      "learning_rate": 1.2533146766564946e-06,
      "loss": 0.4717,
      "step": 12257
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1072951555252075,
      "learning_rate": 1.252635354049833e-06,
      "loss": 0.5145,
      "step": 12258
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9975444674491882,
      "learning_rate": 1.2519561892316606e-06,
      "loss": 0.4958,
      "step": 12259
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1330645084381104,
      "learning_rate": 1.2512771822305742e-06,
      "loss": 0.504,
      "step": 12260
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1020914316177368,
      "learning_rate": 1.2505983330751654e-06,
      "loss": 0.4903,
      "step": 12261
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0297106504440308,
      "learning_rate": 1.2499196417940168e-06,
      "loss": 0.5439,
      "step": 12262
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.07819402217865,
      "learning_rate": 1.2492411084157086e-06,
      "loss": 0.5008,
      "step": 12263
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0378316640853882,
      "learning_rate": 1.2485627329688076e-06,
      "loss": 0.5036,
      "step": 12264
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.012473702430725,
      "learning_rate": 1.2478845154818798e-06,
      "loss": 0.492,
      "step": 12265
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0863486528396606,
      "learning_rate": 1.2472064559834818e-06,
      "loss": 0.5448,
      "step": 12266
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9467172026634216,
      "learning_rate": 1.2465285545021655e-06,
      "loss": 0.481,
      "step": 12267
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9862213134765625,
      "learning_rate": 1.2458508110664758e-06,
      "loss": 0.5023,
      "step": 12268
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0363272428512573,
      "learning_rate": 1.2451732257049458e-06,
      "loss": 0.5312,
      "step": 12269
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0616782903671265,
      "learning_rate": 1.2444957984461103e-06,
      "loss": 0.4936,
      "step": 12270
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.164757490158081,
      "learning_rate": 1.243818529318494e-06,
      "loss": 0.5477,
      "step": 12271
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9252405166625977,
      "learning_rate": 1.2431414183506114e-06,
      "loss": 0.4355,
      "step": 12272
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0829187631607056,
      "learning_rate": 1.2424644655709744e-06,
      "loss": 0.5246,
      "step": 12273
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0668288469314575,
      "learning_rate": 1.2417876710080872e-06,
      "loss": 0.4941,
      "step": 12274
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0414551496505737,
      "learning_rate": 1.2411110346904471e-06,
      "loss": 0.4964,
      "step": 12275
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0581361055374146,
      "learning_rate": 1.2404345566465464e-06,
      "loss": 0.5053,
      "step": 12276
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0122004747390747,
      "learning_rate": 1.2397582369048672e-06,
      "loss": 0.4787,
      "step": 12277
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0736596584320068,
      "learning_rate": 1.2390820754938859e-06,
      "loss": 0.527,
      "step": 12278
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0998107194900513,
      "learning_rate": 1.2384060724420776e-06,
      "loss": 0.56,
      "step": 12279
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0592037439346313,
      "learning_rate": 1.2377302277779029e-06,
      "loss": 0.4759,
      "step": 12280
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0291491746902466,
      "learning_rate": 1.2370545415298207e-06,
      "loss": 0.4809,
      "step": 12281
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0694981813430786,
      "learning_rate": 1.236379013726281e-06,
      "loss": 0.4992,
      "step": 12282
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0465306043624878,
      "learning_rate": 1.2357036443957283e-06,
      "loss": 0.5236,
      "step": 12283
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0182136297225952,
      "learning_rate": 1.2350284335666019e-06,
      "loss": 0.5117,
      "step": 12284
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.052598237991333,
      "learning_rate": 1.2343533812673286e-06,
      "loss": 0.485,
      "step": 12285
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1253355741500854,
      "learning_rate": 1.2336784875263341e-06,
      "loss": 0.4508,
      "step": 12286
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1140406131744385,
      "learning_rate": 1.233003752372039e-06,
      "loss": 0.52,
      "step": 12287
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9919499158859253,
      "learning_rate": 1.23232917583285e-06,
      "loss": 0.4785,
      "step": 12288
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.030924916267395,
      "learning_rate": 1.2316547579371724e-06,
      "loss": 0.4931,
      "step": 12289
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.014090657234192,
      "learning_rate": 1.230980498713404e-06,
      "loss": 0.4911,
      "step": 12290
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1351239681243896,
      "learning_rate": 1.2303063981899355e-06,
      "loss": 0.5271,
      "step": 12291
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0517010688781738,
      "learning_rate": 1.2296324563951517e-06,
      "loss": 0.5212,
      "step": 12292
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9599292874336243,
      "learning_rate": 1.2289586733574283e-06,
      "loss": 0.4797,
      "step": 12293
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0953619480133057,
      "learning_rate": 1.2282850491051363e-06,
      "loss": 0.4853,
      "step": 12294
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.040137529373169,
      "learning_rate": 1.2276115836666396e-06,
      "loss": 0.4951,
      "step": 12295
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0743566751480103,
      "learning_rate": 1.2269382770702964e-06,
      "loss": 0.5165,
      "step": 12296
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9729884266853333,
      "learning_rate": 1.2262651293444572e-06,
      "loss": 0.4798,
      "step": 12297
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1260862350463867,
      "learning_rate": 1.2255921405174664e-06,
      "loss": 0.5019,
      "step": 12298
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9777114391326904,
      "learning_rate": 1.2249193106176578e-06,
      "loss": 0.4741,
      "step": 12299
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9457767009735107,
      "learning_rate": 1.224246639673367e-06,
      "loss": 0.4812,
      "step": 12300
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0699979066848755,
      "learning_rate": 1.2235741277129143e-06,
      "loss": 0.5072,
      "step": 12301
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0439226627349854,
      "learning_rate": 1.2229017747646178e-06,
      "loss": 0.4934,
      "step": 12302
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1002904176712036,
      "learning_rate": 1.2222295808567874e-06,
      "loss": 0.5215,
      "step": 12303
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.052224040031433,
      "learning_rate": 1.2215575460177282e-06,
      "loss": 0.5105,
      "step": 12304
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.048862338066101,
      "learning_rate": 1.220885670275736e-06,
      "loss": 0.519,
      "step": 12305
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0804308652877808,
      "learning_rate": 1.2202139536591035e-06,
      "loss": 0.5023,
      "step": 12306
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.988083004951477,
      "learning_rate": 1.2195423961961089e-06,
      "loss": 0.511,
      "step": 12307
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.086592435836792,
      "learning_rate": 1.2188709979150366e-06,
      "loss": 0.515,
      "step": 12308
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1539126634597778,
      "learning_rate": 1.2181997588441507e-06,
      "loss": 0.5196,
      "step": 12309
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.000229001045227,
      "learning_rate": 1.2175286790117174e-06,
      "loss": 0.4879,
      "step": 12310
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0030760765075684,
      "learning_rate": 1.2168577584459944e-06,
      "loss": 0.5261,
      "step": 12311
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0990442037582397,
      "learning_rate": 1.2161869971752283e-06,
      "loss": 0.5291,
      "step": 12312
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.148339867591858,
      "learning_rate": 1.2155163952276654e-06,
      "loss": 0.5428,
      "step": 12313
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.043141484260559,
      "learning_rate": 1.2148459526315442e-06,
      "loss": 0.534,
      "step": 12314
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0665313005447388,
      "learning_rate": 1.2141756694150903e-06,
      "loss": 0.5132,
      "step": 12315
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.044087290763855,
      "learning_rate": 1.2135055456065292e-06,
      "loss": 0.47,
      "step": 12316
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1001490354537964,
      "learning_rate": 1.2128355812340776e-06,
      "loss": 0.5133,
      "step": 12317
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.056571125984192,
      "learning_rate": 1.2121657763259448e-06,
      "loss": 0.4569,
      "step": 12318
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0614756345748901,
      "learning_rate": 1.211496130910334e-06,
      "loss": 0.4624,
      "step": 12319
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0753448009490967,
      "learning_rate": 1.2108266450154422e-06,
      "loss": 0.5151,
      "step": 12320
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0368281602859497,
      "learning_rate": 1.2101573186694587e-06,
      "loss": 0.5118,
      "step": 12321
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.035905361175537,
      "learning_rate": 1.209488151900568e-06,
      "loss": 0.4938,
      "step": 12322
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1701600551605225,
      "learning_rate": 1.2088191447369436e-06,
      "loss": 0.5398,
      "step": 12323
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1100108623504639,
      "learning_rate": 1.2081502972067567e-06,
      "loss": 0.46,
      "step": 12324
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0216463804244995,
      "learning_rate": 1.2074816093381696e-06,
      "loss": 0.4951,
      "step": 12325
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0515730381011963,
      "learning_rate": 1.2068130811593387e-06,
      "loss": 0.4909,
      "step": 12326
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0393810272216797,
      "learning_rate": 1.2061447126984138e-06,
      "loss": 0.4834,
      "step": 12327
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0972084999084473,
      "learning_rate": 1.2054765039835382e-06,
      "loss": 0.5074,
      "step": 12328
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.992595374584198,
      "learning_rate": 1.2048084550428442e-06,
      "loss": 0.4412,
      "step": 12329
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1025792360305786,
      "learning_rate": 1.2041405659044664e-06,
      "loss": 0.5186,
      "step": 12330
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0410696268081665,
      "learning_rate": 1.203472836596523e-06,
      "loss": 0.4979,
      "step": 12331
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0511598587036133,
      "learning_rate": 1.2028052671471318e-06,
      "loss": 0.5426,
      "step": 12332
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1000758409500122,
      "learning_rate": 1.2021378575844005e-06,
      "loss": 0.5021,
      "step": 12333
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0427312850952148,
      "learning_rate": 1.201470607936433e-06,
      "loss": 0.4636,
      "step": 12334
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.076611042022705,
      "learning_rate": 1.2008035182313237e-06,
      "loss": 0.51,
      "step": 12335
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.98536616563797,
      "learning_rate": 1.2001365884971634e-06,
      "loss": 0.4575,
      "step": 12336
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1343871355056763,
      "learning_rate": 1.1994698187620297e-06,
      "loss": 0.5158,
      "step": 12337
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1061893701553345,
      "learning_rate": 1.1988032090540036e-06,
      "loss": 0.5396,
      "step": 12338
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0203694105148315,
      "learning_rate": 1.1981367594011496e-06,
      "loss": 0.5111,
      "step": 12339
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0427228212356567,
      "learning_rate": 1.1974704698315309e-06,
      "loss": 0.4431,
      "step": 12340
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.022687554359436,
      "learning_rate": 1.1968043403732044e-06,
      "loss": 0.493,
      "step": 12341
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0888246297836304,
      "learning_rate": 1.1961383710542135e-06,
      "loss": 0.5025,
      "step": 12342
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9961660504341125,
      "learning_rate": 1.1954725619026048e-06,
      "loss": 0.4856,
      "step": 12343
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9393571615219116,
      "learning_rate": 1.1948069129464128e-06,
      "loss": 0.5036,
      "step": 12344
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0751937627792358,
      "learning_rate": 1.1941414242136635e-06,
      "loss": 0.4683,
      "step": 12345
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0670708417892456,
      "learning_rate": 1.1934760957323782e-06,
      "loss": 0.495,
      "step": 12346
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.130021095275879,
      "learning_rate": 1.1928109275305734e-06,
      "loss": 0.5344,
      "step": 12347
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.95157390832901,
      "learning_rate": 1.1921459196362562e-06,
      "loss": 0.4909,
      "step": 12348
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0562176704406738,
      "learning_rate": 1.1914810720774289e-06,
      "loss": 0.517,
      "step": 12349
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0687183141708374,
      "learning_rate": 1.190816384882082e-06,
      "loss": 0.5289,
      "step": 12350
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0604063272476196,
      "learning_rate": 1.1901518580782073e-06,
      "loss": 0.5122,
      "step": 12351
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.074355125427246,
      "learning_rate": 1.1894874916937855e-06,
      "loss": 0.5271,
      "step": 12352
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9688604474067688,
      "learning_rate": 1.1888232857567888e-06,
      "loss": 0.4705,
      "step": 12353
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0829731225967407,
      "learning_rate": 1.1881592402951853e-06,
      "loss": 0.518,
      "step": 12354
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.146059513092041,
      "learning_rate": 1.1874953553369351e-06,
      "loss": 0.4958,
      "step": 12355
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.050464153289795,
      "learning_rate": 1.1868316309099937e-06,
      "loss": 0.4865,
      "step": 12356
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.046895980834961,
      "learning_rate": 1.186168067042308e-06,
      "loss": 0.5281,
      "step": 12357
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1184728145599365,
      "learning_rate": 1.1855046637618168e-06,
      "loss": 0.5722,
      "step": 12358
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1339380741119385,
      "learning_rate": 1.1848414210964526e-06,
      "loss": 0.5566,
      "step": 12359
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0997258424758911,
      "learning_rate": 1.1841783390741473e-06,
      "loss": 0.4693,
      "step": 12360
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.177841305732727,
      "learning_rate": 1.1835154177228165e-06,
      "loss": 0.4936,
      "step": 12361
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.146742582321167,
      "learning_rate": 1.1828526570703747e-06,
      "loss": 0.562,
      "step": 12362
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.091923713684082,
      "learning_rate": 1.1821900571447286e-06,
      "loss": 0.4978,
      "step": 12363
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0691144466400146,
      "learning_rate": 1.1815276179737778e-06,
      "loss": 0.4421,
      "step": 12364
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0965749025344849,
      "learning_rate": 1.1808653395854174e-06,
      "loss": 0.516,
      "step": 12365
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.002099633216858,
      "learning_rate": 1.1802032220075299e-06,
      "loss": 0.4794,
      "step": 12366
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1064283847808838,
      "learning_rate": 1.1795412652679955e-06,
      "loss": 0.523,
      "step": 12367
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9925868511199951,
      "learning_rate": 1.178879469394691e-06,
      "loss": 0.5006,
      "step": 12368
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0517547130584717,
      "learning_rate": 1.1782178344154776e-06,
      "loss": 0.4492,
      "step": 12369
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0431710481643677,
      "learning_rate": 1.1775563603582162e-06,
      "loss": 0.5079,
      "step": 12370
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9839174151420593,
      "learning_rate": 1.1768950472507605e-06,
      "loss": 0.4542,
      "step": 12371
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1189147233963013,
      "learning_rate": 1.1762338951209524e-06,
      "loss": 0.5292,
      "step": 12372
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1259547472000122,
      "learning_rate": 1.1755729039966358e-06,
      "loss": 0.5137,
      "step": 12373
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0271570682525635,
      "learning_rate": 1.174912073905638e-06,
      "loss": 0.4744,
      "step": 12374
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0150748491287231,
      "learning_rate": 1.174251404875787e-06,
      "loss": 0.4976,
      "step": 12375
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9843624830245972,
      "learning_rate": 1.1735908969349002e-06,
      "loss": 0.4748,
      "step": 12376
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.01045823097229,
      "learning_rate": 1.1729305501107897e-06,
      "loss": 0.4913,
      "step": 12377
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0301995277404785,
      "learning_rate": 1.1722703644312599e-06,
      "loss": 0.4744,
      "step": 12378
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0072860717773438,
      "learning_rate": 1.1716103399241113e-06,
      "loss": 0.479,
      "step": 12379
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1030186414718628,
      "learning_rate": 1.1709504766171298e-06,
      "loss": 0.4761,
      "step": 12380
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0800237655639648,
      "learning_rate": 1.170290774538107e-06,
      "loss": 0.5464,
      "step": 12381
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0646288394927979,
      "learning_rate": 1.1696312337148152e-06,
      "loss": 0.4738,
      "step": 12382
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.122464656829834,
      "learning_rate": 1.1689718541750278e-06,
      "loss": 0.5275,
      "step": 12383
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1252233982086182,
      "learning_rate": 1.168312635946508e-06,
      "loss": 0.5514,
      "step": 12384
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0471551418304443,
      "learning_rate": 1.1676535790570137e-06,
      "loss": 0.4772,
      "step": 12385
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1040979623794556,
      "learning_rate": 1.1669946835342956e-06,
      "loss": 0.4702,
      "step": 12386
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.230148434638977,
      "learning_rate": 1.1663359494060983e-06,
      "loss": 0.5516,
      "step": 12387
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.084335446357727,
      "learning_rate": 1.1656773767001566e-06,
      "loss": 0.529,
      "step": 12388
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1386499404907227,
      "learning_rate": 1.1650189654442024e-06,
      "loss": 0.4936,
      "step": 12389
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0516639947891235,
      "learning_rate": 1.1643607156659582e-06,
      "loss": 0.51,
      "step": 12390
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1283698081970215,
      "learning_rate": 1.1637026273931413e-06,
      "loss": 0.4913,
      "step": 12391
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.097561001777649,
      "learning_rate": 1.1630447006534606e-06,
      "loss": 0.5207,
      "step": 12392
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1390267610549927,
      "learning_rate": 1.1623869354746203e-06,
      "loss": 0.5143,
      "step": 12393
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1013120412826538,
      "learning_rate": 1.1617293318843164e-06,
      "loss": 0.5651,
      "step": 12394
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.054701805114746,
      "learning_rate": 1.1610718899102392e-06,
      "loss": 0.487,
      "step": 12395
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1136512756347656,
      "learning_rate": 1.1604146095800684e-06,
      "loss": 0.5257,
      "step": 12396
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.004216194152832,
      "learning_rate": 1.1597574909214808e-06,
      "loss": 0.5358,
      "step": 12397
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.12004554271698,
      "learning_rate": 1.159100533962147e-06,
      "loss": 0.4837,
      "step": 12398
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1613850593566895,
      "learning_rate": 1.1584437387297283e-06,
      "loss": 0.581,
      "step": 12399
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0097615718841553,
      "learning_rate": 1.157787105251879e-06,
      "loss": 0.4788,
      "step": 12400
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0033386945724487,
      "learning_rate": 1.157130633556251e-06,
      "loss": 0.4392,
      "step": 12401
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0544471740722656,
      "learning_rate": 1.1564743236704801e-06,
      "loss": 0.4531,
      "step": 12402
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1057604551315308,
      "learning_rate": 1.1558181756222081e-06,
      "loss": 0.4915,
      "step": 12403
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0928549766540527,
      "learning_rate": 1.1551621894390586e-06,
      "loss": 0.475,
      "step": 12404
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1400675773620605,
      "learning_rate": 1.1545063651486533e-06,
      "loss": 0.5263,
      "step": 12405
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1011698246002197,
      "learning_rate": 1.1538507027786077e-06,
      "loss": 0.5582,
      "step": 12406
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0064760446548462,
      "learning_rate": 1.1531952023565295e-06,
      "loss": 0.4861,
      "step": 12407
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0369073152542114,
      "learning_rate": 1.1525398639100194e-06,
      "loss": 0.5001,
      "step": 12408
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.210459589958191,
      "learning_rate": 1.1518846874666723e-06,
      "loss": 0.5119,
      "step": 12409
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2517964839935303,
      "learning_rate": 1.1512296730540717e-06,
      "loss": 0.51,
      "step": 12410
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0143394470214844,
      "learning_rate": 1.1505748206998036e-06,
      "loss": 0.4776,
      "step": 12411
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9268616437911987,
      "learning_rate": 1.1499201304314372e-06,
      "loss": 0.5016,
      "step": 12412
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0749627351760864,
      "learning_rate": 1.149265602276541e-06,
      "loss": 0.5425,
      "step": 12413
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0563760995864868,
      "learning_rate": 1.1486112362626738e-06,
      "loss": 0.5445,
      "step": 12414
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0509696006774902,
      "learning_rate": 1.14795703241739e-06,
      "loss": 0.509,
      "step": 12415
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9929944276809692,
      "learning_rate": 1.1473029907682348e-06,
      "loss": 0.427,
      "step": 12416
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.095852255821228,
      "learning_rate": 1.1466491113427503e-06,
      "loss": 0.4672,
      "step": 12417
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0767621994018555,
      "learning_rate": 1.1459953941684648e-06,
      "loss": 0.5021,
      "step": 12418
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0170347690582275,
      "learning_rate": 1.1453418392729065e-06,
      "loss": 0.4952,
      "step": 12419
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0994138717651367,
      "learning_rate": 1.1446884466835933e-06,
      "loss": 0.5337,
      "step": 12420
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1458686590194702,
      "learning_rate": 1.1440352164280388e-06,
      "loss": 0.4907,
      "step": 12421
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0651253461837769,
      "learning_rate": 1.1433821485337487e-06,
      "loss": 0.4825,
      "step": 12422
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.042712688446045,
      "learning_rate": 1.1427292430282165e-06,
      "loss": 0.4977,
      "step": 12423
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.02010178565979,
      "learning_rate": 1.14207649993894e-06,
      "loss": 0.5145,
      "step": 12424
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.131560206413269,
      "learning_rate": 1.1414239192934019e-06,
      "loss": 0.4912,
      "step": 12425
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0900567770004272,
      "learning_rate": 1.1407715011190784e-06,
      "loss": 0.4849,
      "step": 12426
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.135135531425476,
      "learning_rate": 1.1401192454434418e-06,
      "loss": 0.5114,
      "step": 12427
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0785999298095703,
      "learning_rate": 1.139467152293956e-06,
      "loss": 0.5069,
      "step": 12428
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1728732585906982,
      "learning_rate": 1.138815221698079e-06,
      "loss": 0.5358,
      "step": 12429
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.074499249458313,
      "learning_rate": 1.138163453683262e-06,
      "loss": 0.475,
      "step": 12430
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9937154650688171,
      "learning_rate": 1.1375118482769447e-06,
      "loss": 0.4616,
      "step": 12431
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9972637295722961,
      "learning_rate": 1.136860405506569e-06,
      "loss": 0.4687,
      "step": 12432
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9380080699920654,
      "learning_rate": 1.1362091253995632e-06,
      "loss": 0.4475,
      "step": 12433
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0433030128479004,
      "learning_rate": 1.1355580079833496e-06,
      "loss": 0.5259,
      "step": 12434
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0440255403518677,
      "learning_rate": 1.134907053285344e-06,
      "loss": 0.4836,
      "step": 12435
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0331069231033325,
      "learning_rate": 1.1342562613329571e-06,
      "loss": 0.5182,
      "step": 12436
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0718039274215698,
      "learning_rate": 1.133605632153591e-06,
      "loss": 0.5,
      "step": 12437
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1010985374450684,
      "learning_rate": 1.1329551657746429e-06,
      "loss": 0.5704,
      "step": 12438
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.12299382686615,
      "learning_rate": 1.132304862223499e-06,
      "loss": 0.5177,
      "step": 12439
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9886751174926758,
      "learning_rate": 1.1316547215275409e-06,
      "loss": 0.5398,
      "step": 12440
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0287259817123413,
      "learning_rate": 1.1310047437141485e-06,
      "loss": 0.5166,
      "step": 12441
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0754035711288452,
      "learning_rate": 1.1303549288106857e-06,
      "loss": 0.4811,
      "step": 12442
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0982261896133423,
      "learning_rate": 1.1297052768445154e-06,
      "loss": 0.5062,
      "step": 12443
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1342655420303345,
      "learning_rate": 1.129055787842992e-06,
      "loss": 0.4888,
      "step": 12444
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.09312105178833,
      "learning_rate": 1.1284064618334634e-06,
      "loss": 0.4614,
      "step": 12445
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1874736547470093,
      "learning_rate": 1.1277572988432716e-06,
      "loss": 0.5141,
      "step": 12446
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9963446855545044,
      "learning_rate": 1.1271082988997485e-06,
      "loss": 0.5013,
      "step": 12447
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1956878900527954,
      "learning_rate": 1.1264594620302216e-06,
      "loss": 0.5259,
      "step": 12448
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9786106944084167,
      "learning_rate": 1.1258107882620117e-06,
      "loss": 0.4864,
      "step": 12449
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1049765348434448,
      "learning_rate": 1.1251622776224325e-06,
      "loss": 0.5224,
      "step": 12450
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.047175407409668,
      "learning_rate": 1.1245139301387903e-06,
      "loss": 0.4617,
      "step": 12451
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.107480764389038,
      "learning_rate": 1.1238657458383857e-06,
      "loss": 0.5228,
      "step": 12452
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.088758111000061,
      "learning_rate": 1.1232177247485076e-06,
      "loss": 0.5175,
      "step": 12453
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0597591400146484,
      "learning_rate": 1.122569866896448e-06,
      "loss": 0.4779,
      "step": 12454
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.032039999961853,
      "learning_rate": 1.1219221723094815e-06,
      "loss": 0.4954,
      "step": 12455
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1140142679214478,
      "learning_rate": 1.1212746410148807e-06,
      "loss": 0.5409,
      "step": 12456
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0760780572891235,
      "learning_rate": 1.120627273039912e-06,
      "loss": 0.5255,
      "step": 12457
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0337908267974854,
      "learning_rate": 1.119980068411834e-06,
      "loss": 0.5136,
      "step": 12458
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0888824462890625,
      "learning_rate": 1.1193330271578968e-06,
      "loss": 0.5351,
      "step": 12459
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0786190032958984,
      "learning_rate": 1.118686149305348e-06,
      "loss": 0.4628,
      "step": 12460
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0877033472061157,
      "learning_rate": 1.1180394348814206e-06,
      "loss": 0.4994,
      "step": 12461
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0420441627502441,
      "learning_rate": 1.117392883913349e-06,
      "loss": 0.4942,
      "step": 12462
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0242724418640137,
      "learning_rate": 1.1167464964283587e-06,
      "loss": 0.4835,
      "step": 12463
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9864571690559387,
      "learning_rate": 1.1161002724536623e-06,
      "loss": 0.4727,
      "step": 12464
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0759340524673462,
      "learning_rate": 1.115454212016473e-06,
      "loss": 0.4859,
      "step": 12465
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0306402444839478,
      "learning_rate": 1.1148083151439932e-06,
      "loss": 0.488,
      "step": 12466
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0664716958999634,
      "learning_rate": 1.1141625818634194e-06,
      "loss": 0.511,
      "step": 12467
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2387864589691162,
      "learning_rate": 1.1135170122019433e-06,
      "loss": 0.5537,
      "step": 12468
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0770725011825562,
      "learning_rate": 1.112871606186744e-06,
      "loss": 0.5181,
      "step": 12469
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0454281568527222,
      "learning_rate": 1.112226363844998e-06,
      "loss": 0.5427,
      "step": 12470
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9640543460845947,
      "learning_rate": 1.1115812852038777e-06,
      "loss": 0.4233,
      "step": 12471
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0597995519638062,
      "learning_rate": 1.1109363702905419e-06,
      "loss": 0.5358,
      "step": 12472
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1093391180038452,
      "learning_rate": 1.1102916191321456e-06,
      "loss": 0.5271,
      "step": 12473
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0604957342147827,
      "learning_rate": 1.1096470317558384e-06,
      "loss": 0.4954,
      "step": 12474
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0820348262786865,
      "learning_rate": 1.1090026081887611e-06,
      "loss": 0.504,
      "step": 12475
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0919123888015747,
      "learning_rate": 1.1083583484580495e-06,
      "loss": 0.4784,
      "step": 12476
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9372173547744751,
      "learning_rate": 1.107714252590828e-06,
      "loss": 0.4721,
      "step": 12477
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9823314547538757,
      "learning_rate": 1.1070703206142186e-06,
      "loss": 0.5098,
      "step": 12478
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1057475805282593,
      "learning_rate": 1.1064265525553375e-06,
      "loss": 0.5691,
      "step": 12479
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.030718207359314,
      "learning_rate": 1.1057829484412885e-06,
      "loss": 0.4894,
      "step": 12480
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0543081760406494,
      "learning_rate": 1.1051395082991722e-06,
      "loss": 0.5059,
      "step": 12481
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0306936502456665,
      "learning_rate": 1.1044962321560837e-06,
      "loss": 0.5019,
      "step": 12482
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9948222637176514,
      "learning_rate": 1.1038531200391045e-06,
      "loss": 0.5566,
      "step": 12483
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0176924467086792,
      "learning_rate": 1.1032101719753197e-06,
      "loss": 0.4679,
      "step": 12484
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9939400553703308,
      "learning_rate": 1.102567387991797e-06,
      "loss": 0.4832,
      "step": 12485
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0428471565246582,
      "learning_rate": 1.101924768115603e-06,
      "loss": 0.486,
      "step": 12486
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0242252349853516,
      "learning_rate": 1.101282312373797e-06,
      "loss": 0.5002,
      "step": 12487
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1221117973327637,
      "learning_rate": 1.1006400207934304e-06,
      "loss": 0.5324,
      "step": 12488
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0698509216308594,
      "learning_rate": 1.0999978934015475e-06,
      "loss": 0.5184,
      "step": 12489
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.080450177192688,
      "learning_rate": 1.0993559302251878e-06,
      "loss": 0.5573,
      "step": 12490
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.034711480140686,
      "learning_rate": 1.0987141312913773e-06,
      "loss": 0.5067,
      "step": 12491
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1353464126586914,
      "learning_rate": 1.098072496627146e-06,
      "loss": 0.5728,
      "step": 12492
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.089824914932251,
      "learning_rate": 1.0974310262595067e-06,
      "loss": 0.5395,
      "step": 12493
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1430447101593018,
      "learning_rate": 1.096789720215471e-06,
      "loss": 0.6045,
      "step": 12494
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.062886357307434,
      "learning_rate": 1.0961485785220434e-06,
      "loss": 0.5036,
      "step": 12495
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.996558666229248,
      "learning_rate": 1.0955076012062155e-06,
      "loss": 0.5283,
      "step": 12496
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.056496024131775,
      "learning_rate": 1.094866788294981e-06,
      "loss": 0.4628,
      "step": 12497
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0023682117462158,
      "learning_rate": 1.094226139815323e-06,
      "loss": 0.473,
      "step": 12498
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1194649934768677,
      "learning_rate": 1.0935856557942132e-06,
      "loss": 0.4728,
      "step": 12499
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0746665000915527,
      "learning_rate": 1.0929453362586223e-06,
      "loss": 0.4718,
      "step": 12500
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.132388949394226,
      "learning_rate": 1.0923051812355117e-06,
      "loss": 0.5671,
      "step": 12501
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1181532144546509,
      "learning_rate": 1.091665190751836e-06,
      "loss": 0.5218,
      "step": 12502
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.140037178993225,
      "learning_rate": 1.0910253648345442e-06,
      "loss": 0.5484,
      "step": 12503
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.057293176651001,
      "learning_rate": 1.0903857035105736e-06,
      "loss": 0.5199,
      "step": 12504
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9680564403533936,
      "learning_rate": 1.0897462068068616e-06,
      "loss": 0.4586,
      "step": 12505
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.089263916015625,
      "learning_rate": 1.0891068747503353e-06,
      "loss": 0.5194,
      "step": 12506
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0306979417800903,
      "learning_rate": 1.0884677073679123e-06,
      "loss": 0.4676,
      "step": 12507
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9302442669868469,
      "learning_rate": 1.0878287046865072e-06,
      "loss": 0.4554,
      "step": 12508
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.081356167793274,
      "learning_rate": 1.0871898667330249e-06,
      "loss": 0.4919,
      "step": 12509
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.027951717376709,
      "learning_rate": 1.0865511935343664e-06,
      "loss": 0.4478,
      "step": 12510
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.006633996963501,
      "learning_rate": 1.0859126851174246e-06,
      "loss": 0.4938,
      "step": 12511
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0576258897781372,
      "learning_rate": 1.0852743415090823e-06,
      "loss": 0.5065,
      "step": 12512
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1270864009857178,
      "learning_rate": 1.0846361627362174e-06,
      "loss": 0.473,
      "step": 12513
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.988420844078064,
      "learning_rate": 1.0839981488257061e-06,
      "loss": 0.477,
      "step": 12514
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.015462875366211,
      "learning_rate": 1.0833602998044085e-06,
      "loss": 0.47,
      "step": 12515
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.111742615699768,
      "learning_rate": 1.0827226156991838e-06,
      "loss": 0.524,
      "step": 12516
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0591188669204712,
      "learning_rate": 1.0820850965368822e-06,
      "loss": 0.4825,
      "step": 12517
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.140297293663025,
      "learning_rate": 1.0814477423443482e-06,
      "loss": 0.5592,
      "step": 12518
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.121349573135376,
      "learning_rate": 1.0808105531484192e-06,
      "loss": 0.5092,
      "step": 12519
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.060700535774231,
      "learning_rate": 1.0801735289759225e-06,
      "loss": 0.4999,
      "step": 12520
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1008962392807007,
      "learning_rate": 1.0795366698536812e-06,
      "loss": 0.5153,
      "step": 12521
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0303106307983398,
      "learning_rate": 1.078899975808515e-06,
      "loss": 0.5057,
      "step": 12522
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.054789662361145,
      "learning_rate": 1.0782634468672293e-06,
      "loss": 0.4603,
      "step": 12523
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9646784067153931,
      "learning_rate": 1.0776270830566266e-06,
      "loss": 0.485,
      "step": 12524
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0548442602157593,
      "learning_rate": 1.0769908844035032e-06,
      "loss": 0.4672,
      "step": 12525
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0208356380462646,
      "learning_rate": 1.0763548509346461e-06,
      "loss": 0.4431,
      "step": 12526
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0277953147888184,
      "learning_rate": 1.0757189826768367e-06,
      "loss": 0.4811,
      "step": 12527
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0759657621383667,
      "learning_rate": 1.075083279656851e-06,
      "loss": 0.5032,
      "step": 12528
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0658326148986816,
      "learning_rate": 1.0744477419014532e-06,
      "loss": 0.5109,
      "step": 12529
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0466490983963013,
      "learning_rate": 1.0738123694374047e-06,
      "loss": 0.4807,
      "step": 12530
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9918878078460693,
      "learning_rate": 1.0731771622914595e-06,
      "loss": 0.4838,
      "step": 12531
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0270483493804932,
      "learning_rate": 1.072542120490363e-06,
      "loss": 0.5023,
      "step": 12532
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0566339492797852,
      "learning_rate": 1.0719072440608575e-06,
      "loss": 0.5024,
      "step": 12533
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9983569979667664,
      "learning_rate": 1.0712725330296697e-06,
      "loss": 0.4908,
      "step": 12534
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0100479125976562,
      "learning_rate": 1.07063798742353e-06,
      "loss": 0.5212,
      "step": 12535
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1355971097946167,
      "learning_rate": 1.0700036072691566e-06,
      "loss": 0.5449,
      "step": 12536
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1706911325454712,
      "learning_rate": 1.0693693925932585e-06,
      "loss": 0.5524,
      "step": 12537
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0609681606292725,
      "learning_rate": 1.0687353434225418e-06,
      "loss": 0.5237,
      "step": 12538
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0677683353424072,
      "learning_rate": 1.0681014597837042e-06,
      "loss": 0.5273,
      "step": 12539
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0673065185546875,
      "learning_rate": 1.0674677417034358e-06,
      "loss": 0.4832,
      "step": 12540
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9804611802101135,
      "learning_rate": 1.0668341892084217e-06,
      "loss": 0.4591,
      "step": 12541
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1132515668869019,
      "learning_rate": 1.0662008023253356e-06,
      "loss": 0.4985,
      "step": 12542
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.080709457397461,
      "learning_rate": 1.0655675810808485e-06,
      "loss": 0.5226,
      "step": 12543
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0782188177108765,
      "learning_rate": 1.0649345255016258e-06,
      "loss": 0.5295,
      "step": 12544
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0711718797683716,
      "learning_rate": 1.0643016356143204e-06,
      "loss": 0.5439,
      "step": 12545
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0208290815353394,
      "learning_rate": 1.0636689114455811e-06,
      "loss": 0.5282,
      "step": 12546
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.088685154914856,
      "learning_rate": 1.063036353022051e-06,
      "loss": 0.5668,
      "step": 12547
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.036378264427185,
      "learning_rate": 1.0624039603703645e-06,
      "loss": 0.4408,
      "step": 12548
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0257773399353027,
      "learning_rate": 1.06177173351715e-06,
      "loss": 0.5045,
      "step": 12549
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0909396409988403,
      "learning_rate": 1.061139672489027e-06,
      "loss": 0.5179,
      "step": 12550
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0610034465789795,
      "learning_rate": 1.0605077773126083e-06,
      "loss": 0.5166,
      "step": 12551
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.014702558517456,
      "learning_rate": 1.059876048014506e-06,
      "loss": 0.5213,
      "step": 12552
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1166365146636963,
      "learning_rate": 1.0592444846213145e-06,
      "loss": 0.5204,
      "step": 12553
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9917412996292114,
      "learning_rate": 1.058613087159629e-06,
      "loss": 0.4676,
      "step": 12554
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0502556562423706,
      "learning_rate": 1.0579818556560357e-06,
      "loss": 0.5403,
      "step": 12555
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1649173498153687,
      "learning_rate": 1.0573507901371126e-06,
      "loss": 0.5067,
      "step": 12556
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.035313606262207,
      "learning_rate": 1.0567198906294341e-06,
      "loss": 0.4929,
      "step": 12557
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0535595417022705,
      "learning_rate": 1.0560891571595616e-06,
      "loss": 0.5031,
      "step": 12558
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.067043662071228,
      "learning_rate": 1.0554585897540553e-06,
      "loss": 0.4907,
      "step": 12559
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.138853907585144,
      "learning_rate": 1.0548281884394657e-06,
      "loss": 0.4944,
      "step": 12560
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0570776462554932,
      "learning_rate": 1.0541979532423362e-06,
      "loss": 0.4762,
      "step": 12561
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9952700138092041,
      "learning_rate": 1.053567884189205e-06,
      "loss": 0.4968,
      "step": 12562
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.04869544506073,
      "learning_rate": 1.0529379813066026e-06,
      "loss": 0.4726,
      "step": 12563
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0134539604187012,
      "learning_rate": 1.0523082446210487e-06,
      "loss": 0.4516,
      "step": 12564
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0992261171340942,
      "learning_rate": 1.051678674159064e-06,
      "loss": 0.5417,
      "step": 12565
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0802074670791626,
      "learning_rate": 1.0510492699471536e-06,
      "loss": 0.5175,
      "step": 12566
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.125693917274475,
      "learning_rate": 1.0504200320118214e-06,
      "loss": 0.5146,
      "step": 12567
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.204516887664795,
      "learning_rate": 1.049790960379562e-06,
      "loss": 0.5596,
      "step": 12568
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0837621688842773,
      "learning_rate": 1.0491620550768633e-06,
      "loss": 0.5103,
      "step": 12569
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0523866415023804,
      "learning_rate": 1.048533316130207e-06,
      "loss": 0.4998,
      "step": 12570
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1099904775619507,
      "learning_rate": 1.0479047435660671e-06,
      "loss": 0.4525,
      "step": 12571
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0851279497146606,
      "learning_rate": 1.047276337410908e-06,
      "loss": 0.5323,
      "step": 12572
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0258618593215942,
      "learning_rate": 1.0466480976911947e-06,
      "loss": 0.5285,
      "step": 12573
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9892868995666504,
      "learning_rate": 1.0460200244333758e-06,
      "loss": 0.4928,
      "step": 12574
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9799687266349792,
      "learning_rate": 1.0453921176638981e-06,
      "loss": 0.4839,
      "step": 12575
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9586300849914551,
      "learning_rate": 1.044764377409203e-06,
      "loss": 0.4929,
      "step": 12576
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1096810102462769,
      "learning_rate": 1.0441368036957184e-06,
      "loss": 0.5154,
      "step": 12577
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0250980854034424,
      "learning_rate": 1.0435093965498727e-06,
      "loss": 0.4814,
      "step": 12578
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0203561782836914,
      "learning_rate": 1.0428821559980839e-06,
      "loss": 0.4966,
      "step": 12579
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9630902409553528,
      "learning_rate": 1.0422550820667605e-06,
      "loss": 0.4975,
      "step": 12580
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0307133197784424,
      "learning_rate": 1.0416281747823076e-06,
      "loss": 0.4603,
      "step": 12581
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1498281955718994,
      "learning_rate": 1.0410014341711216e-06,
      "loss": 0.5362,
      "step": 12582
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0730465650558472,
      "learning_rate": 1.0403748602595937e-06,
      "loss": 0.5233,
      "step": 12583
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0637001991271973,
      "learning_rate": 1.0397484530741053e-06,
      "loss": 0.5503,
      "step": 12584
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1022603511810303,
      "learning_rate": 1.0391222126410327e-06,
      "loss": 0.5288,
      "step": 12585
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.012472152709961,
      "learning_rate": 1.0384961389867454e-06,
      "loss": 0.4724,
      "step": 12586
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0934635400772095,
      "learning_rate": 1.0378702321376054e-06,
      "loss": 0.5138,
      "step": 12587
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9922604560852051,
      "learning_rate": 1.037244492119966e-06,
      "loss": 0.5094,
      "step": 12588
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9959739446640015,
      "learning_rate": 1.036618918960175e-06,
      "loss": 0.5135,
      "step": 12589
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0212490558624268,
      "learning_rate": 1.0359935126845738e-06,
      "loss": 0.4835,
      "step": 12590
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9797313809394836,
      "learning_rate": 1.0353682733194965e-06,
      "loss": 0.4893,
      "step": 12591
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0796663761138916,
      "learning_rate": 1.0347432008912688e-06,
      "loss": 0.5523,
      "step": 12592
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.037741780281067,
      "learning_rate": 1.0341182954262125e-06,
      "loss": 0.4667,
      "step": 12593
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9964542984962463,
      "learning_rate": 1.0334935569506355e-06,
      "loss": 0.4916,
      "step": 12594
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.066288709640503,
      "learning_rate": 1.0328689854908492e-06,
      "loss": 0.5088,
      "step": 12595
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9630863666534424,
      "learning_rate": 1.032244581073148e-06,
      "loss": 0.4728,
      "step": 12596
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1112786531448364,
      "learning_rate": 1.0316203437238242e-06,
      "loss": 0.5292,
      "step": 12597
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0189820528030396,
      "learning_rate": 1.0309962734691632e-06,
      "loss": 0.5186,
      "step": 12598
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1797800064086914,
      "learning_rate": 1.0303723703354418e-06,
      "loss": 0.5682,
      "step": 12599
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0767300128936768,
      "learning_rate": 1.0297486343489304e-06,
      "loss": 0.5095,
      "step": 12600
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0390448570251465,
      "learning_rate": 1.0291250655358942e-06,
      "loss": 0.5208,
      "step": 12601
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9824992418289185,
      "learning_rate": 1.0285016639225849e-06,
      "loss": 0.5199,
      "step": 12602
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0326634645462036,
      "learning_rate": 1.0278784295352572e-06,
      "loss": 0.5177,
      "step": 12603
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0143102407455444,
      "learning_rate": 1.0272553624001502e-06,
      "loss": 0.4865,
      "step": 12604
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0329055786132812,
      "learning_rate": 1.0266324625434992e-06,
      "loss": 0.4971,
      "step": 12605
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9900029301643372,
      "learning_rate": 1.0260097299915345e-06,
      "loss": 0.4573,
      "step": 12606
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0251998901367188,
      "learning_rate": 1.0253871647704722e-06,
      "loss": 0.4938,
      "step": 12607
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.018629550933838,
      "learning_rate": 1.024764766906532e-06,
      "loss": 0.4997,
      "step": 12608
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0895016193389893,
      "learning_rate": 1.0241425364259195e-06,
      "loss": 0.5291,
      "step": 12609
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0606317520141602,
      "learning_rate": 1.0235204733548321e-06,
      "loss": 0.4863,
      "step": 12610
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0265743732452393,
      "learning_rate": 1.022898577719465e-06,
      "loss": 0.5231,
      "step": 12611
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0301939249038696,
      "learning_rate": 1.0222768495460029e-06,
      "loss": 0.4679,
      "step": 12612
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0592392683029175,
      "learning_rate": 1.0216552888606256e-06,
      "loss": 0.4683,
      "step": 12613
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0133415460586548,
      "learning_rate": 1.0210338956895054e-06,
      "loss": 0.5066,
      "step": 12614
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0261272192001343,
      "learning_rate": 1.020412670058804e-06,
      "loss": 0.5144,
      "step": 12615
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1378453969955444,
      "learning_rate": 1.0197916119946821e-06,
      "loss": 0.5321,
      "step": 12616
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1250720024108887,
      "learning_rate": 1.0191707215232905e-06,
      "loss": 0.5413,
      "step": 12617
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0722566843032837,
      "learning_rate": 1.0185499986707702e-06,
      "loss": 0.5324,
      "step": 12618
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0596920251846313,
      "learning_rate": 1.0179294434632593e-06,
      "loss": 0.4888,
      "step": 12619
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0051017999649048,
      "learning_rate": 1.0173090559268867e-06,
      "loss": 0.4899,
      "step": 12620
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1246610879898071,
      "learning_rate": 1.0166888360877747e-06,
      "loss": 0.5003,
      "step": 12621
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0212544202804565,
      "learning_rate": 1.0160687839720407e-06,
      "loss": 0.5072,
      "step": 12622
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0428470373153687,
      "learning_rate": 1.0154488996057894e-06,
      "loss": 0.5297,
      "step": 12623
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.092565655708313,
      "learning_rate": 1.0148291830151224e-06,
      "loss": 0.5606,
      "step": 12624
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1170252561569214,
      "learning_rate": 1.014209634226138e-06,
      "loss": 0.5372,
      "step": 12625
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0844227075576782,
      "learning_rate": 1.013590253264919e-06,
      "loss": 0.5427,
      "step": 12626
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0359807014465332,
      "learning_rate": 1.0129710401575465e-06,
      "loss": 0.5043,
      "step": 12627
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.022281527519226,
      "learning_rate": 1.0123519949300942e-06,
      "loss": 0.5025,
      "step": 12628
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0547220706939697,
      "learning_rate": 1.0117331176086264e-06,
      "loss": 0.4771,
      "step": 12629
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0500353574752808,
      "learning_rate": 1.0111144082192048e-06,
      "loss": 0.5252,
      "step": 12630
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1121727228164673,
      "learning_rate": 1.0104958667878778e-06,
      "loss": 0.5176,
      "step": 12631
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0687726736068726,
      "learning_rate": 1.0098774933406903e-06,
      "loss": 0.513,
      "step": 12632
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0900862216949463,
      "learning_rate": 1.0092592879036834e-06,
      "loss": 0.5047,
      "step": 12633
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0976719856262207,
      "learning_rate": 1.0086412505028836e-06,
      "loss": 0.4683,
      "step": 12634
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1430429220199585,
      "learning_rate": 1.0080233811643158e-06,
      "loss": 0.5266,
      "step": 12635
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9831097722053528,
      "learning_rate": 1.0074056799139981e-06,
      "loss": 0.5366,
      "step": 12636
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0840317010879517,
      "learning_rate": 1.006788146777935e-06,
      "loss": 0.5049,
      "step": 12637
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0385769605636597,
      "learning_rate": 1.0061707817821343e-06,
      "loss": 0.5058,
      "step": 12638
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0986313819885254,
      "learning_rate": 1.0055535849525872e-06,
      "loss": 0.4705,
      "step": 12639
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1245629787445068,
      "learning_rate": 1.004936556315283e-06,
      "loss": 0.5381,
      "step": 12640
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0714011192321777,
      "learning_rate": 1.004319695896202e-06,
      "loss": 0.5182,
      "step": 12641
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0748082399368286,
      "learning_rate": 1.0037030037213197e-06,
      "loss": 0.4803,
      "step": 12642
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1025422811508179,
      "learning_rate": 1.0030864798166013e-06,
      "loss": 0.5621,
      "step": 12643
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0863114595413208,
      "learning_rate": 1.0024701242080082e-06,
      "loss": 0.5265,
      "step": 12644
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1285635232925415,
      "learning_rate": 1.0018539369214891e-06,
      "loss": 0.5193,
      "step": 12645
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9931705594062805,
      "learning_rate": 1.0012379179829951e-06,
      "loss": 0.5036,
      "step": 12646
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1284775733947754,
      "learning_rate": 1.0006220674184602e-06,
      "loss": 0.5376,
      "step": 12647
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1133114099502563,
      "learning_rate": 1.0000063852538172e-06,
      "loss": 0.4708,
      "step": 12648
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1451929807662964,
      "learning_rate": 9.993908715149902e-07,
      "loss": 0.5042,
      "step": 12649
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0009359121322632,
      "learning_rate": 9.98775526227897e-07,
      "loss": 0.451,
      "step": 12650
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0615708827972412,
      "learning_rate": 9.981603494184473e-07,
      "loss": 0.5368,
      "step": 12651
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0838319063186646,
      "learning_rate": 9.975453411125447e-07,
      "loss": 0.5032,
      "step": 12652
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1050270795822144,
      "learning_rate": 9.969305013360825e-07,
      "loss": 0.4872,
      "step": 12653
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1218594312667847,
      "learning_rate": 9.963158301149522e-07,
      "loss": 0.532,
      "step": 12654
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0388950109481812,
      "learning_rate": 9.957013274750338e-07,
      "loss": 0.526,
      "step": 12655
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.005384922027588,
      "learning_rate": 9.95086993442203e-07,
      "loss": 0.4991,
      "step": 12656
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1161460876464844,
      "learning_rate": 9.944728280423265e-07,
      "loss": 0.5416,
      "step": 12657
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9668041467666626,
      "learning_rate": 9.938588313012655e-07,
      "loss": 0.4708,
      "step": 12658
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1281183958053589,
      "learning_rate": 9.93245003244872e-07,
      "loss": 0.5069,
      "step": 12659
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.054303526878357,
      "learning_rate": 9.92631343898995e-07,
      "loss": 0.5112,
      "step": 12660
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1362591981887817,
      "learning_rate": 9.920178532894698e-07,
      "loss": 0.4875,
      "step": 12661
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.027287483215332,
      "learning_rate": 9.9140453144213e-07,
      "loss": 0.4441,
      "step": 12662
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.027077078819275,
      "learning_rate": 9.907913783828004e-07,
      "loss": 0.4899,
      "step": 12663
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0585702657699585,
      "learning_rate": 9.901783941372988e-07,
      "loss": 0.4836,
      "step": 12664
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0785973072052002,
      "learning_rate": 9.895655787314361e-07,
      "loss": 0.557,
      "step": 12665
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0298945903778076,
      "learning_rate": 9.889529321910169e-07,
      "loss": 0.4785,
      "step": 12666
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.070832371711731,
      "learning_rate": 9.88340454541834e-07,
      "loss": 0.4906,
      "step": 12667
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1544461250305176,
      "learning_rate": 9.87728145809681e-07,
      "loss": 0.5258,
      "step": 12668
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0338704586029053,
      "learning_rate": 9.871160060203371e-07,
      "loss": 0.4249,
      "step": 12669
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9609220027923584,
      "learning_rate": 9.865040351995787e-07,
      "loss": 0.4871,
      "step": 12670
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0077117681503296,
      "learning_rate": 9.85892233373173e-07,
      "loss": 0.4835,
      "step": 12671
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9985730648040771,
      "learning_rate": 9.852806005668813e-07,
      "loss": 0.4722,
      "step": 12672
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1201215982437134,
      "learning_rate": 9.846691368064577e-07,
      "loss": 0.4947,
      "step": 12673
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.046615719795227,
      "learning_rate": 9.840578421176495e-07,
      "loss": 0.4741,
      "step": 12674
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0141711235046387,
      "learning_rate": 9.834467165261924e-07,
      "loss": 0.4817,
      "step": 12675
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1131911277770996,
      "learning_rate": 9.828357600578242e-07,
      "loss": 0.4825,
      "step": 12676
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1036888360977173,
      "learning_rate": 9.82224972738266e-07,
      "loss": 0.5249,
      "step": 12677
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0062801837921143,
      "learning_rate": 9.816143545932378e-07,
      "loss": 0.4517,
      "step": 12678
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1002241373062134,
      "learning_rate": 9.8100390564845e-07,
      "loss": 0.4809,
      "step": 12679
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9574532508850098,
      "learning_rate": 9.803936259296066e-07,
      "loss": 0.4358,
      "step": 12680
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0709331035614014,
      "learning_rate": 9.797835154624041e-07,
      "loss": 0.5312,
      "step": 12681
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.082556128501892,
      "learning_rate": 9.791735742725339e-07,
      "loss": 0.5543,
      "step": 12682
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1491870880126953,
      "learning_rate": 9.78563802385676e-07,
      "loss": 0.4738,
      "step": 12683
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0681147575378418,
      "learning_rate": 9.779541998275067e-07,
      "loss": 0.496,
      "step": 12684
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.092307209968567,
      "learning_rate": 9.773447666236946e-07,
      "loss": 0.471,
      "step": 12685
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9956874251365662,
      "learning_rate": 9.767355027999004e-07,
      "loss": 0.492,
      "step": 12686
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9814101457595825,
      "learning_rate": 9.761264083817795e-07,
      "loss": 0.4715,
      "step": 12687
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.137710690498352,
      "learning_rate": 9.755174833949749e-07,
      "loss": 0.4593,
      "step": 12688
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1179423332214355,
      "learning_rate": 9.749087278651304e-07,
      "loss": 0.5595,
      "step": 12689
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.055477261543274,
      "learning_rate": 9.743001418178782e-07,
      "loss": 0.5269,
      "step": 12690
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.06284499168396,
      "learning_rate": 9.736917252788414e-07,
      "loss": 0.4771,
      "step": 12691
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.029338002204895,
      "learning_rate": 9.730834782736393e-07,
      "loss": 0.4957,
      "step": 12692
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0246593952178955,
      "learning_rate": 9.724754008278836e-07,
      "loss": 0.491,
      "step": 12693
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9944297671318054,
      "learning_rate": 9.718674929671778e-07,
      "loss": 0.4525,
      "step": 12694
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0613635778427124,
      "learning_rate": 9.71259754717121e-07,
      "loss": 0.523,
      "step": 12695
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.114700198173523,
      "learning_rate": 9.706521861032974e-07,
      "loss": 0.4554,
      "step": 12696
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.114327073097229,
      "learning_rate": 9.700447871512953e-07,
      "loss": 0.5158,
      "step": 12697
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0712555646896362,
      "learning_rate": 9.694375578866889e-07,
      "loss": 0.5325,
      "step": 12698
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0607600212097168,
      "learning_rate": 9.688304983350443e-07,
      "loss": 0.5167,
      "step": 12699
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.012575626373291,
      "learning_rate": 9.682236085219243e-07,
      "loss": 0.4688,
      "step": 12700
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0543360710144043,
      "learning_rate": 9.67616888472882e-07,
      "loss": 0.4889,
      "step": 12701
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0162805318832397,
      "learning_rate": 9.670103382134655e-07,
      "loss": 0.4771,
      "step": 12702
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0929137468338013,
      "learning_rate": 9.664039577692152e-07,
      "loss": 0.5128,
      "step": 12703
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0132461786270142,
      "learning_rate": 9.65797747165661e-07,
      "loss": 0.505,
      "step": 12704
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.038873314857483,
      "learning_rate": 9.65191706428328e-07,
      "loss": 0.5092,
      "step": 12705
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0783709287643433,
      "learning_rate": 9.645858355827392e-07,
      "loss": 0.5183,
      "step": 12706
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9915413856506348,
      "learning_rate": 9.639801346544015e-07,
      "loss": 0.5139,
      "step": 12707
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0379064083099365,
      "learning_rate": 9.633746036688196e-07,
      "loss": 0.4709,
      "step": 12708
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9953266978263855,
      "learning_rate": 9.627692426514907e-07,
      "loss": 0.4872,
      "step": 12709
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0961235761642456,
      "learning_rate": 9.621640516279047e-07,
      "loss": 0.51,
      "step": 12710
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0647658109664917,
      "learning_rate": 9.61559030623545e-07,
      "loss": 0.5198,
      "step": 12711
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0856093168258667,
      "learning_rate": 9.609541796638848e-07,
      "loss": 0.4598,
      "step": 12712
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0690525770187378,
      "learning_rate": 9.603494987743932e-07,
      "loss": 0.5155,
      "step": 12713
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1249573230743408,
      "learning_rate": 9.597449879805314e-07,
      "loss": 0.4989,
      "step": 12714
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0353507995605469,
      "learning_rate": 9.59140647307753e-07,
      "loss": 0.4712,
      "step": 12715
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1793574094772339,
      "learning_rate": 9.585364767815048e-07,
      "loss": 0.485,
      "step": 12716
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0845911502838135,
      "learning_rate": 9.57932476427228e-07,
      "loss": 0.5272,
      "step": 12717
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0367937088012695,
      "learning_rate": 9.573286462703501e-07,
      "loss": 0.4813,
      "step": 12718
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.994184136390686,
      "learning_rate": 9.567249863363027e-07,
      "loss": 0.479,
      "step": 12719
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0609592199325562,
      "learning_rate": 9.56121496650499e-07,
      "loss": 0.5304,
      "step": 12720
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9854810833930969,
      "learning_rate": 9.55518177238351e-07,
      "loss": 0.5232,
      "step": 12721
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.023314356803894,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.5233,
      "step": 12722
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0726054906845093,
      "learning_rate": 9.54312049336632e-07,
      "loss": 0.5171,
      "step": 12723
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9805082678794861,
      "learning_rate": 9.53709240897846e-07,
      "loss": 0.4573,
      "step": 12724
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0192350149154663,
      "learning_rate": 9.531066028342895e-07,
      "loss": 0.5394,
      "step": 12725
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0108089447021484,
      "learning_rate": 9.525041351713332e-07,
      "loss": 0.4892,
      "step": 12726
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0106862783432007,
      "learning_rate": 9.519018379343486e-07,
      "loss": 0.4612,
      "step": 12727
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0383695363998413,
      "learning_rate": 9.512997111486965e-07,
      "loss": 0.53,
      "step": 12728
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0875706672668457,
      "learning_rate": 9.506977548397284e-07,
      "loss": 0.5628,
      "step": 12729
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0377280712127686,
      "learning_rate": 9.50095969032791e-07,
      "loss": 0.5389,
      "step": 12730
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.090673804283142,
      "learning_rate": 9.494943537532242e-07,
      "loss": 0.494,
      "step": 12731
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9572697281837463,
      "learning_rate": 9.488929090263588e-07,
      "loss": 0.465,
      "step": 12732
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.033835530281067,
      "learning_rate": 9.482916348775217e-07,
      "loss": 0.5165,
      "step": 12733
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1086596250534058,
      "learning_rate": 9.476905313320283e-07,
      "loss": 0.5433,
      "step": 12734
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0501341819763184,
      "learning_rate": 9.470895984151879e-07,
      "loss": 0.4828,
      "step": 12735
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0581494569778442,
      "learning_rate": 9.464888361523078e-07,
      "loss": 0.4805,
      "step": 12736
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1357430219650269,
      "learning_rate": 9.458882445686807e-07,
      "loss": 0.5229,
      "step": 12737
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1199616193771362,
      "learning_rate": 9.452878236895963e-07,
      "loss": 0.5418,
      "step": 12738
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0396251678466797,
      "learning_rate": 9.446875735403366e-07,
      "loss": 0.495,
      "step": 12739
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0419509410858154,
      "learning_rate": 9.440874941461753e-07,
      "loss": 0.4972,
      "step": 12740
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0135390758514404,
      "learning_rate": 9.434875855323816e-07,
      "loss": 0.4716,
      "step": 12741
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.018988013267517,
      "learning_rate": 9.428878477242131e-07,
      "loss": 0.5212,
      "step": 12742
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0946797132492065,
      "learning_rate": 9.422882807469219e-07,
      "loss": 0.466,
      "step": 12743
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0877785682678223,
      "learning_rate": 9.416888846257588e-07,
      "loss": 0.5006,
      "step": 12744
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0570225715637207,
      "learning_rate": 9.41089659385957e-07,
      "loss": 0.5155,
      "step": 12745
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.065837025642395,
      "learning_rate": 9.404906050527496e-07,
      "loss": 0.5261,
      "step": 12746
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.050106406211853,
      "learning_rate": 9.398917216513625e-07,
      "loss": 0.4856,
      "step": 12747
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1717270612716675,
      "learning_rate": 9.39293009207008e-07,
      "loss": 0.5413,
      "step": 12748
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1501052379608154,
      "learning_rate": 9.386944677449017e-07,
      "loss": 0.5275,
      "step": 12749
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9979376196861267,
      "learning_rate": 9.380960972902414e-07,
      "loss": 0.4461,
      "step": 12750
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.067839503288269,
      "learning_rate": 9.374978978682248e-07,
      "loss": 0.4897,
      "step": 12751
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2218809127807617,
      "learning_rate": 9.368998695040387e-07,
      "loss": 0.5476,
      "step": 12752
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0331355333328247,
      "learning_rate": 9.363020122228645e-07,
      "loss": 0.4985,
      "step": 12753
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0841494798660278,
      "learning_rate": 9.357043260498766e-07,
      "loss": 0.5028,
      "step": 12754
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0547598600387573,
      "learning_rate": 9.351068110102418e-07,
      "loss": 0.4473,
      "step": 12755
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2031419277191162,
      "learning_rate": 9.345094671291155e-07,
      "loss": 0.5074,
      "step": 12756
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1683024168014526,
      "learning_rate": 9.339122944316559e-07,
      "loss": 0.541,
      "step": 12757
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0284042358398438,
      "learning_rate": 9.333152929430029e-07,
      "loss": 0.4993,
      "step": 12758
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1034044027328491,
      "learning_rate": 9.327184626882963e-07,
      "loss": 0.5493,
      "step": 12759
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0372745990753174,
      "learning_rate": 9.321218036926677e-07,
      "loss": 0.4963,
      "step": 12760
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0754164457321167,
      "learning_rate": 9.315253159812359e-07,
      "loss": 0.5033,
      "step": 12761
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.061840295791626,
      "learning_rate": 9.30928999579121e-07,
      "loss": 0.4783,
      "step": 12762
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.165432095527649,
      "learning_rate": 9.303328545114321e-07,
      "loss": 0.4691,
      "step": 12763
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0695961713790894,
      "learning_rate": 9.29736880803268e-07,
      "loss": 0.4595,
      "step": 12764
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1859469413757324,
      "learning_rate": 9.29141078479725e-07,
      "loss": 0.5298,
      "step": 12765
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1136317253112793,
      "learning_rate": 9.285454475658889e-07,
      "loss": 0.5012,
      "step": 12766
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.957591712474823,
      "learning_rate": 9.279499880868409e-07,
      "loss": 0.4346,
      "step": 12767
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0229932069778442,
      "learning_rate": 9.273547000676547e-07,
      "loss": 0.5261,
      "step": 12768
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.070711374282837,
      "learning_rate": 9.267595835333915e-07,
      "loss": 0.5153,
      "step": 12769
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.040297269821167,
      "learning_rate": 9.261646385091139e-07,
      "loss": 0.4738,
      "step": 12770
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0666635036468506,
      "learning_rate": 9.25569865019873e-07,
      "loss": 0.4977,
      "step": 12771
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.158909797668457,
      "learning_rate": 9.249752630907094e-07,
      "loss": 0.4999,
      "step": 12772
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0651565790176392,
      "learning_rate": 9.243808327466619e-07,
      "loss": 0.47,
      "step": 12773
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0467365980148315,
      "learning_rate": 9.237865740127594e-07,
      "loss": 0.4714,
      "step": 12774
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0020484924316406,
      "learning_rate": 9.231924869140241e-07,
      "loss": 0.4715,
      "step": 12775
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0114612579345703,
      "learning_rate": 9.225985714754721e-07,
      "loss": 0.4824,
      "step": 12776
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0749523639678955,
      "learning_rate": 9.220048277221089e-07,
      "loss": 0.5071,
      "step": 12777
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.038300633430481,
      "learning_rate": 9.214112556789345e-07,
      "loss": 0.5062,
      "step": 12778
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.062896490097046,
      "learning_rate": 9.208178553709468e-07,
      "loss": 0.5121,
      "step": 12779
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1365878582000732,
      "learning_rate": 9.202246268231274e-07,
      "loss": 0.5414,
      "step": 12780
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1047718524932861,
      "learning_rate": 9.196315700604564e-07,
      "loss": 0.5558,
      "step": 12781
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9735736846923828,
      "learning_rate": 9.190386851079053e-07,
      "loss": 0.4957,
      "step": 12782
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.096001148223877,
      "learning_rate": 9.184459719904388e-07,
      "loss": 0.4893,
      "step": 12783
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0629702806472778,
      "learning_rate": 9.178534307330145e-07,
      "loss": 0.4838,
      "step": 12784
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0845651626586914,
      "learning_rate": 9.17261061360581e-07,
      "loss": 0.5347,
      "step": 12785
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0777674913406372,
      "learning_rate": 9.166688638980791e-07,
      "loss": 0.4936,
      "step": 12786
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0557137727737427,
      "learning_rate": 9.160768383704499e-07,
      "loss": 0.4577,
      "step": 12787
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.97430819272995,
      "learning_rate": 9.154849848026165e-07,
      "loss": 0.4938,
      "step": 12788
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0108232498168945,
      "learning_rate": 9.148933032195013e-07,
      "loss": 0.497,
      "step": 12789
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.094015121459961,
      "learning_rate": 9.14301793646018e-07,
      "loss": 0.5157,
      "step": 12790
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0083976984024048,
      "learning_rate": 9.137104561070736e-07,
      "loss": 0.4662,
      "step": 12791
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0446066856384277,
      "learning_rate": 9.13119290627566e-07,
      "loss": 0.5072,
      "step": 12792
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.00369131565094,
      "learning_rate": 9.125282972323895e-07,
      "loss": 0.4938,
      "step": 12793
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1130354404449463,
      "learning_rate": 9.119374759464261e-07,
      "loss": 0.5223,
      "step": 12794
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.976584792137146,
      "learning_rate": 9.113468267945541e-07,
      "loss": 0.4466,
      "step": 12795
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0781188011169434,
      "learning_rate": 9.107563498016436e-07,
      "loss": 0.5376,
      "step": 12796
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1136902570724487,
      "learning_rate": 9.101660449925576e-07,
      "loss": 0.5402,
      "step": 12797
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1188857555389404,
      "learning_rate": 9.095759123921538e-07,
      "loss": 0.4836,
      "step": 12798
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0546386241912842,
      "learning_rate": 9.089859520252759e-07,
      "loss": 0.4632,
      "step": 12799
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1947174072265625,
      "learning_rate": 9.083961639167693e-07,
      "loss": 0.5401,
      "step": 12800
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2043405771255493,
      "learning_rate": 9.078065480914678e-07,
      "loss": 0.5621,
      "step": 12801
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1328403949737549,
      "learning_rate": 9.072171045741957e-07,
      "loss": 0.52,
      "step": 12802
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0790365934371948,
      "learning_rate": 9.066278333897732e-07,
      "loss": 0.5446,
      "step": 12803
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0958285331726074,
      "learning_rate": 9.060387345630134e-07,
      "loss": 0.487,
      "step": 12804
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1146931648254395,
      "learning_rate": 9.054498081187202e-07,
      "loss": 0.4661,
      "step": 12805
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0177806615829468,
      "learning_rate": 9.048610540816932e-07,
      "loss": 0.4937,
      "step": 12806
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9972603917121887,
      "learning_rate": 9.042724724767199e-07,
      "loss": 0.5211,
      "step": 12807
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0634011030197144,
      "learning_rate": 9.036840633285837e-07,
      "loss": 0.4974,
      "step": 12808
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0534765720367432,
      "learning_rate": 9.030958266620637e-07,
      "loss": 0.5286,
      "step": 12809
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0246983766555786,
      "learning_rate": 9.025077625019252e-07,
      "loss": 0.4781,
      "step": 12810
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0605356693267822,
      "learning_rate": 9.01919870872931e-07,
      "loss": 0.5262,
      "step": 12811
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1038426160812378,
      "learning_rate": 9.013321517998347e-07,
      "loss": 0.4912,
      "step": 12812
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1282594203948975,
      "learning_rate": 9.007446053073832e-07,
      "loss": 0.5649,
      "step": 12813
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9613350629806519,
      "learning_rate": 9.001572314203172e-07,
      "loss": 0.4879,
      "step": 12814
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1107274293899536,
      "learning_rate": 8.99570030163367e-07,
      "loss": 0.5202,
      "step": 12815
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0693176984786987,
      "learning_rate": 8.989830015612566e-07,
      "loss": 0.5076,
      "step": 12816
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0245882272720337,
      "learning_rate": 8.983961456387086e-07,
      "loss": 0.5388,
      "step": 12817
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0226383209228516,
      "learning_rate": 8.978094624204292e-07,
      "loss": 0.4687,
      "step": 12818
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.06915283203125,
      "learning_rate": 8.972229519311227e-07,
      "loss": 0.522,
      "step": 12819
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0821400880813599,
      "learning_rate": 8.966366141954852e-07,
      "loss": 0.5298,
      "step": 12820
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0791270732879639,
      "learning_rate": 8.960504492382055e-07,
      "loss": 0.5111,
      "step": 12821
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0358119010925293,
      "learning_rate": 8.95464457083966e-07,
      "loss": 0.49,
      "step": 12822
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0727996826171875,
      "learning_rate": 8.948786377574382e-07,
      "loss": 0.5114,
      "step": 12823
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0534948110580444,
      "learning_rate": 8.942929912832904e-07,
      "loss": 0.4841,
      "step": 12824
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9961467981338501,
      "learning_rate": 8.93707517686182e-07,
      "loss": 0.4767,
      "step": 12825
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9979651570320129,
      "learning_rate": 8.93122216990765e-07,
      "loss": 0.477,
      "step": 12826
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1751617193222046,
      "learning_rate": 8.92537089221685e-07,
      "loss": 0.5293,
      "step": 12827
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0309288501739502,
      "learning_rate": 8.919521344035808e-07,
      "loss": 0.4804,
      "step": 12828
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.023771047592163,
      "learning_rate": 8.913673525610783e-07,
      "loss": 0.5093,
      "step": 12829
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0918841361999512,
      "learning_rate": 8.907827437188065e-07,
      "loss": 0.5151,
      "step": 12830
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0214347839355469,
      "learning_rate": 8.901983079013771e-07,
      "loss": 0.4703,
      "step": 12831
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.103243350982666,
      "learning_rate": 8.896140451334001e-07,
      "loss": 0.5128,
      "step": 12832
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.016454815864563,
      "learning_rate": 8.890299554394766e-07,
      "loss": 0.4864,
      "step": 12833
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.051961898803711,
      "learning_rate": 8.884460388442006e-07,
      "loss": 0.4967,
      "step": 12834
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.088719129562378,
      "learning_rate": 8.878622953721589e-07,
      "loss": 0.4788,
      "step": 12835
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0930858850479126,
      "learning_rate": 8.87278725047932e-07,
      "loss": 0.5075,
      "step": 12836
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1117326021194458,
      "learning_rate": 8.866953278960888e-07,
      "loss": 0.5197,
      "step": 12837
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0902756452560425,
      "learning_rate": 8.86112103941198e-07,
      "loss": 0.4675,
      "step": 12838
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9956930875778198,
      "learning_rate": 8.855290532078148e-07,
      "loss": 0.4349,
      "step": 12839
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.092830777168274,
      "learning_rate": 8.849461757204897e-07,
      "loss": 0.5405,
      "step": 12840
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0364973545074463,
      "learning_rate": 8.843634715037669e-07,
      "loss": 0.5061,
      "step": 12841
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0219436883926392,
      "learning_rate": 8.83780940582179e-07,
      "loss": 0.541,
      "step": 12842
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9847308993339539,
      "learning_rate": 8.83198582980257e-07,
      "loss": 0.4969,
      "step": 12843
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1531392335891724,
      "learning_rate": 8.826163987225233e-07,
      "loss": 0.5078,
      "step": 12844
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0827713012695312,
      "learning_rate": 8.82034387833488e-07,
      "loss": 0.513,
      "step": 12845
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0626567602157593,
      "learning_rate": 8.814525503376597e-07,
      "loss": 0.4932,
      "step": 12846
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0791380405426025,
      "learning_rate": 8.808708862595367e-07,
      "loss": 0.4864,
      "step": 12847
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0617848634719849,
      "learning_rate": 8.802893956236114e-07,
      "loss": 0.4669,
      "step": 12848
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0819118022918701,
      "learning_rate": 8.797080784543699e-07,
      "loss": 0.4879,
      "step": 12849
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1863682270050049,
      "learning_rate": 8.791269347762849e-07,
      "loss": 0.5103,
      "step": 12850
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0164762735366821,
      "learning_rate": 8.785459646138306e-07,
      "loss": 0.4691,
      "step": 12851
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0647052526474,
      "learning_rate": 8.779651679914692e-07,
      "loss": 0.4773,
      "step": 12852
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0151244401931763,
      "learning_rate": 8.773845449336537e-07,
      "loss": 0.5087,
      "step": 12853
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0956175327301025,
      "learning_rate": 8.768040954648338e-07,
      "loss": 0.5361,
      "step": 12854
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0666496753692627,
      "learning_rate": 8.762238196094502e-07,
      "loss": 0.4673,
      "step": 12855
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1929495334625244,
      "learning_rate": 8.756437173919352e-07,
      "loss": 0.5144,
      "step": 12856
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9840711951255798,
      "learning_rate": 8.750637888367164e-07,
      "loss": 0.4645,
      "step": 12857
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0075687170028687,
      "learning_rate": 8.744840339682126e-07,
      "loss": 0.4712,
      "step": 12858
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9786378145217896,
      "learning_rate": 8.73904452810832e-07,
      "loss": 0.4876,
      "step": 12859
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1230751276016235,
      "learning_rate": 8.733250453889841e-07,
      "loss": 0.461,
      "step": 12860
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1180142164230347,
      "learning_rate": 8.727458117270615e-07,
      "loss": 0.4621,
      "step": 12861
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0221726894378662,
      "learning_rate": 8.721667518494553e-07,
      "loss": 0.5081,
      "step": 12862
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.059422492980957,
      "learning_rate": 8.715878657805471e-07,
      "loss": 0.5142,
      "step": 12863
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0881012678146362,
      "learning_rate": 8.710091535447123e-07,
      "loss": 0.509,
      "step": 12864
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0379509925842285,
      "learning_rate": 8.704306151663184e-07,
      "loss": 0.4673,
      "step": 12865
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0658977031707764,
      "learning_rate": 8.698522506697271e-07,
      "loss": 0.502,
      "step": 12866
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.005908489227295,
      "learning_rate": 8.692740600792871e-07,
      "loss": 0.5111,
      "step": 12867
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0216130018234253,
      "learning_rate": 8.686960434193486e-07,
      "loss": 0.473,
      "step": 12868
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1380136013031006,
      "learning_rate": 8.681182007142475e-07,
      "loss": 0.5199,
      "step": 12869
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0926117897033691,
      "learning_rate": 8.675405319883146e-07,
      "loss": 0.475,
      "step": 12870
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.105962872505188,
      "learning_rate": 8.66963037265876e-07,
      "loss": 0.5099,
      "step": 12871
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9456197619438171,
      "learning_rate": 8.663857165712431e-07,
      "loss": 0.4366,
      "step": 12872
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0361655950546265,
      "learning_rate": 8.658085699287294e-07,
      "loss": 0.5059,
      "step": 12873
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1172592639923096,
      "learning_rate": 8.652315973626362e-07,
      "loss": 0.5217,
      "step": 12874
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0309276580810547,
      "learning_rate": 8.646547988972553e-07,
      "loss": 0.4685,
      "step": 12875
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9785642623901367,
      "learning_rate": 8.64078174556875e-07,
      "loss": 0.5054,
      "step": 12876
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.047278642654419,
      "learning_rate": 8.635017243657751e-07,
      "loss": 0.5194,
      "step": 12877
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.091817021369934,
      "learning_rate": 8.629254483482274e-07,
      "loss": 0.463,
      "step": 12878
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.017472505569458,
      "learning_rate": 8.623493465284987e-07,
      "loss": 0.4432,
      "step": 12879
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.034571647644043,
      "learning_rate": 8.61773418930843e-07,
      "loss": 0.492,
      "step": 12880
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1846771240234375,
      "learning_rate": 8.611976655795135e-07,
      "loss": 0.5708,
      "step": 12881
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1050655841827393,
      "learning_rate": 8.606220864987541e-07,
      "loss": 0.5207,
      "step": 12882
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.016854166984558,
      "learning_rate": 8.600466817127972e-07,
      "loss": 0.4698,
      "step": 12883
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1042780876159668,
      "learning_rate": 8.59471451245873e-07,
      "loss": 0.4675,
      "step": 12884
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.055746078491211,
      "learning_rate": 8.588963951222024e-07,
      "loss": 0.5221,
      "step": 12885
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0114814043045044,
      "learning_rate": 8.583215133659983e-07,
      "loss": 0.5464,
      "step": 12886
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0872836112976074,
      "learning_rate": 8.577468060014688e-07,
      "loss": 0.5093,
      "step": 12887
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.14777672290802,
      "learning_rate": 8.571722730528098e-07,
      "loss": 0.5534,
      "step": 12888
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.007361650466919,
      "learning_rate": 8.565979145442138e-07,
      "loss": 0.5071,
      "step": 12889
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.047307014465332,
      "learning_rate": 8.560237304998681e-07,
      "loss": 0.5,
      "step": 12890
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.188661813735962,
      "learning_rate": 8.554497209439461e-07,
      "loss": 0.5295,
      "step": 12891
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0467709302902222,
      "learning_rate": 8.548758859006184e-07,
      "loss": 0.504,
      "step": 12892
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1030771732330322,
      "learning_rate": 8.543022253940475e-07,
      "loss": 0.4859,
      "step": 12893
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.01668381690979,
      "learning_rate": 8.537287394483878e-07,
      "loss": 0.4997,
      "step": 12894
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.160611867904663,
      "learning_rate": 8.531554280877885e-07,
      "loss": 0.4866,
      "step": 12895
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.108699917793274,
      "learning_rate": 8.525822913363868e-07,
      "loss": 0.539,
      "step": 12896
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0514135360717773,
      "learning_rate": 8.520093292183163e-07,
      "loss": 0.4907,
      "step": 12897
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1542928218841553,
      "learning_rate": 8.514365417577048e-07,
      "loss": 0.479,
      "step": 12898
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0242931842803955,
      "learning_rate": 8.50863928978668e-07,
      "loss": 0.4756,
      "step": 12899
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0869053602218628,
      "learning_rate": 8.502914909053173e-07,
      "loss": 0.508,
      "step": 12900
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0241419076919556,
      "learning_rate": 8.497192275617577e-07,
      "loss": 0.4767,
      "step": 12901
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0105047225952148,
      "learning_rate": 8.491471389720807e-07,
      "loss": 0.5296,
      "step": 12902
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.111020565032959,
      "learning_rate": 8.485752251603807e-07,
      "loss": 0.5032,
      "step": 12903
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0602821111679077,
      "learning_rate": 8.480034861507347e-07,
      "loss": 0.4677,
      "step": 12904
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0135266780853271,
      "learning_rate": 8.474319219672183e-07,
      "loss": 0.5016,
      "step": 12905
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0159438848495483,
      "learning_rate": 8.46860532633898e-07,
      "loss": 0.4794,
      "step": 12906
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0804152488708496,
      "learning_rate": 8.462893181748327e-07,
      "loss": 0.4866,
      "step": 12907
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.025763750076294,
      "learning_rate": 8.457182786140744e-07,
      "loss": 0.5105,
      "step": 12908
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0539695024490356,
      "learning_rate": 8.451474139756693e-07,
      "loss": 0.5277,
      "step": 12909
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0051062107086182,
      "learning_rate": 8.445767242836506e-07,
      "loss": 0.4804,
      "step": 12910
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.993935763835907,
      "learning_rate": 8.440062095620527e-07,
      "loss": 0.4868,
      "step": 12911
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1110044717788696,
      "learning_rate": 8.434358698348944e-07,
      "loss": 0.5325,
      "step": 12912
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0014790296554565,
      "learning_rate": 8.428657051261918e-07,
      "loss": 0.4937,
      "step": 12913
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0092440843582153,
      "learning_rate": 8.422957154599526e-07,
      "loss": 0.4946,
      "step": 12914
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1117631196975708,
      "learning_rate": 8.417259008601775e-07,
      "loss": 0.4752,
      "step": 12915
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0479555130004883,
      "learning_rate": 8.411562613508595e-07,
      "loss": 0.4874,
      "step": 12916
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1826703548431396,
      "learning_rate": 8.405867969559845e-07,
      "loss": 0.5166,
      "step": 12917
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0489535331726074,
      "learning_rate": 8.400175076995287e-07,
      "loss": 0.4758,
      "step": 12918
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0403448343276978,
      "learning_rate": 8.394483936054643e-07,
      "loss": 0.5041,
      "step": 12919
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0457063913345337,
      "learning_rate": 8.388794546977546e-07,
      "loss": 0.4999,
      "step": 12920
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0715712308883667,
      "learning_rate": 8.383106910003552e-07,
      "loss": 0.4867,
      "step": 12921
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1114780902862549,
      "learning_rate": 8.377421025372157e-07,
      "loss": 0.5058,
      "step": 12922
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9921567440032959,
      "learning_rate": 8.371736893322763e-07,
      "loss": 0.4085,
      "step": 12923
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0891449451446533,
      "learning_rate": 8.366054514094718e-07,
      "loss": 0.488,
      "step": 12924
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0652058124542236,
      "learning_rate": 8.360373887927298e-07,
      "loss": 0.5177,
      "step": 12925
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1317729949951172,
      "learning_rate": 8.35469501505966e-07,
      "loss": 0.5144,
      "step": 12926
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0853395462036133,
      "learning_rate": 8.349017895730948e-07,
      "loss": 0.4921,
      "step": 12927
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1067955493927002,
      "learning_rate": 8.343342530180198e-07,
      "loss": 0.4986,
      "step": 12928
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.08108389377594,
      "learning_rate": 8.33766891864638e-07,
      "loss": 0.4834,
      "step": 12929
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1389966011047363,
      "learning_rate": 8.331997061368391e-07,
      "loss": 0.5181,
      "step": 12930
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.072933554649353,
      "learning_rate": 8.326326958585062e-07,
      "loss": 0.5329,
      "step": 12931
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0903488397598267,
      "learning_rate": 8.320658610535115e-07,
      "loss": 0.5116,
      "step": 12932
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.092515230178833,
      "learning_rate": 8.314992017457263e-07,
      "loss": 0.5006,
      "step": 12933
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.00934636592865,
      "learning_rate": 8.30932717959007e-07,
      "loss": 0.4781,
      "step": 12934
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0941412448883057,
      "learning_rate": 8.303664097172087e-07,
      "loss": 0.5036,
      "step": 12935
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0578081607818604,
      "learning_rate": 8.298002770441749e-07,
      "loss": 0.5253,
      "step": 12936
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.956196129322052,
      "learning_rate": 8.292343199637448e-07,
      "loss": 0.5058,
      "step": 12937
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0259432792663574,
      "learning_rate": 8.286685384997484e-07,
      "loss": 0.5223,
      "step": 12938
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0590115785598755,
      "learning_rate": 8.281029326760104e-07,
      "loss": 0.4955,
      "step": 12939
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0075199604034424,
      "learning_rate": 8.275375025163418e-07,
      "loss": 0.5035,
      "step": 12940
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0844745635986328,
      "learning_rate": 8.269722480445569e-07,
      "loss": 0.5579,
      "step": 12941
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0919443368911743,
      "learning_rate": 8.264071692844527e-07,
      "loss": 0.4824,
      "step": 12942
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9813203811645508,
      "learning_rate": 8.258422662598231e-07,
      "loss": 0.483,
      "step": 12943
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1761623620986938,
      "learning_rate": 8.252775389944556e-07,
      "loss": 0.489,
      "step": 12944
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1047816276550293,
      "learning_rate": 8.247129875121274e-07,
      "loss": 0.5299,
      "step": 12945
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0560872554779053,
      "learning_rate": 8.24148611836611e-07,
      "loss": 0.4865,
      "step": 12946
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0733801126480103,
      "learning_rate": 8.235844119916708e-07,
      "loss": 0.4838,
      "step": 12947
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1089249849319458,
      "learning_rate": 8.230203880010612e-07,
      "loss": 0.5146,
      "step": 12948
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.030139446258545,
      "learning_rate": 8.224565398885325e-07,
      "loss": 0.4603,
      "step": 12949
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0687334537506104,
      "learning_rate": 8.218928676778264e-07,
      "loss": 0.5096,
      "step": 12950
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9699050188064575,
      "learning_rate": 8.213293713926767e-07,
      "loss": 0.436,
      "step": 12951
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1305122375488281,
      "learning_rate": 8.207660510568122e-07,
      "loss": 0.5413,
      "step": 12952
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9856038689613342,
      "learning_rate": 8.202029066939482e-07,
      "loss": 0.4482,
      "step": 12953
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9998235106468201,
      "learning_rate": 8.196399383278004e-07,
      "loss": 0.5385,
      "step": 12954
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.179714322090149,
      "learning_rate": 8.190771459820739e-07,
      "loss": 0.5373,
      "step": 12955
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.190774917602539,
      "learning_rate": 8.18514529680463e-07,
      "loss": 0.5386,
      "step": 12956
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0108187198638916,
      "learning_rate": 8.179520894466592e-07,
      "loss": 0.4892,
      "step": 12957
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0275996923446655,
      "learning_rate": 8.173898253043444e-07,
      "loss": 0.4336,
      "step": 12958
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0599771738052368,
      "learning_rate": 8.168277372771937e-07,
      "loss": 0.4618,
      "step": 12959
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9936832785606384,
      "learning_rate": 8.162658253888761e-07,
      "loss": 0.5281,
      "step": 12960
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0354995727539062,
      "learning_rate": 8.157040896630481e-07,
      "loss": 0.4696,
      "step": 12961
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0596461296081543,
      "learning_rate": 8.151425301233656e-07,
      "loss": 0.4751,
      "step": 12962
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1725833415985107,
      "learning_rate": 8.14581146793475e-07,
      "loss": 0.5226,
      "step": 12963
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0640276670455933,
      "learning_rate": 8.140199396970106e-07,
      "loss": 0.5035,
      "step": 12964
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0254626274108887,
      "learning_rate": 8.13458908857605e-07,
      "loss": 0.4465,
      "step": 12965
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0551642179489136,
      "learning_rate": 8.128980542988801e-07,
      "loss": 0.4637,
      "step": 12966
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.061268925666809,
      "learning_rate": 8.12337376044453e-07,
      "loss": 0.4741,
      "step": 12967
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.105828881263733,
      "learning_rate": 8.117768741179322e-07,
      "loss": 0.5346,
      "step": 12968
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0640383958816528,
      "learning_rate": 8.112165485429163e-07,
      "loss": 0.5018,
      "step": 12969
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0903797149658203,
      "learning_rate": 8.106563993429983e-07,
      "loss": 0.4948,
      "step": 12970
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0028164386749268,
      "learning_rate": 8.100964265417682e-07,
      "loss": 0.476,
      "step": 12971
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1049624681472778,
      "learning_rate": 8.09536630162801e-07,
      "loss": 0.5115,
      "step": 12972
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.179762601852417,
      "learning_rate": 8.089770102296685e-07,
      "loss": 0.4664,
      "step": 12973
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0441378355026245,
      "learning_rate": 8.084175667659345e-07,
      "loss": 0.5025,
      "step": 12974
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.024807095527649,
      "learning_rate": 8.078582997951556e-07,
      "loss": 0.5021,
      "step": 12975
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0215458869934082,
      "learning_rate": 8.072992093408816e-07,
      "loss": 0.5306,
      "step": 12976
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0705639123916626,
      "learning_rate": 8.067402954266512e-07,
      "loss": 0.5316,
      "step": 12977
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0934439897537231,
      "learning_rate": 8.061815580759996e-07,
      "loss": 0.4898,
      "step": 12978
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0901581048965454,
      "learning_rate": 8.056229973124529e-07,
      "loss": 0.506,
      "step": 12979
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0371161699295044,
      "learning_rate": 8.050646131595313e-07,
      "loss": 0.5112,
      "step": 12980
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9929707050323486,
      "learning_rate": 8.045064056407453e-07,
      "loss": 0.4959,
      "step": 12981
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.021591305732727,
      "learning_rate": 8.039483747796012e-07,
      "loss": 0.4946,
      "step": 12982
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0111936330795288,
      "learning_rate": 8.033905205995913e-07,
      "loss": 0.4957,
      "step": 12983
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.089510202407837,
      "learning_rate": 8.0283284312421e-07,
      "loss": 0.4813,
      "step": 12984
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0633338689804077,
      "learning_rate": 8.022753423769359e-07,
      "loss": 0.5185,
      "step": 12985
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9597094058990479,
      "learning_rate": 8.017180183812439e-07,
      "loss": 0.4766,
      "step": 12986
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0139012336730957,
      "learning_rate": 8.011608711606017e-07,
      "loss": 0.4829,
      "step": 12987
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.062382698059082,
      "learning_rate": 8.006039007384681e-07,
      "loss": 0.5149,
      "step": 12988
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1986846923828125,
      "learning_rate": 8.000471071382959e-07,
      "loss": 0.5263,
      "step": 12989
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0365426540374756,
      "learning_rate": 7.99490490383531e-07,
      "loss": 0.5102,
      "step": 12990
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0646085739135742,
      "learning_rate": 7.989340504976062e-07,
      "loss": 0.5231,
      "step": 12991
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0220482349395752,
      "learning_rate": 7.983777875039567e-07,
      "loss": 0.4728,
      "step": 12992
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0070226192474365,
      "learning_rate": 7.978217014260009e-07,
      "loss": 0.495,
      "step": 12993
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1144849061965942,
      "learning_rate": 7.972657922871546e-07,
      "loss": 0.4909,
      "step": 12994
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0617939233779907,
      "learning_rate": 7.967100601108258e-07,
      "loss": 0.5118,
      "step": 12995
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0664457082748413,
      "learning_rate": 7.961545049204145e-07,
      "loss": 0.5491,
      "step": 12996
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9803445339202881,
      "learning_rate": 7.955991267393127e-07,
      "loss": 0.4642,
      "step": 12997
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0978424549102783,
      "learning_rate": 7.950439255909065e-07,
      "loss": 0.5544,
      "step": 12998
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0519658327102661,
      "learning_rate": 7.944889014985718e-07,
      "loss": 0.5123,
      "step": 12999
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.081263780593872,
      "learning_rate": 7.939340544856783e-07,
      "loss": 0.5062,
      "step": 13000
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.050601601600647,
      "learning_rate": 7.933793845755922e-07,
      "loss": 0.5167,
      "step": 13001
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0440396070480347,
      "learning_rate": 7.928248917916653e-07,
      "loss": 0.4979,
      "step": 13002
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0332825183868408,
      "learning_rate": 7.922705761572464e-07,
      "loss": 0.4787,
      "step": 13003
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9679126143455505,
      "learning_rate": 7.91716437695676e-07,
      "loss": 0.4428,
      "step": 13004
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1437848806381226,
      "learning_rate": 7.911624764302872e-07,
      "loss": 0.5602,
      "step": 13005
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0696083307266235,
      "learning_rate": 7.906086923844059e-07,
      "loss": 0.4994,
      "step": 13006
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0181982517242432,
      "learning_rate": 7.900550855813477e-07,
      "loss": 0.5032,
      "step": 13007
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1297941207885742,
      "learning_rate": 7.895016560444241e-07,
      "loss": 0.5132,
      "step": 13008
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.085738182067871,
      "learning_rate": 7.889484037969403e-07,
      "loss": 0.531,
      "step": 13009
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0692535638809204,
      "learning_rate": 7.883953288621887e-07,
      "loss": 0.5461,
      "step": 13010
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1931544542312622,
      "learning_rate": 7.878424312634592e-07,
      "loss": 0.5064,
      "step": 13011
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1276174783706665,
      "learning_rate": 7.87289711024033e-07,
      "loss": 0.5223,
      "step": 13012
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0550527572631836,
      "learning_rate": 7.867371681671793e-07,
      "loss": 0.5255,
      "step": 13013
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0752570629119873,
      "learning_rate": 7.861848027161694e-07,
      "loss": 0.5363,
      "step": 13014
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9889829754829407,
      "learning_rate": 7.856326146942572e-07,
      "loss": 0.4606,
      "step": 13015
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.053361415863037,
      "learning_rate": 7.85080604124695e-07,
      "loss": 0.5499,
      "step": 13016
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0334244966506958,
      "learning_rate": 7.845287710307258e-07,
      "loss": 0.4767,
      "step": 13017
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1344623565673828,
      "learning_rate": 7.839771154355858e-07,
      "loss": 0.519,
      "step": 13018
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0209664106369019,
      "learning_rate": 7.834256373625027e-07,
      "loss": 0.5122,
      "step": 13019
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0017036199569702,
      "learning_rate": 7.828743368346991e-07,
      "loss": 0.4465,
      "step": 13020
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0459413528442383,
      "learning_rate": 7.823232138753845e-07,
      "loss": 0.4616,
      "step": 13021
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1484712362289429,
      "learning_rate": 7.817722685077689e-07,
      "loss": 0.4665,
      "step": 13022
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9936702251434326,
      "learning_rate": 7.812215007550483e-07,
      "loss": 0.5223,
      "step": 13023
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1362011432647705,
      "learning_rate": 7.806709106404142e-07,
      "loss": 0.5363,
      "step": 13024
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0711592435836792,
      "learning_rate": 7.801204981870508e-07,
      "loss": 0.5046,
      "step": 13025
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0723801851272583,
      "learning_rate": 7.795702634181318e-07,
      "loss": 0.4827,
      "step": 13026
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0045136213302612,
      "learning_rate": 7.790202063568276e-07,
      "loss": 0.4553,
      "step": 13027
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0694032907485962,
      "learning_rate": 7.784703270263006e-07,
      "loss": 0.504,
      "step": 13028
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0534634590148926,
      "learning_rate": 7.779206254497007e-07,
      "loss": 0.4675,
      "step": 13029
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.056309700012207,
      "learning_rate": 7.773711016501762e-07,
      "loss": 0.529,
      "step": 13030
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.064368486404419,
      "learning_rate": 7.76821755650865e-07,
      "loss": 0.55,
      "step": 13031
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0615298748016357,
      "learning_rate": 7.762725874748983e-07,
      "loss": 0.5037,
      "step": 13032
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1085795164108276,
      "learning_rate": 7.757235971454008e-07,
      "loss": 0.4766,
      "step": 13033
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9743221402168274,
      "learning_rate": 7.751747846854851e-07,
      "loss": 0.4875,
      "step": 13034
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1090971231460571,
      "learning_rate": 7.746261501182633e-07,
      "loss": 0.5021,
      "step": 13035
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.017557978630066,
      "learning_rate": 7.740776934668365e-07,
      "loss": 0.4852,
      "step": 13036
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0135324001312256,
      "learning_rate": 7.73529414754296e-07,
      "loss": 0.4662,
      "step": 13037
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.132223129272461,
      "learning_rate": 7.72981314003729e-07,
      "loss": 0.4664,
      "step": 13038
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.087611198425293,
      "learning_rate": 7.724333912382143e-07,
      "loss": 0.5009,
      "step": 13039
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.010160207748413,
      "learning_rate": 7.718856464808222e-07,
      "loss": 0.4671,
      "step": 13040
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9534567594528198,
      "learning_rate": 7.713380797546188e-07,
      "loss": 0.4645,
      "step": 13041
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1297147274017334,
      "learning_rate": 7.707906910826574e-07,
      "loss": 0.5232,
      "step": 13042
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0202480554580688,
      "learning_rate": 7.702434804879861e-07,
      "loss": 0.4936,
      "step": 13043
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0754560232162476,
      "learning_rate": 7.696964479936497e-07,
      "loss": 0.4756,
      "step": 13044
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0832222700119019,
      "learning_rate": 7.691495936226789e-07,
      "loss": 0.4753,
      "step": 13045
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1323879957199097,
      "learning_rate": 7.686029173981008e-07,
      "loss": 0.5314,
      "step": 13046
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.097445011138916,
      "learning_rate": 7.680564193429336e-07,
      "loss": 0.5147,
      "step": 13047
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1266767978668213,
      "learning_rate": 7.675100994801888e-07,
      "loss": 0.4727,
      "step": 13048
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0783687829971313,
      "learning_rate": 7.669639578328713e-07,
      "loss": 0.4799,
      "step": 13049
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0944496393203735,
      "learning_rate": 7.664179944239746e-07,
      "loss": 0.5092,
      "step": 13050
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0033917427062988,
      "learning_rate": 7.658722092764876e-07,
      "loss": 0.4229,
      "step": 13051
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0313786268234253,
      "learning_rate": 7.653266024133943e-07,
      "loss": 0.4678,
      "step": 13052
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0790505409240723,
      "learning_rate": 7.647811738576655e-07,
      "loss": 0.5071,
      "step": 13053
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.064852237701416,
      "learning_rate": 7.642359236322683e-07,
      "loss": 0.4768,
      "step": 13054
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.03118896484375,
      "learning_rate": 7.63690851760161e-07,
      "loss": 0.4928,
      "step": 13055
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1065315008163452,
      "learning_rate": 7.631459582642947e-07,
      "loss": 0.4912,
      "step": 13056
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.064281702041626,
      "learning_rate": 7.626012431676138e-07,
      "loss": 0.5225,
      "step": 13057
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0789825916290283,
      "learning_rate": 7.620567064930545e-07,
      "loss": 0.5028,
      "step": 13058
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.042149543762207,
      "learning_rate": 7.615123482635433e-07,
      "loss": 0.5042,
      "step": 13059
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.112006664276123,
      "learning_rate": 7.609681685020026e-07,
      "loss": 0.5103,
      "step": 13060
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.028239369392395,
      "learning_rate": 7.604241672313461e-07,
      "loss": 0.4712,
      "step": 13061
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0118327140808105,
      "learning_rate": 7.59880344474479e-07,
      "loss": 0.4952,
      "step": 13062
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0231777429580688,
      "learning_rate": 7.593367002543018e-07,
      "loss": 0.4922,
      "step": 13063
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0467559099197388,
      "learning_rate": 7.587932345937016e-07,
      "loss": 0.4955,
      "step": 13064
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0370157957077026,
      "learning_rate": 7.582499475155653e-07,
      "loss": 0.5388,
      "step": 13065
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0274789333343506,
      "learning_rate": 7.577068390427689e-07,
      "loss": 0.4585,
      "step": 13066
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9453758001327515,
      "learning_rate": 7.571639091981786e-07,
      "loss": 0.463,
      "step": 13067
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0819838047027588,
      "learning_rate": 7.566211580046562e-07,
      "loss": 0.5143,
      "step": 13068
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.992570698261261,
      "learning_rate": 7.56078585485055e-07,
      "loss": 0.463,
      "step": 13069
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0368421077728271,
      "learning_rate": 7.555361916622217e-07,
      "loss": 0.512,
      "step": 13070
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.010269045829773,
      "learning_rate": 7.549939765589942e-07,
      "loss": 0.4379,
      "step": 13071
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.059167742729187,
      "learning_rate": 7.544519401982025e-07,
      "loss": 0.5091,
      "step": 13072
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1328742504119873,
      "learning_rate": 7.539100826026691e-07,
      "loss": 0.527,
      "step": 13073
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1374634504318237,
      "learning_rate": 7.533684037952133e-07,
      "loss": 0.4968,
      "step": 13074
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0074968338012695,
      "learning_rate": 7.528269037986402e-07,
      "loss": 0.5453,
      "step": 13075
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.132983684539795,
      "learning_rate": 7.522855826357511e-07,
      "loss": 0.492,
      "step": 13076
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9377274513244629,
      "learning_rate": 7.517444403293394e-07,
      "loss": 0.4618,
      "step": 13077
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0819823741912842,
      "learning_rate": 7.512034769021909e-07,
      "loss": 0.5267,
      "step": 13078
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1434528827667236,
      "learning_rate": 7.506626923770843e-07,
      "loss": 0.5112,
      "step": 13079
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.046053171157837,
      "learning_rate": 7.501220867767883e-07,
      "loss": 0.5212,
      "step": 13080
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0413637161254883,
      "learning_rate": 7.495816601240664e-07,
      "loss": 0.5084,
      "step": 13081
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0843172073364258,
      "learning_rate": 7.490414124416761e-07,
      "loss": 0.5339,
      "step": 13082
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.107027292251587,
      "learning_rate": 7.485013437523636e-07,
      "loss": 0.5402,
      "step": 13083
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.105094313621521,
      "learning_rate": 7.479614540788687e-07,
      "loss": 0.5307,
      "step": 13084
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0860050916671753,
      "learning_rate": 7.474217434439263e-07,
      "loss": 0.4937,
      "step": 13085
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0668641328811646,
      "learning_rate": 7.468822118702596e-07,
      "loss": 0.4803,
      "step": 13086
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0991917848587036,
      "learning_rate": 7.463428593805894e-07,
      "loss": 0.4805,
      "step": 13087
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0379934310913086,
      "learning_rate": 7.458036859976225e-07,
      "loss": 0.5311,
      "step": 13088
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1096937656402588,
      "learning_rate": 7.452646917440631e-07,
      "loss": 0.5478,
      "step": 13089
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.085133671760559,
      "learning_rate": 7.447258766426063e-07,
      "loss": 0.51,
      "step": 13090
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.214627742767334,
      "learning_rate": 7.441872407159401e-07,
      "loss": 0.4982,
      "step": 13091
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0035841464996338,
      "learning_rate": 7.43648783986744e-07,
      "loss": 0.5197,
      "step": 13092
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0719767808914185,
      "learning_rate": 7.431105064776922e-07,
      "loss": 0.5233,
      "step": 13093
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0103352069854736,
      "learning_rate": 7.425724082114455e-07,
      "loss": 0.4828,
      "step": 13094
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.091241478919983,
      "learning_rate": 7.420344892106674e-07,
      "loss": 0.4777,
      "step": 13095
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0513263940811157,
      "learning_rate": 7.414967494980024e-07,
      "loss": 0.5093,
      "step": 13096
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0273000001907349,
      "learning_rate": 7.40959189096096e-07,
      "loss": 0.4405,
      "step": 13097
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.095465064048767,
      "learning_rate": 7.404218080275816e-07,
      "loss": 0.529,
      "step": 13098
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.068629503250122,
      "learning_rate": 7.398846063150866e-07,
      "loss": 0.4608,
      "step": 13099
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.020584225654602,
      "learning_rate": 7.393475839812314e-07,
      "loss": 0.5202,
      "step": 13100
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0594336986541748,
      "learning_rate": 7.388107410486289e-07,
      "loss": 0.4967,
      "step": 13101
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0724962949752808,
      "learning_rate": 7.3827407753988e-07,
      "loss": 0.496,
      "step": 13102
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9776265025138855,
      "learning_rate": 7.377375934775865e-07,
      "loss": 0.4839,
      "step": 13103
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0062637329101562,
      "learning_rate": 7.372012888843344e-07,
      "loss": 0.506,
      "step": 13104
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9793121814727783,
      "learning_rate": 7.366651637827065e-07,
      "loss": 0.492,
      "step": 13105
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0513062477111816,
      "learning_rate": 7.361292181952795e-07,
      "loss": 0.5025,
      "step": 13106
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2303926944732666,
      "learning_rate": 7.355934521446151e-07,
      "loss": 0.485,
      "step": 13107
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.075791835784912,
      "learning_rate": 7.350578656532776e-07,
      "loss": 0.5139,
      "step": 13108
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0517456531524658,
      "learning_rate": 7.345224587438171e-07,
      "loss": 0.4412,
      "step": 13109
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1046347618103027,
      "learning_rate": 7.339872314387763e-07,
      "loss": 0.5332,
      "step": 13110
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1530647277832031,
      "learning_rate": 7.334521837606934e-07,
      "loss": 0.5211,
      "step": 13111
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9838554263114929,
      "learning_rate": 7.329173157320962e-07,
      "loss": 0.4616,
      "step": 13112
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0944254398345947,
      "learning_rate": 7.323826273755069e-07,
      "loss": 0.5109,
      "step": 13113
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.10060715675354,
      "learning_rate": 7.318481187134408e-07,
      "loss": 0.5204,
      "step": 13114
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1013379096984863,
      "learning_rate": 7.313137897683997e-07,
      "loss": 0.5227,
      "step": 13115
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.012404441833496,
      "learning_rate": 7.30779640562887e-07,
      "loss": 0.5431,
      "step": 13116
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.021898865699768,
      "learning_rate": 7.302456711193928e-07,
      "loss": 0.4886,
      "step": 13117
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.070927381515503,
      "learning_rate": 7.297118814603987e-07,
      "loss": 0.4834,
      "step": 13118
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9357616305351257,
      "learning_rate": 7.291782716083823e-07,
      "loss": 0.4724,
      "step": 13119
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0846022367477417,
      "learning_rate": 7.286448415858116e-07,
      "loss": 0.5166,
      "step": 13120
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.008909821510315,
      "learning_rate": 7.281115914151477e-07,
      "loss": 0.4524,
      "step": 13121
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0392085313796997,
      "learning_rate": 7.275785211188441e-07,
      "loss": 0.5003,
      "step": 13122
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0051409006118774,
      "learning_rate": 7.270456307193474e-07,
      "loss": 0.5078,
      "step": 13123
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1217042207717896,
      "learning_rate": 7.265129202390924e-07,
      "loss": 0.4688,
      "step": 13124
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9774218201637268,
      "learning_rate": 7.259803897005141e-07,
      "loss": 0.488,
      "step": 13125
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0326730012893677,
      "learning_rate": 7.254480391260321e-07,
      "loss": 0.4403,
      "step": 13126
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0553927421569824,
      "learning_rate": 7.249158685380631e-07,
      "loss": 0.5115,
      "step": 13127
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.042048454284668,
      "learning_rate": 7.243838779590151e-07,
      "loss": 0.4818,
      "step": 13128
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0369988679885864,
      "learning_rate": 7.238520674112881e-07,
      "loss": 0.4685,
      "step": 13129
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0481699705123901,
      "learning_rate": 7.233204369172753e-07,
      "loss": 0.476,
      "step": 13130
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0384000539779663,
      "learning_rate": 7.22788986499362e-07,
      "loss": 0.4698,
      "step": 13131
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0750094652175903,
      "learning_rate": 7.222577161799232e-07,
      "loss": 0.5098,
      "step": 13132
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1386425495147705,
      "learning_rate": 7.217266259813332e-07,
      "loss": 0.5112,
      "step": 13133
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0001041889190674,
      "learning_rate": 7.211957159259503e-07,
      "loss": 0.464,
      "step": 13134
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0347130298614502,
      "learning_rate": 7.206649860361314e-07,
      "loss": 0.4863,
      "step": 13135
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1184300184249878,
      "learning_rate": 7.201344363342245e-07,
      "loss": 0.5083,
      "step": 13136
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1076111793518066,
      "learning_rate": 7.196040668425653e-07,
      "loss": 0.5337,
      "step": 13137
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.092587947845459,
      "learning_rate": 7.190738775834894e-07,
      "loss": 0.5188,
      "step": 13138
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0376991033554077,
      "learning_rate": 7.185438685793217e-07,
      "loss": 0.5127,
      "step": 13139
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0892513990402222,
      "learning_rate": 7.180140398523761e-07,
      "loss": 0.5143,
      "step": 13140
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0623342990875244,
      "learning_rate": 7.174843914249636e-07,
      "loss": 0.4956,
      "step": 13141
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0941516160964966,
      "learning_rate": 7.169549233193857e-07,
      "loss": 0.5069,
      "step": 13142
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1275255680084229,
      "learning_rate": 7.164256355579363e-07,
      "loss": 0.5076,
      "step": 13143
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0641416311264038,
      "learning_rate": 7.158965281629027e-07,
      "loss": 0.5074,
      "step": 13144
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0748405456542969,
      "learning_rate": 7.153676011565613e-07,
      "loss": 0.5199,
      "step": 13145
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9913454651832581,
      "learning_rate": 7.148388545611856e-07,
      "loss": 0.4566,
      "step": 13146
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9696449637413025,
      "learning_rate": 7.143102883990405e-07,
      "loss": 0.4981,
      "step": 13147
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0366780757904053,
      "learning_rate": 7.137819026923786e-07,
      "loss": 0.449,
      "step": 13148
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0798648595809937,
      "learning_rate": 7.132536974634508e-07,
      "loss": 0.5199,
      "step": 13149
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.049294352531433,
      "learning_rate": 7.127256727344967e-07,
      "loss": 0.487,
      "step": 13150
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0658286809921265,
      "learning_rate": 7.121978285277503e-07,
      "loss": 0.5006,
      "step": 13151
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0541974306106567,
      "learning_rate": 7.116701648654384e-07,
      "loss": 0.5293,
      "step": 13152
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.053865671157837,
      "learning_rate": 7.11142681769777e-07,
      "loss": 0.4901,
      "step": 13153
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0910694599151611,
      "learning_rate": 7.106153792629761e-07,
      "loss": 0.5032,
      "step": 13154
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9611359238624573,
      "learning_rate": 7.100882573672419e-07,
      "loss": 0.4711,
      "step": 13155
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0032509565353394,
      "learning_rate": 7.095613161047666e-07,
      "loss": 0.4942,
      "step": 13156
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0794392824172974,
      "learning_rate": 7.09034555497739e-07,
      "loss": 0.4868,
      "step": 13157
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0678043365478516,
      "learning_rate": 7.085079755683389e-07,
      "loss": 0.4887,
      "step": 13158
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0658090114593506,
      "learning_rate": 7.079815763387393e-07,
      "loss": 0.5243,
      "step": 13159
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0863361358642578,
      "learning_rate": 7.074553578311055e-07,
      "loss": 0.4692,
      "step": 13160
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.046759843826294,
      "learning_rate": 7.06929320067593e-07,
      "loss": 0.5326,
      "step": 13161
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1394504308700562,
      "learning_rate": 7.064034630703515e-07,
      "loss": 0.5297,
      "step": 13162
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1047179698944092,
      "learning_rate": 7.058777868615258e-07,
      "loss": 0.5167,
      "step": 13163
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.058111548423767,
      "learning_rate": 7.053522914632466e-07,
      "loss": 0.518,
      "step": 13164
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0377641916275024,
      "learning_rate": 7.048269768976429e-07,
      "loss": 0.4877,
      "step": 13165
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1154805421829224,
      "learning_rate": 7.043018431868348e-07,
      "loss": 0.4773,
      "step": 13166
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.08950936794281,
      "learning_rate": 7.037768903529302e-07,
      "loss": 0.5348,
      "step": 13167
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.030245065689087,
      "learning_rate": 7.032521184180369e-07,
      "loss": 0.5246,
      "step": 13168
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0297592878341675,
      "learning_rate": 7.027275274042489e-07,
      "loss": 0.5093,
      "step": 13169
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.093056321144104,
      "learning_rate": 7.022031173336557e-07,
      "loss": 0.4911,
      "step": 13170
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0303655862808228,
      "learning_rate": 7.016788882283382e-07,
      "loss": 0.5291,
      "step": 13171
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1379703283309937,
      "learning_rate": 7.011548401103696e-07,
      "loss": 0.5298,
      "step": 13172
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0572259426116943,
      "learning_rate": 7.006309730018168e-07,
      "loss": 0.5104,
      "step": 13173
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0219038724899292,
      "learning_rate": 7.001072869247378e-07,
      "loss": 0.5306,
      "step": 13174
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0922071933746338,
      "learning_rate": 6.995837819011808e-07,
      "loss": 0.5321,
      "step": 13175
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9969636797904968,
      "learning_rate": 6.990604579531929e-07,
      "loss": 0.4978,
      "step": 13176
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1060311794281006,
      "learning_rate": 6.985373151028058e-07,
      "loss": 0.5127,
      "step": 13177
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.023337721824646,
      "learning_rate": 6.980143533720491e-07,
      "loss": 0.4736,
      "step": 13178
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0479949712753296,
      "learning_rate": 6.974915727829423e-07,
      "loss": 0.4933,
      "step": 13179
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1665982007980347,
      "learning_rate": 6.96968973357498e-07,
      "loss": 0.5469,
      "step": 13180
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.982250452041626,
      "learning_rate": 6.964465551177208e-07,
      "loss": 0.4244,
      "step": 13181
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.029781460762024,
      "learning_rate": 6.959243180856096e-07,
      "loss": 0.5439,
      "step": 13182
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0114187002182007,
      "learning_rate": 6.954022622831514e-07,
      "loss": 0.4516,
      "step": 13183
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0797160863876343,
      "learning_rate": 6.948803877323296e-07,
      "loss": 0.4391,
      "step": 13184
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0194454193115234,
      "learning_rate": 6.943586944551178e-07,
      "loss": 0.4993,
      "step": 13185
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0202571153640747,
      "learning_rate": 6.938371824734835e-07,
      "loss": 0.5219,
      "step": 13186
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1259692907333374,
      "learning_rate": 6.933158518093852e-07,
      "loss": 0.5369,
      "step": 13187
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1246204376220703,
      "learning_rate": 6.927947024847748e-07,
      "loss": 0.5125,
      "step": 13188
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1178438663482666,
      "learning_rate": 6.922737345215952e-07,
      "loss": 0.4701,
      "step": 13189
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9730929732322693,
      "learning_rate": 6.91752947941785e-07,
      "loss": 0.4824,
      "step": 13190
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0151828527450562,
      "learning_rate": 6.912323427672691e-07,
      "loss": 0.4812,
      "step": 13191
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0726484060287476,
      "learning_rate": 6.907119190199706e-07,
      "loss": 0.4955,
      "step": 13192
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.059969425201416,
      "learning_rate": 6.901916767218019e-07,
      "loss": 0.4861,
      "step": 13193
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0365266799926758,
      "learning_rate": 6.896716158946692e-07,
      "loss": 0.4832,
      "step": 13194
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0280224084854126,
      "learning_rate": 6.891517365604705e-07,
      "loss": 0.528,
      "step": 13195
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0622800588607788,
      "learning_rate": 6.886320387410967e-07,
      "loss": 0.5447,
      "step": 13196
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9725284576416016,
      "learning_rate": 6.881125224584273e-07,
      "loss": 0.4566,
      "step": 13197
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.090033769607544,
      "learning_rate": 6.875931877343417e-07,
      "loss": 0.4939,
      "step": 13198
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.004368543624878,
      "learning_rate": 6.870740345907046e-07,
      "loss": 0.5051,
      "step": 13199
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1019244194030762,
      "learning_rate": 6.865550630493756e-07,
      "loss": 0.4604,
      "step": 13200
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1146312952041626,
      "learning_rate": 6.860362731322079e-07,
      "loss": 0.5348,
      "step": 13201
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0931274890899658,
      "learning_rate": 6.855176648610457e-07,
      "loss": 0.5044,
      "step": 13202
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.031827449798584,
      "learning_rate": 6.849992382577253e-07,
      "loss": 0.4897,
      "step": 13203
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0737539529800415,
      "learning_rate": 6.844809933440776e-07,
      "loss": 0.4928,
      "step": 13204
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1895108222961426,
      "learning_rate": 6.839629301419204e-07,
      "loss": 0.476,
      "step": 13205
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0393731594085693,
      "learning_rate": 6.83445048673072e-07,
      "loss": 0.4769,
      "step": 13206
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9944177269935608,
      "learning_rate": 6.829273489593352e-07,
      "loss": 0.4725,
      "step": 13207
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0164682865142822,
      "learning_rate": 6.824098310225097e-07,
      "loss": 0.4965,
      "step": 13208
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1275815963745117,
      "learning_rate": 6.818924948843863e-07,
      "loss": 0.5223,
      "step": 13209
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0147440433502197,
      "learning_rate": 6.81375340566749e-07,
      "loss": 0.5011,
      "step": 13210
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0180290937423706,
      "learning_rate": 6.808583680913722e-07,
      "loss": 0.4858,
      "step": 13211
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0238807201385498,
      "learning_rate": 6.803415774800253e-07,
      "loss": 0.4898,
      "step": 13212
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0540781021118164,
      "learning_rate": 6.798249687544667e-07,
      "loss": 0.4693,
      "step": 13213
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0231060981750488,
      "learning_rate": 6.793085419364498e-07,
      "loss": 0.4837,
      "step": 13214
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.021544098854065,
      "learning_rate": 6.787922970477196e-07,
      "loss": 0.4942,
      "step": 13215
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0937020778656006,
      "learning_rate": 6.782762341100135e-07,
      "loss": 0.5429,
      "step": 13216
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9730556607246399,
      "learning_rate": 6.777603531450617e-07,
      "loss": 0.4739,
      "step": 13217
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0430997610092163,
      "learning_rate": 6.772446541745836e-07,
      "loss": 0.5048,
      "step": 13218
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0744502544403076,
      "learning_rate": 6.767291372202967e-07,
      "loss": 0.5076,
      "step": 13219
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.154815912246704,
      "learning_rate": 6.762138023039072e-07,
      "loss": 0.5316,
      "step": 13220
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0704598426818848,
      "learning_rate": 6.756986494471119e-07,
      "loss": 0.5289,
      "step": 13221
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0126025676727295,
      "learning_rate": 6.751836786716032e-07,
      "loss": 0.517,
      "step": 13222
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.066314458847046,
      "learning_rate": 6.74668889999065e-07,
      "loss": 0.4967,
      "step": 13223
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0445787906646729,
      "learning_rate": 6.741542834511727e-07,
      "loss": 0.4835,
      "step": 13224
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.01438570022583,
      "learning_rate": 6.736398590495968e-07,
      "loss": 0.4443,
      "step": 13225
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1149402856826782,
      "learning_rate": 6.731256168159939e-07,
      "loss": 0.5227,
      "step": 13226
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0016542673110962,
      "learning_rate": 6.726115567720198e-07,
      "loss": 0.4791,
      "step": 13227
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0544713735580444,
      "learning_rate": 6.720976789393202e-07,
      "loss": 0.5225,
      "step": 13228
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.021567702293396,
      "learning_rate": 6.71583983339531e-07,
      "loss": 0.503,
      "step": 13229
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.028765082359314,
      "learning_rate": 6.710704699942827e-07,
      "loss": 0.4991,
      "step": 13230
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0623314380645752,
      "learning_rate": 6.705571389251975e-07,
      "loss": 0.4521,
      "step": 13231
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1302554607391357,
      "learning_rate": 6.700439901538902e-07,
      "loss": 0.4692,
      "step": 13232
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.114856243133545,
      "learning_rate": 6.695310237019692e-07,
      "loss": 0.4972,
      "step": 13233
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0923891067504883,
      "learning_rate": 6.690182395910305e-07,
      "loss": 0.5548,
      "step": 13234
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0090917348861694,
      "learning_rate": 6.685056378426663e-07,
      "loss": 0.4718,
      "step": 13235
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1440528631210327,
      "learning_rate": 6.679932184784638e-07,
      "loss": 0.5182,
      "step": 13236
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0652399063110352,
      "learning_rate": 6.674809815199962e-07,
      "loss": 0.4616,
      "step": 13237
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1093815565109253,
      "learning_rate": 6.669689269888325e-07,
      "loss": 0.5176,
      "step": 13238
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0881990194320679,
      "learning_rate": 6.664570549065336e-07,
      "loss": 0.4498,
      "step": 13239
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0387650728225708,
      "learning_rate": 6.659453652946529e-07,
      "loss": 0.486,
      "step": 13240
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0223878622055054,
      "learning_rate": 6.654338581747366e-07,
      "loss": 0.5146,
      "step": 13241
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1007391214370728,
      "learning_rate": 6.649225335683213e-07,
      "loss": 0.5279,
      "step": 13242
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2238249778747559,
      "learning_rate": 6.644113914969369e-07,
      "loss": 0.5656,
      "step": 13243
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9655773639678955,
      "learning_rate": 6.639004319821063e-07,
      "loss": 0.4897,
      "step": 13244
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0541054010391235,
      "learning_rate": 6.63389655045345e-07,
      "loss": 0.5037,
      "step": 13245
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0961520671844482,
      "learning_rate": 6.628790607081586e-07,
      "loss": 0.5069,
      "step": 13246
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.033692717552185,
      "learning_rate": 6.623686489920489e-07,
      "loss": 0.4749,
      "step": 13247
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1215088367462158,
      "learning_rate": 6.61858419918503e-07,
      "loss": 0.4896,
      "step": 13248
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1026906967163086,
      "learning_rate": 6.613483735090104e-07,
      "loss": 0.5227,
      "step": 13249
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0389769077301025,
      "learning_rate": 6.608385097850439e-07,
      "loss": 0.4894,
      "step": 13250
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.054916501045227,
      "learning_rate": 6.603288287680726e-07,
      "loss": 0.5115,
      "step": 13251
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0588915348052979,
      "learning_rate": 6.598193304795575e-07,
      "loss": 0.4646,
      "step": 13252
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0616204738616943,
      "learning_rate": 6.593100149409521e-07,
      "loss": 0.5235,
      "step": 13253
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1210933923721313,
      "learning_rate": 6.588008821737019e-07,
      "loss": 0.5045,
      "step": 13254
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0720219612121582,
      "learning_rate": 6.582919321992459e-07,
      "loss": 0.5223,
      "step": 13255
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1221983432769775,
      "learning_rate": 6.577831650390104e-07,
      "loss": 0.4741,
      "step": 13256
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.176119327545166,
      "learning_rate": 6.572745807144226e-07,
      "loss": 0.4837,
      "step": 13257
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0502431392669678,
      "learning_rate": 6.567661792468944e-07,
      "loss": 0.4666,
      "step": 13258
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.075438380241394,
      "learning_rate": 6.562579606578328e-07,
      "loss": 0.4939,
      "step": 13259
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0331934690475464,
      "learning_rate": 6.557499249686377e-07,
      "loss": 0.5035,
      "step": 13260
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0521024465560913,
      "learning_rate": 6.552420722007008e-07,
      "loss": 0.4854,
      "step": 13261
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1086888313293457,
      "learning_rate": 6.547344023754065e-07,
      "loss": 0.5503,
      "step": 13262
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1222712993621826,
      "learning_rate": 6.542269155141306e-07,
      "loss": 0.4943,
      "step": 13263
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0259684324264526,
      "learning_rate": 6.537196116382411e-07,
      "loss": 0.4563,
      "step": 13264
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0945380926132202,
      "learning_rate": 6.532124907690979e-07,
      "loss": 0.4994,
      "step": 13265
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9539095163345337,
      "learning_rate": 6.527055529280574e-07,
      "loss": 0.4748,
      "step": 13266
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1312283277511597,
      "learning_rate": 6.521987981364614e-07,
      "loss": 0.5096,
      "step": 13267
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0389736890792847,
      "learning_rate": 6.516922264156495e-07,
      "loss": 0.4795,
      "step": 13268
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.989928662776947,
      "learning_rate": 6.511858377869517e-07,
      "loss": 0.4759,
      "step": 13269
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0273138284683228,
      "learning_rate": 6.506796322716891e-07,
      "loss": 0.4625,
      "step": 13270
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0603842735290527,
      "learning_rate": 6.501736098911787e-07,
      "loss": 0.4699,
      "step": 13271
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0442575216293335,
      "learning_rate": 6.496677706667243e-07,
      "loss": 0.504,
      "step": 13272
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9706287384033203,
      "learning_rate": 6.491621146196253e-07,
      "loss": 0.4683,
      "step": 13273
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0698890686035156,
      "learning_rate": 6.486566417711765e-07,
      "loss": 0.4817,
      "step": 13274
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.066401720046997,
      "learning_rate": 6.481513521426581e-07,
      "loss": 0.4917,
      "step": 13275
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9655542373657227,
      "learning_rate": 6.476462457553473e-07,
      "loss": 0.4641,
      "step": 13276
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.100365400314331,
      "learning_rate": 6.471413226305134e-07,
      "loss": 0.51,
      "step": 13277
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1870086193084717,
      "learning_rate": 6.466365827894133e-07,
      "loss": 0.5375,
      "step": 13278
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0047991275787354,
      "learning_rate": 6.461320262533055e-07,
      "loss": 0.5118,
      "step": 13279
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0439016819000244,
      "learning_rate": 6.456276530434302e-07,
      "loss": 0.4938,
      "step": 13280
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.066996455192566,
      "learning_rate": 6.451234631810271e-07,
      "loss": 0.5259,
      "step": 13281
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1347441673278809,
      "learning_rate": 6.446194566873254e-07,
      "loss": 0.5433,
      "step": 13282
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1079368591308594,
      "learning_rate": 6.441156335835474e-07,
      "loss": 0.5256,
      "step": 13283
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9577980637550354,
      "learning_rate": 6.436119938909069e-07,
      "loss": 0.4615,
      "step": 13284
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1125235557556152,
      "learning_rate": 6.431085376306112e-07,
      "loss": 0.5352,
      "step": 13285
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9927465319633484,
      "learning_rate": 6.426052648238568e-07,
      "loss": 0.4886,
      "step": 13286
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0493212938308716,
      "learning_rate": 6.421021754918383e-07,
      "loss": 0.4716,
      "step": 13287
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0924726724624634,
      "learning_rate": 6.415992696557361e-07,
      "loss": 0.4976,
      "step": 13288
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0519717931747437,
      "learning_rate": 6.41096547336727e-07,
      "loss": 0.5018,
      "step": 13289
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0670619010925293,
      "learning_rate": 6.405940085559797e-07,
      "loss": 0.4831,
      "step": 13290
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0152946710586548,
      "learning_rate": 6.400916533346518e-07,
      "loss": 0.4259,
      "step": 13291
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0688623189926147,
      "learning_rate": 6.39589481693898e-07,
      "loss": 0.5016,
      "step": 13292
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0448710918426514,
      "learning_rate": 6.390874936548635e-07,
      "loss": 0.523,
      "step": 13293
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0543631315231323,
      "learning_rate": 6.385856892386826e-07,
      "loss": 0.5235,
      "step": 13294
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1219706535339355,
      "learning_rate": 6.380840684664869e-07,
      "loss": 0.4887,
      "step": 13295
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1632959842681885,
      "learning_rate": 6.375826313593963e-07,
      "loss": 0.5444,
      "step": 13296
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0708070993423462,
      "learning_rate": 6.37081377938526e-07,
      "loss": 0.5215,
      "step": 13297
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0890955924987793,
      "learning_rate": 6.365803082249822e-07,
      "loss": 0.4947,
      "step": 13298
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9379157423973083,
      "learning_rate": 6.360794222398603e-07,
      "loss": 0.4903,
      "step": 13299
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0575850009918213,
      "learning_rate": 6.35578720004254e-07,
      "loss": 0.5211,
      "step": 13300
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0986931324005127,
      "learning_rate": 6.350782015392459e-07,
      "loss": 0.5327,
      "step": 13301
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2144449949264526,
      "learning_rate": 6.345778668659097e-07,
      "loss": 0.5434,
      "step": 13302
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0637465715408325,
      "learning_rate": 6.34077716005313e-07,
      "loss": 0.5019,
      "step": 13303
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1095032691955566,
      "learning_rate": 6.335777489785161e-07,
      "loss": 0.5243,
      "step": 13304
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0209369659423828,
      "learning_rate": 6.3307796580657e-07,
      "loss": 0.5526,
      "step": 13305
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0439844131469727,
      "learning_rate": 6.325783665105206e-07,
      "loss": 0.5125,
      "step": 13306
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.094522476196289,
      "learning_rate": 6.320789511114022e-07,
      "loss": 0.4866,
      "step": 13307
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0305488109588623,
      "learning_rate": 6.315797196302432e-07,
      "loss": 0.4716,
      "step": 13308
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0392119884490967,
      "learning_rate": 6.310806720880675e-07,
      "loss": 0.5054,
      "step": 13309
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.055533528327942,
      "learning_rate": 6.305818085058852e-07,
      "loss": 0.4753,
      "step": 13310
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0672519207000732,
      "learning_rate": 6.300831289047027e-07,
      "loss": 0.5125,
      "step": 13311
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0750885009765625,
      "learning_rate": 6.295846333055184e-07,
      "loss": 0.4882,
      "step": 13312
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0400210618972778,
      "learning_rate": 6.290863217293214e-07,
      "loss": 0.5058,
      "step": 13313
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0726042985916138,
      "learning_rate": 6.285881941970951e-07,
      "loss": 0.4813,
      "step": 13314
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.069843053817749,
      "learning_rate": 6.280902507298115e-07,
      "loss": 0.5179,
      "step": 13315
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.089877724647522,
      "learning_rate": 6.275924913484377e-07,
      "loss": 0.4994,
      "step": 13316
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9835137724876404,
      "learning_rate": 6.270949160739359e-07,
      "loss": 0.4661,
      "step": 13317
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0019054412841797,
      "learning_rate": 6.265975249272544e-07,
      "loss": 0.4595,
      "step": 13318
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1164153814315796,
      "learning_rate": 6.261003179293368e-07,
      "loss": 0.4822,
      "step": 13319
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1330068111419678,
      "learning_rate": 6.256032951011188e-07,
      "loss": 0.5266,
      "step": 13320
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.100870132446289,
      "learning_rate": 6.25106456463529e-07,
      "loss": 0.5282,
      "step": 13321
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.029121994972229,
      "learning_rate": 6.246098020374869e-07,
      "loss": 0.5301,
      "step": 13322
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.088587999343872,
      "learning_rate": 6.241133318439063e-07,
      "loss": 0.5308,
      "step": 13323
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0542081594467163,
      "learning_rate": 6.236170459036894e-07,
      "loss": 0.4574,
      "step": 13324
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1169860363006592,
      "learning_rate": 6.23120944237735e-07,
      "loss": 0.5238,
      "step": 13325
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9798452854156494,
      "learning_rate": 6.226250268669309e-07,
      "loss": 0.5135,
      "step": 13326
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0330060720443726,
      "learning_rate": 6.221292938121598e-07,
      "loss": 0.5312,
      "step": 13327
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0352720022201538,
      "learning_rate": 6.216337450942955e-07,
      "loss": 0.4789,
      "step": 13328
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0385183095932007,
      "learning_rate": 6.211383807342008e-07,
      "loss": 0.4893,
      "step": 13329
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.119275689125061,
      "learning_rate": 6.206432007527368e-07,
      "loss": 0.4774,
      "step": 13330
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0935087203979492,
      "learning_rate": 6.201482051707542e-07,
      "loss": 0.5279,
      "step": 13331
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.020162582397461,
      "learning_rate": 6.196533940090932e-07,
      "loss": 0.4822,
      "step": 13332
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2218085527420044,
      "learning_rate": 6.191587672885896e-07,
      "loss": 0.4954,
      "step": 13333
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.037545084953308,
      "learning_rate": 6.186643250300706e-07,
      "loss": 0.5241,
      "step": 13334
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.091447353363037,
      "learning_rate": 6.18170067254355e-07,
      "loss": 0.4714,
      "step": 13335
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.047592043876648,
      "learning_rate": 6.176759939822557e-07,
      "loss": 0.4809,
      "step": 13336
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0553518533706665,
      "learning_rate": 6.171821052345744e-07,
      "loss": 0.4992,
      "step": 13337
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0779601335525513,
      "learning_rate": 6.166884010321072e-07,
      "loss": 0.5394,
      "step": 13338
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.006421685218811,
      "learning_rate": 6.161948813956447e-07,
      "loss": 0.5238,
      "step": 13339
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.026556372642517,
      "learning_rate": 6.157015463459648e-07,
      "loss": 0.4894,
      "step": 13340
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9980022311210632,
      "learning_rate": 6.152083959038407e-07,
      "loss": 0.446,
      "step": 13341
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0848935842514038,
      "learning_rate": 6.147154300900377e-07,
      "loss": 0.5094,
      "step": 13342
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.112890601158142,
      "learning_rate": 6.142226489253122e-07,
      "loss": 0.5098,
      "step": 13343
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0533113479614258,
      "learning_rate": 6.137300524304151e-07,
      "loss": 0.5044,
      "step": 13344
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0517538785934448,
      "learning_rate": 6.132376406260865e-07,
      "loss": 0.4979,
      "step": 13345
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0184341669082642,
      "learning_rate": 6.127454135330585e-07,
      "loss": 0.5236,
      "step": 13346
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0783300399780273,
      "learning_rate": 6.122533711720613e-07,
      "loss": 0.4872,
      "step": 13347
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1193021535873413,
      "learning_rate": 6.1176151356381e-07,
      "loss": 0.5704,
      "step": 13348
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0210407972335815,
      "learning_rate": 6.112698407290158e-07,
      "loss": 0.5419,
      "step": 13349
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0369386672973633,
      "learning_rate": 6.107783526883809e-07,
      "loss": 0.523,
      "step": 13350
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9796282052993774,
      "learning_rate": 6.102870494626006e-07,
      "loss": 0.4769,
      "step": 13351
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1247459650039673,
      "learning_rate": 6.097959310723633e-07,
      "loss": 0.5422,
      "step": 13352
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.99430251121521,
      "learning_rate": 6.093049975383458e-07,
      "loss": 0.5098,
      "step": 13353
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0856844186782837,
      "learning_rate": 6.08814248881221e-07,
      "loss": 0.4806,
      "step": 13354
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0806077718734741,
      "learning_rate": 6.083236851216517e-07,
      "loss": 0.4805,
      "step": 13355
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1206717491149902,
      "learning_rate": 6.078333062802949e-07,
      "loss": 0.5672,
      "step": 13356
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1182076930999756,
      "learning_rate": 6.073431123777984e-07,
      "loss": 0.5323,
      "step": 13357
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1024640798568726,
      "learning_rate": 6.068531034348035e-07,
      "loss": 0.5219,
      "step": 13358
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0542242527008057,
      "learning_rate": 6.063632794719399e-07,
      "loss": 0.4919,
      "step": 13359
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.079557180404663,
      "learning_rate": 6.058736405098359e-07,
      "loss": 0.5392,
      "step": 13360
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1475123167037964,
      "learning_rate": 6.053841865691063e-07,
      "loss": 0.5783,
      "step": 13361
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0788050889968872,
      "learning_rate": 6.048949176703606e-07,
      "loss": 0.4897,
      "step": 13362
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0291560888290405,
      "learning_rate": 6.044058338342002e-07,
      "loss": 0.4777,
      "step": 13363
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0231040716171265,
      "learning_rate": 6.039169350812191e-07,
      "loss": 0.5104,
      "step": 13364
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0467113256454468,
      "learning_rate": 6.034282214320031e-07,
      "loss": 0.5164,
      "step": 13365
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2185094356536865,
      "learning_rate": 6.029396929071313e-07,
      "loss": 0.5592,
      "step": 13366
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0959972143173218,
      "learning_rate": 6.024513495271705e-07,
      "loss": 0.5077,
      "step": 13367
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0685462951660156,
      "learning_rate": 6.019631913126877e-07,
      "loss": 0.4392,
      "step": 13368
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1245286464691162,
      "learning_rate": 6.014752182842343e-07,
      "loss": 0.488,
      "step": 13369
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0407520532608032,
      "learning_rate": 6.009874304623576e-07,
      "loss": 0.4951,
      "step": 13370
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9670149087905884,
      "learning_rate": 6.004998278675988e-07,
      "loss": 0.4493,
      "step": 13371
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0624332427978516,
      "learning_rate": 6.000124105204847e-07,
      "loss": 0.4983,
      "step": 13372
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.092924952507019,
      "learning_rate": 5.995251784415435e-07,
      "loss": 0.5028,
      "step": 13373
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0477136373519897,
      "learning_rate": 5.990381316512894e-07,
      "loss": 0.5134,
      "step": 13374
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.074385404586792,
      "learning_rate": 5.985512701702284e-07,
      "loss": 0.5002,
      "step": 13375
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0472429990768433,
      "learning_rate": 5.980645940188623e-07,
      "loss": 0.4914,
      "step": 13376
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0674313306808472,
      "learning_rate": 5.975781032176831e-07,
      "loss": 0.4902,
      "step": 13377
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0908169746398926,
      "learning_rate": 5.970917977871749e-07,
      "loss": 0.513,
      "step": 13378
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0883985757827759,
      "learning_rate": 5.966056777478152e-07,
      "loss": 0.4675,
      "step": 13379
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1221798658370972,
      "learning_rate": 5.961197431200705e-07,
      "loss": 0.5164,
      "step": 13380
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1040724515914917,
      "learning_rate": 5.956339939244044e-07,
      "loss": 0.4699,
      "step": 13381
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0564048290252686,
      "learning_rate": 5.951484301812699e-07,
      "loss": 0.5189,
      "step": 13382
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.131378412246704,
      "learning_rate": 5.946630519111107e-07,
      "loss": 0.511,
      "step": 13383
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0738738775253296,
      "learning_rate": 5.941778591343656e-07,
      "loss": 0.5174,
      "step": 13384
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0545551776885986,
      "learning_rate": 5.936928518714641e-07,
      "loss": 0.4968,
      "step": 13385
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0524842739105225,
      "learning_rate": 5.932080301428278e-07,
      "loss": 0.5021,
      "step": 13386
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0227655172348022,
      "learning_rate": 5.927233939688714e-07,
      "loss": 0.5357,
      "step": 13387
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.021254301071167,
      "learning_rate": 5.922389433700021e-07,
      "loss": 0.4929,
      "step": 13388
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0238847732543945,
      "learning_rate": 5.917546783666156e-07,
      "loss": 0.4754,
      "step": 13389
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.070908546447754,
      "learning_rate": 5.912705989791062e-07,
      "loss": 0.5137,
      "step": 13390
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0081439018249512,
      "learning_rate": 5.907867052278543e-07,
      "loss": 0.4707,
      "step": 13391
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.061011791229248,
      "learning_rate": 5.903029971332353e-07,
      "loss": 0.536,
      "step": 13392
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0461562871932983,
      "learning_rate": 5.898194747156171e-07,
      "loss": 0.5196,
      "step": 13393
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0973914861679077,
      "learning_rate": 5.893361379953588e-07,
      "loss": 0.5192,
      "step": 13394
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1432956457138062,
      "learning_rate": 5.888529869928122e-07,
      "loss": 0.5257,
      "step": 13395
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1039844751358032,
      "learning_rate": 5.883700217283223e-07,
      "loss": 0.468,
      "step": 13396
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1045359373092651,
      "learning_rate": 5.878872422222215e-07,
      "loss": 0.4658,
      "step": 13397
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1638327836990356,
      "learning_rate": 5.874046484948426e-07,
      "loss": 0.5342,
      "step": 13398
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0851562023162842,
      "learning_rate": 5.869222405665026e-07,
      "loss": 0.4768,
      "step": 13399
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1951812505722046,
      "learning_rate": 5.864400184575153e-07,
      "loss": 0.5113,
      "step": 13400
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1096984148025513,
      "learning_rate": 5.859579821881855e-07,
      "loss": 0.5218,
      "step": 13401
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0537545680999756,
      "learning_rate": 5.854761317788082e-07,
      "loss": 0.4826,
      "step": 13402
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0303213596343994,
      "learning_rate": 5.849944672496749e-07,
      "loss": 0.5577,
      "step": 13403
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1699107885360718,
      "learning_rate": 5.845129886210671e-07,
      "loss": 0.546,
      "step": 13404
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0213823318481445,
      "learning_rate": 5.840316959132558e-07,
      "loss": 0.4528,
      "step": 13405
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.068701982498169,
      "learning_rate": 5.835505891465076e-07,
      "loss": 0.5116,
      "step": 13406
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0392073392868042,
      "learning_rate": 5.830696683410802e-07,
      "loss": 0.4792,
      "step": 13407
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9517161250114441,
      "learning_rate": 5.825889335172241e-07,
      "loss": 0.4691,
      "step": 13408
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0567563772201538,
      "learning_rate": 5.821083846951819e-07,
      "loss": 0.5436,
      "step": 13409
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.060482144355774,
      "learning_rate": 5.816280218951847e-07,
      "loss": 0.4599,
      "step": 13410
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0229899883270264,
      "learning_rate": 5.811478451374625e-07,
      "loss": 0.4913,
      "step": 13411
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1489890813827515,
      "learning_rate": 5.806678544422334e-07,
      "loss": 0.5356,
      "step": 13412
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0392504930496216,
      "learning_rate": 5.801880498297057e-07,
      "loss": 0.4936,
      "step": 13413
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0875258445739746,
      "learning_rate": 5.797084313200846e-07,
      "loss": 0.4962,
      "step": 13414
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0216097831726074,
      "learning_rate": 5.792289989335637e-07,
      "loss": 0.5191,
      "step": 13415
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1515929698944092,
      "learning_rate": 5.787497526903313e-07,
      "loss": 0.5328,
      "step": 13416
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0613532066345215,
      "learning_rate": 5.782706926105674e-07,
      "loss": 0.5331,
      "step": 13417
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.016359806060791,
      "learning_rate": 5.777918187144416e-07,
      "loss": 0.5008,
      "step": 13418
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1669195890426636,
      "learning_rate": 5.773131310221169e-07,
      "loss": 0.5416,
      "step": 13419
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0914517641067505,
      "learning_rate": 5.768346295537536e-07,
      "loss": 0.5339,
      "step": 13420
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0393308401107788,
      "learning_rate": 5.76356314329496e-07,
      "loss": 0.5353,
      "step": 13421
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.111310362815857,
      "learning_rate": 5.758781853694845e-07,
      "loss": 0.5443,
      "step": 13422
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0661511421203613,
      "learning_rate": 5.754002426938532e-07,
      "loss": 0.5195,
      "step": 13423
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.052367091178894,
      "learning_rate": 5.749224863227249e-07,
      "loss": 0.4722,
      "step": 13424
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0971262454986572,
      "learning_rate": 5.744449162762183e-07,
      "loss": 0.4774,
      "step": 13425
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.053396463394165,
      "learning_rate": 5.739675325744398e-07,
      "loss": 0.51,
      "step": 13426
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.101646900177002,
      "learning_rate": 5.734903352374904e-07,
      "loss": 0.4877,
      "step": 13427
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9976493716239929,
      "learning_rate": 5.730133242854663e-07,
      "loss": 0.4812,
      "step": 13428
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.06601083278656,
      "learning_rate": 5.725364997384498e-07,
      "loss": 0.5059,
      "step": 13429
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0498607158660889,
      "learning_rate": 5.720598616165196e-07,
      "loss": 0.5217,
      "step": 13430
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.089664101600647,
      "learning_rate": 5.715834099397455e-07,
      "loss": 0.4993,
      "step": 13431
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9954808354377747,
      "learning_rate": 5.711071447281868e-07,
      "loss": 0.4651,
      "step": 13432
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9758763909339905,
      "learning_rate": 5.70631066001901e-07,
      "loss": 0.4788,
      "step": 13433
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0360801219940186,
      "learning_rate": 5.701551737809319e-07,
      "loss": 0.4746,
      "step": 13434
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1086175441741943,
      "learning_rate": 5.696794680853179e-07,
      "loss": 0.4847,
      "step": 13435
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0172804594039917,
      "learning_rate": 5.692039489350892e-07,
      "loss": 0.4974,
      "step": 13436
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1112823486328125,
      "learning_rate": 5.687286163502687e-07,
      "loss": 0.4819,
      "step": 13437
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.057182788848877,
      "learning_rate": 5.682534703508713e-07,
      "loss": 0.4543,
      "step": 13438
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0339446067810059,
      "learning_rate": 5.67778510956904e-07,
      "loss": 0.471,
      "step": 13439
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9820137619972229,
      "learning_rate": 5.673037381883634e-07,
      "loss": 0.4218,
      "step": 13440
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.03311288356781,
      "learning_rate": 5.668291520652436e-07,
      "loss": 0.4846,
      "step": 13441
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9906759262084961,
      "learning_rate": 5.663547526075258e-07,
      "loss": 0.4565,
      "step": 13442
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0246329307556152,
      "learning_rate": 5.658805398351858e-07,
      "loss": 0.5291,
      "step": 13443
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9982864260673523,
      "learning_rate": 5.654065137681907e-07,
      "loss": 0.4907,
      "step": 13444
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.123137354850769,
      "learning_rate": 5.64932674426501e-07,
      "loss": 0.5616,
      "step": 13445
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0283350944519043,
      "learning_rate": 5.644590218300672e-07,
      "loss": 0.4707,
      "step": 13446
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0369606018066406,
      "learning_rate": 5.639855559988356e-07,
      "loss": 0.4627,
      "step": 13447
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.040816068649292,
      "learning_rate": 5.63512276952739e-07,
      "loss": 0.5324,
      "step": 13448
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.088789701461792,
      "learning_rate": 5.630391847117073e-07,
      "loss": 0.5435,
      "step": 13449
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0620760917663574,
      "learning_rate": 5.625662792956604e-07,
      "loss": 0.5114,
      "step": 13450
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0261839628219604,
      "learning_rate": 5.620935607245109e-07,
      "loss": 0.5098,
      "step": 13451
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9988752007484436,
      "learning_rate": 5.616210290181628e-07,
      "loss": 0.459,
      "step": 13452
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1079870462417603,
      "learning_rate": 5.611486841965136e-07,
      "loss": 0.48,
      "step": 13453
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0706554651260376,
      "learning_rate": 5.606765262794512e-07,
      "loss": 0.5064,
      "step": 13454
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.140230655670166,
      "learning_rate": 5.602045552868585e-07,
      "loss": 0.5137,
      "step": 13455
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.059227466583252,
      "learning_rate": 5.597327712386058e-07,
      "loss": 0.536,
      "step": 13456
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9997138977050781,
      "learning_rate": 5.592611741545594e-07,
      "loss": 0.446,
      "step": 13457
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0232324600219727,
      "learning_rate": 5.58789764054577e-07,
      "loss": 0.507,
      "step": 13458
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0606647729873657,
      "learning_rate": 5.583185409585079e-07,
      "loss": 0.481,
      "step": 13459
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1186909675598145,
      "learning_rate": 5.578475048861931e-07,
      "loss": 0.4988,
      "step": 13460
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9839795827865601,
      "learning_rate": 5.573766558574684e-07,
      "loss": 0.4766,
      "step": 13461
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9404541254043579,
      "learning_rate": 5.569059938921551e-07,
      "loss": 0.488,
      "step": 13462
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9731743931770325,
      "learning_rate": 5.564355190100768e-07,
      "loss": 0.4615,
      "step": 13463
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9581971764564514,
      "learning_rate": 5.559652312310393e-07,
      "loss": 0.4714,
      "step": 13464
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0430355072021484,
      "learning_rate": 5.554951305748462e-07,
      "loss": 0.4696,
      "step": 13465
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0668957233428955,
      "learning_rate": 5.550252170612924e-07,
      "loss": 0.4743,
      "step": 13466
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0566219091415405,
      "learning_rate": 5.545554907101636e-07,
      "loss": 0.4965,
      "step": 13467
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9783183336257935,
      "learning_rate": 5.540859515412378e-07,
      "loss": 0.4866,
      "step": 13468
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9738695025444031,
      "learning_rate": 5.536165995742882e-07,
      "loss": 0.4563,
      "step": 13469
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.023001790046692,
      "learning_rate": 5.531474348290733e-07,
      "loss": 0.4656,
      "step": 13470
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0590389966964722,
      "learning_rate": 5.526784573253525e-07,
      "loss": 0.51,
      "step": 13471
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0875754356384277,
      "learning_rate": 5.522096670828703e-07,
      "loss": 0.5145,
      "step": 13472
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.018993854522705,
      "learning_rate": 5.517410641213656e-07,
      "loss": 0.5021,
      "step": 13473
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9939917325973511,
      "learning_rate": 5.512726484605707e-07,
      "loss": 0.484,
      "step": 13474
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0694676637649536,
      "learning_rate": 5.508044201202084e-07,
      "loss": 0.4823,
      "step": 13475
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0037014484405518,
      "learning_rate": 5.503363791199945e-07,
      "loss": 0.4544,
      "step": 13476
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0620406866073608,
      "learning_rate": 5.49868525479637e-07,
      "loss": 0.5536,
      "step": 13477
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.023963451385498,
      "learning_rate": 5.494008592188344e-07,
      "loss": 0.4545,
      "step": 13478
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0016241073608398,
      "learning_rate": 5.489333803572788e-07,
      "loss": 0.4767,
      "step": 13479
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9408761858940125,
      "learning_rate": 5.484660889146548e-07,
      "loss": 0.453,
      "step": 13480
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.110805869102478,
      "learning_rate": 5.479989849106381e-07,
      "loss": 0.5043,
      "step": 13481
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0312732458114624,
      "learning_rate": 5.475320683648977e-07,
      "loss": 0.491,
      "step": 13482
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9720163941383362,
      "learning_rate": 5.470653392970904e-07,
      "loss": 0.4894,
      "step": 13483
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0498781204223633,
      "learning_rate": 5.465987977268727e-07,
      "loss": 0.4994,
      "step": 13484
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0908483266830444,
      "learning_rate": 5.46132443673888e-07,
      "loss": 0.5265,
      "step": 13485
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0789039134979248,
      "learning_rate": 5.456662771577714e-07,
      "loss": 0.4853,
      "step": 13486
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1271196603775024,
      "learning_rate": 5.452002981981519e-07,
      "loss": 0.5378,
      "step": 13487
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0779871940612793,
      "learning_rate": 5.447345068146515e-07,
      "loss": 0.5562,
      "step": 13488
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0521538257598877,
      "learning_rate": 5.442689030268816e-07,
      "loss": 0.4833,
      "step": 13489
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.155379295349121,
      "learning_rate": 5.438034868544495e-07,
      "loss": 0.5332,
      "step": 13490
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0768070220947266,
      "learning_rate": 5.433382583169478e-07,
      "loss": 0.5016,
      "step": 13491
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1185115575790405,
      "learning_rate": 5.428732174339702e-07,
      "loss": 0.5098,
      "step": 13492
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0764923095703125,
      "learning_rate": 5.424083642250966e-07,
      "loss": 0.4944,
      "step": 13493
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1410459280014038,
      "learning_rate": 5.419436987098991e-07,
      "loss": 0.546,
      "step": 13494
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.112058162689209,
      "learning_rate": 5.414792209079445e-07,
      "loss": 0.5081,
      "step": 13495
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0271916389465332,
      "learning_rate": 5.410149308387891e-07,
      "loss": 0.5599,
      "step": 13496
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0328270196914673,
      "learning_rate": 5.405508285219835e-07,
      "loss": 0.5023,
      "step": 13497
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9662725925445557,
      "learning_rate": 5.400869139770704e-07,
      "loss": 0.438,
      "step": 13498
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0837187767028809,
      "learning_rate": 5.396231872235819e-07,
      "loss": 0.4805,
      "step": 13499
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.087839961051941,
      "learning_rate": 5.391596482810424e-07,
      "loss": 0.5013,
      "step": 13500
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0441499948501587,
      "learning_rate": 5.386962971689746e-07,
      "loss": 0.4961,
      "step": 13501
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0059573650360107,
      "learning_rate": 5.382331339068853e-07,
      "loss": 0.506,
      "step": 13502
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.980630099773407,
      "learning_rate": 5.377701585142769e-07,
      "loss": 0.4787,
      "step": 13503
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1318646669387817,
      "learning_rate": 5.373073710106441e-07,
      "loss": 0.5448,
      "step": 13504
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1532589197158813,
      "learning_rate": 5.368447714154734e-07,
      "loss": 0.5069,
      "step": 13505
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0172760486602783,
      "learning_rate": 5.363823597482443e-07,
      "loss": 0.5132,
      "step": 13506
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0407923460006714,
      "learning_rate": 5.359201360284255e-07,
      "loss": 0.488,
      "step": 13507
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1686497926712036,
      "learning_rate": 5.354581002754799e-07,
      "loss": 0.5275,
      "step": 13508
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9421904683113098,
      "learning_rate": 5.349962525088631e-07,
      "loss": 0.4569,
      "step": 13509
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0419373512268066,
      "learning_rate": 5.345345927480211e-07,
      "loss": 0.4937,
      "step": 13510
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0212336778640747,
      "learning_rate": 5.340731210123934e-07,
      "loss": 0.4955,
      "step": 13511
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0964690446853638,
      "learning_rate": 5.336118373214116e-07,
      "loss": 0.5284,
      "step": 13512
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.086272954940796,
      "learning_rate": 5.331507416944965e-07,
      "loss": 0.5009,
      "step": 13513
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.071381688117981,
      "learning_rate": 5.326898341510655e-07,
      "loss": 0.5464,
      "step": 13514
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0455056428909302,
      "learning_rate": 5.322291147105246e-07,
      "loss": 0.4958,
      "step": 13515
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.069616675376892,
      "learning_rate": 5.317685833922737e-07,
      "loss": 0.5581,
      "step": 13516
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0819993019104004,
      "learning_rate": 5.313082402157039e-07,
      "loss": 0.4991,
      "step": 13517
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9846544861793518,
      "learning_rate": 5.308480852001979e-07,
      "loss": 0.4838,
      "step": 13518
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.094940423965454,
      "learning_rate": 5.303881183651327e-07,
      "loss": 0.4385,
      "step": 13519
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0644207000732422,
      "learning_rate": 5.29928339729876e-07,
      "loss": 0.5435,
      "step": 13520
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0141444206237793,
      "learning_rate": 5.294687493137845e-07,
      "loss": 0.5139,
      "step": 13521
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.046487808227539,
      "learning_rate": 5.290093471362145e-07,
      "loss": 0.5192,
      "step": 13522
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9801287055015564,
      "learning_rate": 5.28550133216506e-07,
      "loss": 0.5298,
      "step": 13523
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0636112689971924,
      "learning_rate": 5.28091107573997e-07,
      "loss": 0.4557,
      "step": 13524
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9705425500869751,
      "learning_rate": 5.27632270228014e-07,
      "loss": 0.4822,
      "step": 13525
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1575100421905518,
      "learning_rate": 5.271736211978784e-07,
      "loss": 0.483,
      "step": 13526
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.126613736152649,
      "learning_rate": 5.267151605029014e-07,
      "loss": 0.4884,
      "step": 13527
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0410032272338867,
      "learning_rate": 5.262568881623892e-07,
      "loss": 0.4944,
      "step": 13528
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0787334442138672,
      "learning_rate": 5.257988041956347e-07,
      "loss": 0.4754,
      "step": 13529
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1417659521102905,
      "learning_rate": 5.253409086219274e-07,
      "loss": 0.5649,
      "step": 13530
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.06132972240448,
      "learning_rate": 5.248832014605503e-07,
      "loss": 0.4725,
      "step": 13531
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1390949487686157,
      "learning_rate": 5.244256827307726e-07,
      "loss": 0.5635,
      "step": 13532
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9856281876564026,
      "learning_rate": 5.239683524518596e-07,
      "loss": 0.4919,
      "step": 13533
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1564993858337402,
      "learning_rate": 5.23511210643069e-07,
      "loss": 0.5237,
      "step": 13534
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9941248297691345,
      "learning_rate": 5.230542573236485e-07,
      "loss": 0.4986,
      "step": 13535
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0197328329086304,
      "learning_rate": 5.225974925128402e-07,
      "loss": 0.4711,
      "step": 13536
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.131434679031372,
      "learning_rate": 5.221409162298741e-07,
      "loss": 0.5358,
      "step": 13537
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0401612520217896,
      "learning_rate": 5.216845284939764e-07,
      "loss": 0.5269,
      "step": 13538
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0301513671875,
      "learning_rate": 5.212283293243658e-07,
      "loss": 0.48,
      "step": 13539
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0342973470687866,
      "learning_rate": 5.207723187402491e-07,
      "loss": 0.4466,
      "step": 13540
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1259351968765259,
      "learning_rate": 5.203164967608282e-07,
      "loss": 0.5571,
      "step": 13541
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.121213436126709,
      "learning_rate": 5.198608634052965e-07,
      "loss": 0.572,
      "step": 13542
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0400069952011108,
      "learning_rate": 5.194054186928365e-07,
      "loss": 0.4837,
      "step": 13543
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0414228439331055,
      "learning_rate": 5.189501626426297e-07,
      "loss": 0.4879,
      "step": 13544
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0761221647262573,
      "learning_rate": 5.184950952738421e-07,
      "loss": 0.5487,
      "step": 13545
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.091073989868164,
      "learning_rate": 5.180402166056359e-07,
      "loss": 0.4429,
      "step": 13546
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.127923607826233,
      "learning_rate": 5.175855266571644e-07,
      "loss": 0.5364,
      "step": 13547
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1597503423690796,
      "learning_rate": 5.171310254475737e-07,
      "loss": 0.514,
      "step": 13548
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.162801742553711,
      "learning_rate": 5.166767129960004e-07,
      "loss": 0.5011,
      "step": 13549
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0138014554977417,
      "learning_rate": 5.162225893215755e-07,
      "loss": 0.4867,
      "step": 13550
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0033013820648193,
      "learning_rate": 5.157686544434176e-07,
      "loss": 0.4832,
      "step": 13551
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0404198169708252,
      "learning_rate": 5.153149083806436e-07,
      "loss": 0.4695,
      "step": 13552
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.141335368156433,
      "learning_rate": 5.14861351152357e-07,
      "loss": 0.5374,
      "step": 13553
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.062104344367981,
      "learning_rate": 5.144079827776566e-07,
      "loss": 0.4696,
      "step": 13554
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9700568318367004,
      "learning_rate": 5.139548032756325e-07,
      "loss": 0.4859,
      "step": 13555
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.995707631111145,
      "learning_rate": 5.13501812665364e-07,
      "loss": 0.4342,
      "step": 13556
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.113817811012268,
      "learning_rate": 5.130490109659275e-07,
      "loss": 0.4639,
      "step": 13557
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0634162425994873,
      "learning_rate": 5.125963981963894e-07,
      "loss": 0.513,
      "step": 13558
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9440078139305115,
      "learning_rate": 5.12143974375805e-07,
      "loss": 0.4778,
      "step": 13559
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0164488554000854,
      "learning_rate": 5.116917395232262e-07,
      "loss": 0.5417,
      "step": 13560
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.050866723060608,
      "learning_rate": 5.112396936576947e-07,
      "loss": 0.4614,
      "step": 13561
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.016036868095398,
      "learning_rate": 5.107878367982438e-07,
      "loss": 0.4636,
      "step": 13562
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.132187008857727,
      "learning_rate": 5.103361689639019e-07,
      "loss": 0.5277,
      "step": 13563
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0902912616729736,
      "learning_rate": 5.098846901736832e-07,
      "loss": 0.5161,
      "step": 13564
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0292311906814575,
      "learning_rate": 5.094334004466012e-07,
      "loss": 0.5256,
      "step": 13565
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0934474468231201,
      "learning_rate": 5.089822998016586e-07,
      "loss": 0.4878,
      "step": 13566
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1122615337371826,
      "learning_rate": 5.085313882578469e-07,
      "loss": 0.4703,
      "step": 13567
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.130413293838501,
      "learning_rate": 5.080806658341536e-07,
      "loss": 0.5113,
      "step": 13568
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1992076635360718,
      "learning_rate": 5.076301325495575e-07,
      "loss": 0.5321,
      "step": 13569
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9438257813453674,
      "learning_rate": 5.071797884230284e-07,
      "loss": 0.4599,
      "step": 13570
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0700342655181885,
      "learning_rate": 5.067296334735306e-07,
      "loss": 0.5489,
      "step": 13571
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1255141496658325,
      "learning_rate": 5.062796677200154e-07,
      "loss": 0.5352,
      "step": 13572
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0605268478393555,
      "learning_rate": 5.058298911814302e-07,
      "loss": 0.5164,
      "step": 13573
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0894030332565308,
      "learning_rate": 5.053803038767158e-07,
      "loss": 0.5137,
      "step": 13574
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0053369998931885,
      "learning_rate": 5.049309058248004e-07,
      "loss": 0.4574,
      "step": 13575
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0016471147537231,
      "learning_rate": 5.044816970446076e-07,
      "loss": 0.4495,
      "step": 13576
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.084174394607544,
      "learning_rate": 5.040326775550514e-07,
      "loss": 0.5058,
      "step": 13577
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1791083812713623,
      "learning_rate": 5.035838473750393e-07,
      "loss": 0.5271,
      "step": 13578
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.030188798904419,
      "learning_rate": 5.031352065234702e-07,
      "loss": 0.4744,
      "step": 13579
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.043033480644226,
      "learning_rate": 5.026867550192327e-07,
      "loss": 0.4716,
      "step": 13580
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.157036304473877,
      "learning_rate": 5.022384928812107e-07,
      "loss": 0.5134,
      "step": 13581
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9832121133804321,
      "learning_rate": 5.017904201282808e-07,
      "loss": 0.454,
      "step": 13582
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0457446575164795,
      "learning_rate": 5.013425367793074e-07,
      "loss": 0.4794,
      "step": 13583
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0089260339736938,
      "learning_rate": 5.008948428531496e-07,
      "loss": 0.4778,
      "step": 13584
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0331329107284546,
      "learning_rate": 5.004473383686592e-07,
      "loss": 0.515,
      "step": 13585
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1153910160064697,
      "learning_rate": 5.000000233446783e-07,
      "loss": 0.4699,
      "step": 13586
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0856069326400757,
      "learning_rate": 4.99552897800043e-07,
      "loss": 0.4885,
      "step": 13587
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0789910554885864,
      "learning_rate": 4.991059617535781e-07,
      "loss": 0.4944,
      "step": 13588
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.072582483291626,
      "learning_rate": 4.986592152241043e-07,
      "loss": 0.5377,
      "step": 13589
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.044414758682251,
      "learning_rate": 4.982126582304314e-07,
      "loss": 0.4907,
      "step": 13590
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.141349196434021,
      "learning_rate": 4.977662907913633e-07,
      "loss": 0.5417,
      "step": 13591
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0667126178741455,
      "learning_rate": 4.973201129256943e-07,
      "loss": 0.4893,
      "step": 13592
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.069313406944275,
      "learning_rate": 4.968741246522129e-07,
      "loss": 0.526,
      "step": 13593
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.127071738243103,
      "learning_rate": 4.964283259896945e-07,
      "loss": 0.496,
      "step": 13594
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0764715671539307,
      "learning_rate": 4.959827169569136e-07,
      "loss": 0.4625,
      "step": 13595
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0825448036193848,
      "learning_rate": 4.955372975726336e-07,
      "loss": 0.464,
      "step": 13596
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0636563301086426,
      "learning_rate": 4.950920678556065e-07,
      "loss": 0.5054,
      "step": 13597
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9773215651512146,
      "learning_rate": 4.946470278245813e-07,
      "loss": 0.479,
      "step": 13598
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9716424942016602,
      "learning_rate": 4.942021774982969e-07,
      "loss": 0.5029,
      "step": 13599
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0770186185836792,
      "learning_rate": 4.937575168954845e-07,
      "loss": 0.4665,
      "step": 13600
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0549659729003906,
      "learning_rate": 4.933130460348673e-07,
      "loss": 0.4864,
      "step": 13601
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0820561647415161,
      "learning_rate": 4.928687649351594e-07,
      "loss": 0.4769,
      "step": 13602
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0644725561141968,
      "learning_rate": 4.924246736150679e-07,
      "loss": 0.4813,
      "step": 13603
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1527262926101685,
      "learning_rate": 4.919807720932946e-07,
      "loss": 0.5019,
      "step": 13604
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9755449295043945,
      "learning_rate": 4.915370603885272e-07,
      "loss": 0.4708,
      "step": 13605
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0473021268844604,
      "learning_rate": 4.91093538519451e-07,
      "loss": 0.4953,
      "step": 13606
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0419485569000244,
      "learning_rate": 4.906502065047403e-07,
      "loss": 0.5103,
      "step": 13607
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9475518465042114,
      "learning_rate": 4.902070643630624e-07,
      "loss": 0.4805,
      "step": 13608
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0797600746154785,
      "learning_rate": 4.89764112113078e-07,
      "loss": 0.5072,
      "step": 13609
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0618468523025513,
      "learning_rate": 4.893213497734356e-07,
      "loss": 0.4919,
      "step": 13610
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1053194999694824,
      "learning_rate": 4.888787773627785e-07,
      "loss": 0.5046,
      "step": 13611
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0452958345413208,
      "learning_rate": 4.884363948997455e-07,
      "loss": 0.4605,
      "step": 13612
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0648913383483887,
      "learning_rate": 4.879942024029599e-07,
      "loss": 0.4763,
      "step": 13613
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0711185932159424,
      "learning_rate": 4.875521998910426e-07,
      "loss": 0.4512,
      "step": 13614
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.075517177581787,
      "learning_rate": 4.871103873826044e-07,
      "loss": 0.508,
      "step": 13615
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9884278178215027,
      "learning_rate": 4.866687648962487e-07,
      "loss": 0.5261,
      "step": 13616
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0627617835998535,
      "learning_rate": 4.862273324505712e-07,
      "loss": 0.4483,
      "step": 13617
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9698841571807861,
      "learning_rate": 4.857860900641576e-07,
      "loss": 0.4589,
      "step": 13618
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0453863143920898,
      "learning_rate": 4.853450377555879e-07,
      "loss": 0.4927,
      "step": 13619
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0251530408859253,
      "learning_rate": 4.849041755434336e-07,
      "loss": 0.4854,
      "step": 13620
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.01449453830719,
      "learning_rate": 4.844635034462574e-07,
      "loss": 0.4369,
      "step": 13621
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0162572860717773,
      "learning_rate": 4.840230214826147e-07,
      "loss": 0.5273,
      "step": 13622
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.072426438331604,
      "learning_rate": 4.835827296710537e-07,
      "loss": 0.5463,
      "step": 13623
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1377872228622437,
      "learning_rate": 4.831426280301105e-07,
      "loss": 0.5315,
      "step": 13624
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9945191144943237,
      "learning_rate": 4.8270271657832e-07,
      "loss": 0.4296,
      "step": 13625
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0929838418960571,
      "learning_rate": 4.822629953342028e-07,
      "loss": 0.4947,
      "step": 13626
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0578333139419556,
      "learning_rate": 4.81823464316275e-07,
      "loss": 0.4908,
      "step": 13627
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0581110715866089,
      "learning_rate": 4.813841235430433e-07,
      "loss": 0.4542,
      "step": 13628
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0441462993621826,
      "learning_rate": 4.809449730330068e-07,
      "loss": 0.4749,
      "step": 13629
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0819425582885742,
      "learning_rate": 4.805060128046574e-07,
      "loss": 0.5002,
      "step": 13630
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9921013712882996,
      "learning_rate": 4.80067242876478e-07,
      "loss": 0.5028,
      "step": 13631
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1062136888504028,
      "learning_rate": 4.796286632669417e-07,
      "loss": 0.5031,
      "step": 13632
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0803297758102417,
      "learning_rate": 4.791902739945187e-07,
      "loss": 0.4738,
      "step": 13633
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0163581371307373,
      "learning_rate": 4.787520750776658e-07,
      "loss": 0.4973,
      "step": 13634
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0785454511642456,
      "learning_rate": 4.783140665348352e-07,
      "loss": 0.4376,
      "step": 13635
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1214227676391602,
      "learning_rate": 4.778762483844701e-07,
      "loss": 0.493,
      "step": 13636
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9712678790092468,
      "learning_rate": 4.774386206450027e-07,
      "loss": 0.4666,
      "step": 13637
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0110604763031006,
      "learning_rate": 4.770011833348631e-07,
      "loss": 0.4826,
      "step": 13638
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0145424604415894,
      "learning_rate": 4.7656393647247054e-07,
      "loss": 0.4737,
      "step": 13639
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9978539943695068,
      "learning_rate": 4.7612688007623363e-07,
      "loss": 0.4696,
      "step": 13640
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9833884239196777,
      "learning_rate": 4.756900141645565e-07,
      "loss": 0.5214,
      "step": 13641
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0501322746276855,
      "learning_rate": 4.752533387558339e-07,
      "loss": 0.5091,
      "step": 13642
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0006647109985352,
      "learning_rate": 4.748168538684528e-07,
      "loss": 0.4968,
      "step": 13643
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2387149333953857,
      "learning_rate": 4.7438055952079287e-07,
      "loss": 0.5373,
      "step": 13644
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0665277242660522,
      "learning_rate": 4.739444557312223e-07,
      "loss": 0.485,
      "step": 13645
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0650554895401,
      "learning_rate": 4.735085425181063e-07,
      "loss": 0.4833,
      "step": 13646
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.059370756149292,
      "learning_rate": 4.730728198998008e-07,
      "loss": 0.5535,
      "step": 13647
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.110365629196167,
      "learning_rate": 4.726372878946489e-07,
      "loss": 0.4718,
      "step": 13648
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.028216004371643,
      "learning_rate": 4.7220194652099204e-07,
      "loss": 0.5179,
      "step": 13649
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0740399360656738,
      "learning_rate": 4.7176679579716e-07,
      "loss": 0.5044,
      "step": 13650
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1303333044052124,
      "learning_rate": 4.7133183574147534e-07,
      "loss": 0.554,
      "step": 13651
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0639551877975464,
      "learning_rate": 4.7089706637225283e-07,
      "loss": 0.528,
      "step": 13652
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0426877737045288,
      "learning_rate": 4.7046248770780065e-07,
      "loss": 0.5154,
      "step": 13653
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9748163819313049,
      "learning_rate": 4.7002809976641417e-07,
      "loss": 0.466,
      "step": 13654
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1808016300201416,
      "learning_rate": 4.6959390256638703e-07,
      "loss": 0.5148,
      "step": 13655
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0725035667419434,
      "learning_rate": 4.691598961260002e-07,
      "loss": 0.508,
      "step": 13656
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1180516481399536,
      "learning_rate": 4.68726080463528e-07,
      "loss": 0.5635,
      "step": 13657
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.062372088432312,
      "learning_rate": 4.682924555972379e-07,
      "loss": 0.4866,
      "step": 13658
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0931804180145264,
      "learning_rate": 4.6785902154538763e-07,
      "loss": 0.54,
      "step": 13659
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1557106971740723,
      "learning_rate": 4.674257783262276e-07,
      "loss": 0.5277,
      "step": 13660
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.065765380859375,
      "learning_rate": 4.669927259580015e-07,
      "loss": 0.5072,
      "step": 13661
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9811139702796936,
      "learning_rate": 4.665598644589409e-07,
      "loss": 0.4384,
      "step": 13662
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0448228120803833,
      "learning_rate": 4.6612719384727556e-07,
      "loss": 0.5043,
      "step": 13663
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1161221265792847,
      "learning_rate": 4.656947141412205e-07,
      "loss": 0.5172,
      "step": 13664
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0317866802215576,
      "learning_rate": 4.652624253589877e-07,
      "loss": 0.5474,
      "step": 13665
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0991182327270508,
      "learning_rate": 4.6483032751877987e-07,
      "loss": 0.5154,
      "step": 13666
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1278705596923828,
      "learning_rate": 4.6439842063878803e-07,
      "loss": 0.4723,
      "step": 13667
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0951365232467651,
      "learning_rate": 4.639667047372015e-07,
      "loss": 0.5117,
      "step": 13668
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9740217924118042,
      "learning_rate": 4.6353517983219856e-07,
      "loss": 0.4419,
      "step": 13669
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.073346734046936,
      "learning_rate": 4.631038459419468e-07,
      "loss": 0.4737,
      "step": 13670
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0972206592559814,
      "learning_rate": 4.6267270308460955e-07,
      "loss": 0.5103,
      "step": 13671
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9710244536399841,
      "learning_rate": 4.6224175127834057e-07,
      "loss": 0.4978,
      "step": 13672
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.994428813457489,
      "learning_rate": 4.61810990541286e-07,
      "loss": 0.446,
      "step": 13673
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0358386039733887,
      "learning_rate": 4.61380420891584e-07,
      "loss": 0.5114,
      "step": 13674
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1287455558776855,
      "learning_rate": 4.6095004234736175e-07,
      "loss": 0.5337,
      "step": 13675
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.040019154548645,
      "learning_rate": 4.6051985492674425e-07,
      "loss": 0.4942,
      "step": 13676
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1022597551345825,
      "learning_rate": 4.6008985864784473e-07,
      "loss": 0.5078,
      "step": 13677
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.071824312210083,
      "learning_rate": 4.596600535287671e-07,
      "loss": 0.5336,
      "step": 13678
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.101735234260559,
      "learning_rate": 4.592304395876102e-07,
      "loss": 0.5188,
      "step": 13679
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0816458463668823,
      "learning_rate": 4.588010168424628e-07,
      "loss": 0.5067,
      "step": 13680
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0752999782562256,
      "learning_rate": 4.5837178531140723e-07,
      "loss": 0.4947,
      "step": 13681
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0864089727401733,
      "learning_rate": 4.579427450125179e-07,
      "loss": 0.4854,
      "step": 13682
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9950405359268188,
      "learning_rate": 4.5751389596385755e-07,
      "loss": 0.4512,
      "step": 13683
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1778031587600708,
      "learning_rate": 4.570852381834839e-07,
      "loss": 0.4902,
      "step": 13684
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0931570529937744,
      "learning_rate": 4.5665677168944935e-07,
      "loss": 0.5498,
      "step": 13685
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0452144145965576,
      "learning_rate": 4.562284964997915e-07,
      "loss": 0.4889,
      "step": 13686
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0706392526626587,
      "learning_rate": 4.5580041263254547e-07,
      "loss": 0.5363,
      "step": 13687
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0081599950790405,
      "learning_rate": 4.553725201057363e-07,
      "loss": 0.4986,
      "step": 13688
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1238662004470825,
      "learning_rate": 4.5494481893738005e-07,
      "loss": 0.4684,
      "step": 13689
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1825469732284546,
      "learning_rate": 4.5451730914548744e-07,
      "loss": 0.5582,
      "step": 13690
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9957408308982849,
      "learning_rate": 4.540899907480578e-07,
      "loss": 0.4999,
      "step": 13691
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0893901586532593,
      "learning_rate": 4.536628637630836e-07,
      "loss": 0.4806,
      "step": 13692
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0604897737503052,
      "learning_rate": 4.532359282085519e-07,
      "loss": 0.5332,
      "step": 13693
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0288249254226685,
      "learning_rate": 4.528091841024379e-07,
      "loss": 0.481,
      "step": 13694
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0741475820541382,
      "learning_rate": 4.5238263146271053e-07,
      "loss": 0.5142,
      "step": 13695
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0991498231887817,
      "learning_rate": 4.5195627030733156e-07,
      "loss": 0.5221,
      "step": 13696
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0029066801071167,
      "learning_rate": 4.5153010065425054e-07,
      "loss": 0.4646,
      "step": 13697
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0602319240570068,
      "learning_rate": 4.511041225214158e-07,
      "loss": 0.4879,
      "step": 13698
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9987397789955139,
      "learning_rate": 4.5067833592676136e-07,
      "loss": 0.4726,
      "step": 13699
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0657809972763062,
      "learning_rate": 4.502527408882157e-07,
      "loss": 0.4613,
      "step": 13700
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.000096082687378,
      "learning_rate": 4.498273374237e-07,
      "loss": 0.5043,
      "step": 13701
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1645472049713135,
      "learning_rate": 4.494021255511266e-07,
      "loss": 0.5265,
      "step": 13702
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0900377035140991,
      "learning_rate": 4.48977105288399e-07,
      "loss": 0.4996,
      "step": 13703
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1540076732635498,
      "learning_rate": 4.485522766534145e-07,
      "loss": 0.5581,
      "step": 13704
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9702553749084473,
      "learning_rate": 4.4812763966405825e-07,
      "loss": 0.4524,
      "step": 13705
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0510414838790894,
      "learning_rate": 4.4770319433821487e-07,
      "loss": 0.506,
      "step": 13706
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0558525323867798,
      "learning_rate": 4.472789406937522e-07,
      "loss": 0.5163,
      "step": 13707
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0879522562026978,
      "learning_rate": 4.468548787485355e-07,
      "loss": 0.5177,
      "step": 13708
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1908034086227417,
      "learning_rate": 4.4643100852042097e-07,
      "loss": 0.5214,
      "step": 13709
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0457243919372559,
      "learning_rate": 4.4600733002725547e-07,
      "loss": 0.4969,
      "step": 13710
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.009068250656128,
      "learning_rate": 4.4558384328687975e-07,
      "loss": 0.4928,
      "step": 13711
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0318927764892578,
      "learning_rate": 4.451605483171251e-07,
      "loss": 0.4985,
      "step": 13712
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0235506296157837,
      "learning_rate": 4.4473744513581384e-07,
      "loss": 0.442,
      "step": 13713
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0120404958724976,
      "learning_rate": 4.443145337607624e-07,
      "loss": 0.4789,
      "step": 13714
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0483548641204834,
      "learning_rate": 4.4389181420977814e-07,
      "loss": 0.5485,
      "step": 13715
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.004766583442688,
      "learning_rate": 4.4346928650065957e-07,
      "loss": 0.4707,
      "step": 13716
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0503003597259521,
      "learning_rate": 4.4304695065119807e-07,
      "loss": 0.4899,
      "step": 13717
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0432121753692627,
      "learning_rate": 4.4262480667917774e-07,
      "loss": 0.4798,
      "step": 13718
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0445311069488525,
      "learning_rate": 4.422028546023721e-07,
      "loss": 0.4877,
      "step": 13719
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.067617654800415,
      "learning_rate": 4.4178109443855033e-07,
      "loss": 0.5021,
      "step": 13720
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.034996747970581,
      "learning_rate": 4.4135952620546876e-07,
      "loss": 0.4994,
      "step": 13721
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0224899053573608,
      "learning_rate": 4.409381499208787e-07,
      "loss": 0.4401,
      "step": 13722
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0104342699050903,
      "learning_rate": 4.405169656025238e-07,
      "loss": 0.4929,
      "step": 13723
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1080906391143799,
      "learning_rate": 4.400959732681381e-07,
      "loss": 0.5192,
      "step": 13724
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0382052659988403,
      "learning_rate": 4.3967517293544814e-07,
      "loss": 0.4673,
      "step": 13725
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1179478168487549,
      "learning_rate": 4.3925456462217244e-07,
      "loss": 0.521,
      "step": 13726
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0882370471954346,
      "learning_rate": 4.3883414834602125e-07,
      "loss": 0.4925,
      "step": 13727
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0186153650283813,
      "learning_rate": 4.384139241246982e-07,
      "loss": 0.5467,
      "step": 13728
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0723274946212769,
      "learning_rate": 4.3799389197589525e-07,
      "loss": 0.5052,
      "step": 13729
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0477349758148193,
      "learning_rate": 4.375740519172994e-07,
      "loss": 0.5355,
      "step": 13730
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0270122289657593,
      "learning_rate": 4.3715440396658816e-07,
      "loss": 0.4898,
      "step": 13731
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1825838088989258,
      "learning_rate": 4.3673494814143234e-07,
      "loss": 0.5149,
      "step": 13732
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2040338516235352,
      "learning_rate": 4.3631568445949403e-07,
      "loss": 0.4735,
      "step": 13733
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0329917669296265,
      "learning_rate": 4.3589661293842624e-07,
      "loss": 0.4847,
      "step": 13734
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.091172218322754,
      "learning_rate": 4.3547773359587377e-07,
      "loss": 0.5329,
      "step": 13735
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0232014656066895,
      "learning_rate": 4.350590464494764e-07,
      "loss": 0.4495,
      "step": 13736
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0511703491210938,
      "learning_rate": 4.346405515168617e-07,
      "loss": 0.487,
      "step": 13737
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.158629298210144,
      "learning_rate": 4.342222488156511e-07,
      "loss": 0.522,
      "step": 13738
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0807676315307617,
      "learning_rate": 4.3380413836345893e-07,
      "loss": 0.5195,
      "step": 13739
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0331394672393799,
      "learning_rate": 4.333862201778899e-07,
      "loss": 0.4433,
      "step": 13740
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.116027593612671,
      "learning_rate": 4.329684942765411e-07,
      "loss": 0.4983,
      "step": 13741
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1004745960235596,
      "learning_rate": 4.3255096067700176e-07,
      "loss": 0.4979,
      "step": 13742
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0236526727676392,
      "learning_rate": 4.321336193968523e-07,
      "loss": 0.5051,
      "step": 13743
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0533157587051392,
      "learning_rate": 4.3171647045366525e-07,
      "loss": 0.4792,
      "step": 13744
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0958071947097778,
      "learning_rate": 4.312995138650056e-07,
      "loss": 0.5173,
      "step": 13745
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9877141714096069,
      "learning_rate": 4.3088274964843027e-07,
      "loss": 0.474,
      "step": 13746
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1106044054031372,
      "learning_rate": 4.3046617782148857e-07,
      "loss": 0.5205,
      "step": 13747
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.08500075340271,
      "learning_rate": 4.300497984017182e-07,
      "loss": 0.4695,
      "step": 13748
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0543922185897827,
      "learning_rate": 4.2963361140665405e-07,
      "loss": 0.4411,
      "step": 13749
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.01895010471344,
      "learning_rate": 4.292176168538198e-07,
      "loss": 0.4968,
      "step": 13750
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.195759654045105,
      "learning_rate": 4.2880181476073034e-07,
      "loss": 0.5137,
      "step": 13751
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.080329179763794,
      "learning_rate": 4.283862051448945e-07,
      "loss": 0.5073,
      "step": 13752
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0563124418258667,
      "learning_rate": 4.279707880238121e-07,
      "loss": 0.4968,
      "step": 13753
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.100950837135315,
      "learning_rate": 4.275555634149753e-07,
      "loss": 0.525,
      "step": 13754
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0707536935806274,
      "learning_rate": 4.2714053133586785e-07,
      "loss": 0.4671,
      "step": 13755
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0423051118850708,
      "learning_rate": 4.267256918039625e-07,
      "loss": 0.4598,
      "step": 13756
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9561262130737305,
      "learning_rate": 4.263110448367308e-07,
      "loss": 0.4481,
      "step": 13757
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.06647789478302,
      "learning_rate": 4.2589659045163044e-07,
      "loss": 0.5527,
      "step": 13758
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.132185697555542,
      "learning_rate": 4.254823286661125e-07,
      "loss": 0.5088,
      "step": 13759
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0688728094100952,
      "learning_rate": 4.250682594976191e-07,
      "loss": 0.4909,
      "step": 13760
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0723905563354492,
      "learning_rate": 4.2465438296358685e-07,
      "loss": 0.5008,
      "step": 13761
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0533455610275269,
      "learning_rate": 4.2424069908144236e-07,
      "loss": 0.4684,
      "step": 13762
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1853251457214355,
      "learning_rate": 4.2382720786860453e-07,
      "loss": 0.5125,
      "step": 13763
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1141843795776367,
      "learning_rate": 4.2341390934248273e-07,
      "loss": 0.477,
      "step": 13764
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0213943719863892,
      "learning_rate": 4.230008035204797e-07,
      "loss": 0.5005,
      "step": 13765
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.955684244632721,
      "learning_rate": 4.225878904199926e-07,
      "loss": 0.4232,
      "step": 13766
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.043104648590088,
      "learning_rate": 4.2217517005840423e-07,
      "loss": 0.4865,
      "step": 13767
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0324668884277344,
      "learning_rate": 4.2176264245309517e-07,
      "loss": 0.5011,
      "step": 13768
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0168111324310303,
      "learning_rate": 4.2135030762143424e-07,
      "loss": 0.4794,
      "step": 13769
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0939329862594604,
      "learning_rate": 4.2093816558078373e-07,
      "loss": 0.571,
      "step": 13770
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0314226150512695,
      "learning_rate": 4.205262163484991e-07,
      "loss": 0.4627,
      "step": 13771
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0353028774261475,
      "learning_rate": 4.2011445994192324e-07,
      "loss": 0.4867,
      "step": 13772
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1914775371551514,
      "learning_rate": 4.1970289637839556e-07,
      "loss": 0.499,
      "step": 13773
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0128769874572754,
      "learning_rate": 4.19291525675245e-07,
      "loss": 0.4959,
      "step": 13774
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2244237661361694,
      "learning_rate": 4.1888034784979326e-07,
      "loss": 0.5313,
      "step": 13775
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9951699376106262,
      "learning_rate": 4.184693629193537e-07,
      "loss": 0.5494,
      "step": 13776
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0847532749176025,
      "learning_rate": 4.180585709012319e-07,
      "loss": 0.5002,
      "step": 13777
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0222808122634888,
      "learning_rate": 4.1764797181272296e-07,
      "loss": 0.506,
      "step": 13778
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0069555044174194,
      "learning_rate": 4.172375656711181e-07,
      "loss": 0.4506,
      "step": 13779
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0545144081115723,
      "learning_rate": 4.1682735249369663e-07,
      "loss": 0.5265,
      "step": 13780
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0830049514770508,
      "learning_rate": 4.1641733229773163e-07,
      "loss": 0.4831,
      "step": 13781
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0026416778564453,
      "learning_rate": 4.1600750510048805e-07,
      "loss": 0.5108,
      "step": 13782
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0848588943481445,
      "learning_rate": 4.1559787091922153e-07,
      "loss": 0.5119,
      "step": 13783
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0871912240982056,
      "learning_rate": 4.151884297711806e-07,
      "loss": 0.4547,
      "step": 13784
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0597244501113892,
      "learning_rate": 4.147791816736063e-07,
      "loss": 0.4979,
      "step": 13785
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0201863050460815,
      "learning_rate": 4.143701266437283e-07,
      "loss": 0.4842,
      "step": 13786
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.080246925354004,
      "learning_rate": 4.139612646987734e-07,
      "loss": 0.5132,
      "step": 13787
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1408929824829102,
      "learning_rate": 4.135525958559555e-07,
      "loss": 0.5508,
      "step": 13788
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9834961295127869,
      "learning_rate": 4.131441201324826e-07,
      "loss": 0.4859,
      "step": 13789
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.138939380645752,
      "learning_rate": 4.1273583754555424e-07,
      "loss": 0.4789,
      "step": 13790
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0835551023483276,
      "learning_rate": 4.123277481123622e-07,
      "loss": 0.4838,
      "step": 13791
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9488256573677063,
      "learning_rate": 4.1191985185008887e-07,
      "loss": 0.4945,
      "step": 13792
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.041617512702942,
      "learning_rate": 4.1151214877591105e-07,
      "loss": 0.4989,
      "step": 13793
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.108404278755188,
      "learning_rate": 4.1110463890699336e-07,
      "loss": 0.5205,
      "step": 13794
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.07442045211792,
      "learning_rate": 4.1069732226049484e-07,
      "loss": 0.5459,
      "step": 13795
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.000274419784546,
      "learning_rate": 4.102901988535685e-07,
      "loss": 0.4816,
      "step": 13796
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0229767560958862,
      "learning_rate": 4.0988326870335494e-07,
      "loss": 0.4857,
      "step": 13797
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0853655338287354,
      "learning_rate": 4.0947653182698887e-07,
      "loss": 0.5183,
      "step": 13798
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1095023155212402,
      "learning_rate": 4.0906998824159715e-07,
      "loss": 0.5336,
      "step": 13799
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0596293210983276,
      "learning_rate": 4.086636379642972e-07,
      "loss": 0.5024,
      "step": 13800
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1000126600265503,
      "learning_rate": 4.0825748101220087e-07,
      "loss": 0.5201,
      "step": 13801
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0342031717300415,
      "learning_rate": 4.078515174024067e-07,
      "loss": 0.5001,
      "step": 13802
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0115686655044556,
      "learning_rate": 4.074457471520099e-07,
      "loss": 0.5209,
      "step": 13803
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0109944343566895,
      "learning_rate": 4.0704017027809797e-07,
      "loss": 0.5039,
      "step": 13804
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0330100059509277,
      "learning_rate": 4.0663478679774604e-07,
      "loss": 0.4766,
      "step": 13805
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0435158014297485,
      "learning_rate": 4.062295967280239e-07,
      "loss": 0.4726,
      "step": 13806
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9838164448738098,
      "learning_rate": 4.058246000859939e-07,
      "loss": 0.5299,
      "step": 13807
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0656788349151611,
      "learning_rate": 4.054197968887064e-07,
      "loss": 0.4478,
      "step": 13808
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.037368893623352,
      "learning_rate": 4.0501518715320933e-07,
      "loss": 0.4994,
      "step": 13809
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0878825187683105,
      "learning_rate": 4.046107708965369e-07,
      "loss": 0.527,
      "step": 13810
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0550358295440674,
      "learning_rate": 4.042065481357188e-07,
      "loss": 0.5076,
      "step": 13811
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0873349905014038,
      "learning_rate": 4.038025188877753e-07,
      "loss": 0.5078,
      "step": 13812
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.109261155128479,
      "learning_rate": 4.03398683169719e-07,
      "loss": 0.5092,
      "step": 13813
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1236189603805542,
      "learning_rate": 4.029950409985539e-07,
      "loss": 0.5157,
      "step": 13814
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1359589099884033,
      "learning_rate": 4.0259159239127656e-07,
      "loss": 0.5106,
      "step": 13815
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.962879478931427,
      "learning_rate": 4.021883373648722e-07,
      "loss": 0.4692,
      "step": 13816
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9297695755958557,
      "learning_rate": 4.017852759363239e-07,
      "loss": 0.3834,
      "step": 13817
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.05164635181427,
      "learning_rate": 4.013824081226009e-07,
      "loss": 0.5041,
      "step": 13818
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0964182615280151,
      "learning_rate": 4.009797339406674e-07,
      "loss": 0.4969,
      "step": 13819
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0226235389709473,
      "learning_rate": 4.005772534074792e-07,
      "loss": 0.493,
      "step": 13820
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1254289150238037,
      "learning_rate": 4.001749665399807e-07,
      "loss": 0.5216,
      "step": 13821
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0507532358169556,
      "learning_rate": 3.997728733551137e-07,
      "loss": 0.4617,
      "step": 13822
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0152353048324585,
      "learning_rate": 3.993709738698093e-07,
      "loss": 0.5004,
      "step": 13823
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.008558988571167,
      "learning_rate": 3.989692681009877e-07,
      "loss": 0.5099,
      "step": 13824
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0981887578964233,
      "learning_rate": 3.985677560655643e-07,
      "loss": 0.5035,
      "step": 13825
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0890809297561646,
      "learning_rate": 3.9816643778044506e-07,
      "loss": 0.5457,
      "step": 13826
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0290725231170654,
      "learning_rate": 3.977653132625292e-07,
      "loss": 0.4723,
      "step": 13827
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0328872203826904,
      "learning_rate": 3.9736438252870655e-07,
      "loss": 0.5104,
      "step": 13828
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0733381509780884,
      "learning_rate": 3.969636455958564e-07,
      "loss": 0.52,
      "step": 13829
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.997137188911438,
      "learning_rate": 3.965631024808553e-07,
      "loss": 0.526,
      "step": 13830
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0118529796600342,
      "learning_rate": 3.961627532005691e-07,
      "loss": 0.4546,
      "step": 13831
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.021236777305603,
      "learning_rate": 3.957625977718527e-07,
      "loss": 0.497,
      "step": 13832
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0267452001571655,
      "learning_rate": 3.953626362115559e-07,
      "loss": 0.49,
      "step": 13833
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.10640549659729,
      "learning_rate": 3.949628685365203e-07,
      "loss": 0.5001,
      "step": 13834
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0515050888061523,
      "learning_rate": 3.945632947635791e-07,
      "loss": 0.4786,
      "step": 13835
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.076259970664978,
      "learning_rate": 3.941639149095566e-07,
      "loss": 0.5328,
      "step": 13836
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9872125387191772,
      "learning_rate": 3.9376472899126884e-07,
      "loss": 0.4885,
      "step": 13837
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0869286060333252,
      "learning_rate": 3.933657370255228e-07,
      "loss": 0.5239,
      "step": 13838
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1174988746643066,
      "learning_rate": 3.9296693902912244e-07,
      "loss": 0.5198,
      "step": 13839
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.062361478805542,
      "learning_rate": 3.9256833501885693e-07,
      "loss": 0.5302,
      "step": 13840
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0280914306640625,
      "learning_rate": 3.9216992501151074e-07,
      "loss": 0.4902,
      "step": 13841
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0504766702651978,
      "learning_rate": 3.917717090238593e-07,
      "loss": 0.5444,
      "step": 13842
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0632145404815674,
      "learning_rate": 3.913736870726703e-07,
      "loss": 0.4899,
      "step": 13843
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0664408206939697,
      "learning_rate": 3.909758591747037e-07,
      "loss": 0.5637,
      "step": 13844
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.01212739944458,
      "learning_rate": 3.905782253467094e-07,
      "loss": 0.4518,
      "step": 13845
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0291129350662231,
      "learning_rate": 3.9018078560543015e-07,
      "loss": 0.5055,
      "step": 13846
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1620970964431763,
      "learning_rate": 3.8978353996760365e-07,
      "loss": 0.4721,
      "step": 13847
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9888899922370911,
      "learning_rate": 3.8938648844995374e-07,
      "loss": 0.4705,
      "step": 13848
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0864684581756592,
      "learning_rate": 3.889896310691993e-07,
      "loss": 0.5069,
      "step": 13849
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1696125268936157,
      "learning_rate": 3.885929678420508e-07,
      "loss": 0.5196,
      "step": 13850
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.006962537765503,
      "learning_rate": 3.881964987852105e-07,
      "loss": 0.4866,
      "step": 13851
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1452676057815552,
      "learning_rate": 3.878002239153739e-07,
      "loss": 0.5542,
      "step": 13852
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0906898975372314,
      "learning_rate": 3.874041432492237e-07,
      "loss": 0.5263,
      "step": 13853
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1190228462219238,
      "learning_rate": 3.870082568034389e-07,
      "loss": 0.5016,
      "step": 13854
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0661256313323975,
      "learning_rate": 3.866125645946894e-07,
      "loss": 0.4567,
      "step": 13855
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0436792373657227,
      "learning_rate": 3.862170666396359e-07,
      "loss": 0.498,
      "step": 13856
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9427409172058105,
      "learning_rate": 3.858217629549316e-07,
      "loss": 0.4355,
      "step": 13857
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1033661365509033,
      "learning_rate": 3.8542665355722154e-07,
      "loss": 0.4829,
      "step": 13858
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9851303100585938,
      "learning_rate": 3.8503173846314137e-07,
      "loss": 0.4218,
      "step": 13859
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0885196924209595,
      "learning_rate": 3.846370176893205e-07,
      "loss": 0.5067,
      "step": 13860
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0836544036865234,
      "learning_rate": 3.8424249125238065e-07,
      "loss": 0.5167,
      "step": 13861
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0435181856155396,
      "learning_rate": 3.838481591689308e-07,
      "loss": 0.487,
      "step": 13862
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1410058736801147,
      "learning_rate": 3.834540214555771e-07,
      "loss": 0.5455,
      "step": 13863
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9805917143821716,
      "learning_rate": 3.830600781289151e-07,
      "loss": 0.4689,
      "step": 13864
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0559532642364502,
      "learning_rate": 3.826663292055316e-07,
      "loss": 0.4885,
      "step": 13865
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0752135515213013,
      "learning_rate": 3.822727747020072e-07,
      "loss": 0.5398,
      "step": 13866
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1543159484863281,
      "learning_rate": 3.818794146349114e-07,
      "loss": 0.5181,
      "step": 13867
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0757665634155273,
      "learning_rate": 3.8148624902080764e-07,
      "loss": 0.4785,
      "step": 13868
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0365145206451416,
      "learning_rate": 3.8109327787625273e-07,
      "loss": 0.5054,
      "step": 13869
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.090109944343567,
      "learning_rate": 3.807005012177911e-07,
      "loss": 0.4763,
      "step": 13870
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.19999361038208,
      "learning_rate": 3.803079190619624e-07,
      "loss": 0.5465,
      "step": 13871
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0682469606399536,
      "learning_rate": 3.7991553142529616e-07,
      "loss": 0.4847,
      "step": 13872
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0509440898895264,
      "learning_rate": 3.7952333832431466e-07,
      "loss": 0.4749,
      "step": 13873
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0399044752120972,
      "learning_rate": 3.7913133977553306e-07,
      "loss": 0.5162,
      "step": 13874
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0946909189224243,
      "learning_rate": 3.7873953579545486e-07,
      "loss": 0.4957,
      "step": 13875
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.099408745765686,
      "learning_rate": 3.783479264005779e-07,
      "loss": 0.4898,
      "step": 13876
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0778411626815796,
      "learning_rate": 3.779565116073941e-07,
      "loss": 0.4739,
      "step": 13877
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0826133489608765,
      "learning_rate": 3.775652914323813e-07,
      "loss": 0.5707,
      "step": 13878
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1412365436553955,
      "learning_rate": 3.771742658920141e-07,
      "loss": 0.5085,
      "step": 13879
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9909276366233826,
      "learning_rate": 3.767834350027572e-07,
      "loss": 0.4776,
      "step": 13880
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0669307708740234,
      "learning_rate": 3.7639279878106616e-07,
      "loss": 0.5215,
      "step": 13881
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0728931427001953,
      "learning_rate": 3.7600235724339127e-07,
      "loss": 0.5108,
      "step": 13882
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0628244876861572,
      "learning_rate": 3.756121104061705e-07,
      "loss": 0.4737,
      "step": 13883
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0442898273468018,
      "learning_rate": 3.752220582858368e-07,
      "loss": 0.4864,
      "step": 13884
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0579360723495483,
      "learning_rate": 3.748322008988137e-07,
      "loss": 0.5235,
      "step": 13885
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1031591892242432,
      "learning_rate": 3.744425382615169e-07,
      "loss": 0.5246,
      "step": 13886
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1032261848449707,
      "learning_rate": 3.7405307039035387e-07,
      "loss": 0.5555,
      "step": 13887
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0538815259933472,
      "learning_rate": 3.7366379730172376e-07,
      "loss": 0.539,
      "step": 13888
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0281118154525757,
      "learning_rate": 3.732747190120162e-07,
      "loss": 0.5217,
      "step": 13889
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.055055022239685,
      "learning_rate": 3.728858355376164e-07,
      "loss": 0.5277,
      "step": 13890
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0404201745986938,
      "learning_rate": 3.724971468948968e-07,
      "loss": 0.5128,
      "step": 13891
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0271530151367188,
      "learning_rate": 3.721086531002244e-07,
      "loss": 0.5059,
      "step": 13892
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.06915283203125,
      "learning_rate": 3.7172035416995765e-07,
      "loss": 0.5106,
      "step": 13893
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.000438928604126,
      "learning_rate": 3.7133225012044585e-07,
      "loss": 0.4452,
      "step": 13894
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0894824266433716,
      "learning_rate": 3.709443409680308e-07,
      "loss": 0.4878,
      "step": 13895
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0660288333892822,
      "learning_rate": 3.7055662672904723e-07,
      "loss": 0.495,
      "step": 13896
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0462357997894287,
      "learning_rate": 3.7016910741981825e-07,
      "loss": 0.5012,
      "step": 13897
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9886908531188965,
      "learning_rate": 3.6978178305666357e-07,
      "loss": 0.4258,
      "step": 13898
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0848771333694458,
      "learning_rate": 3.693946536558896e-07,
      "loss": 0.4953,
      "step": 13899
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1187751293182373,
      "learning_rate": 3.6900771923379817e-07,
      "loss": 0.5036,
      "step": 13900
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0687930583953857,
      "learning_rate": 3.6862097980668255e-07,
      "loss": 0.5525,
      "step": 13901
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0529654026031494,
      "learning_rate": 3.68234435390824e-07,
      "loss": 0.5304,
      "step": 13902
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1089003086090088,
      "learning_rate": 3.6784808600250186e-07,
      "loss": 0.5179,
      "step": 13903
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1205693483352661,
      "learning_rate": 3.674619316579836e-07,
      "loss": 0.5054,
      "step": 13904
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0788072347640991,
      "learning_rate": 3.670759723735273e-07,
      "loss": 0.4904,
      "step": 13905
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0418163537979126,
      "learning_rate": 3.666902081653845e-07,
      "loss": 0.4827,
      "step": 13906
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1107317209243774,
      "learning_rate": 3.663046390497993e-07,
      "loss": 0.5121,
      "step": 13907
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0846924781799316,
      "learning_rate": 3.659192650430066e-07,
      "loss": 0.5148,
      "step": 13908
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0620695352554321,
      "learning_rate": 3.655340861612333e-07,
      "loss": 0.4897,
      "step": 13909
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9974926710128784,
      "learning_rate": 3.6514910242069547e-07,
      "loss": 0.4973,
      "step": 13910
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0115067958831787,
      "learning_rate": 3.647643138376067e-07,
      "loss": 0.486,
      "step": 13911
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.189638376235962,
      "learning_rate": 3.6437972042816904e-07,
      "loss": 0.5415,
      "step": 13912
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0261467695236206,
      "learning_rate": 3.6399532220857403e-07,
      "loss": 0.4823,
      "step": 13913
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1787561178207397,
      "learning_rate": 3.6361111919500815e-07,
      "loss": 0.5182,
      "step": 13914
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0854473114013672,
      "learning_rate": 3.6322711140364953e-07,
      "loss": 0.5527,
      "step": 13915
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0096983909606934,
      "learning_rate": 3.628432988506675e-07,
      "loss": 0.4698,
      "step": 13916
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1122733354568481,
      "learning_rate": 3.6245968155222243e-07,
      "loss": 0.4852,
      "step": 13917
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.03513765335083,
      "learning_rate": 3.6207625952446756e-07,
      "loss": 0.5188,
      "step": 13918
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0341960191726685,
      "learning_rate": 3.616930327835466e-07,
      "loss": 0.46,
      "step": 13919
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0552611351013184,
      "learning_rate": 3.613100013455972e-07,
      "loss": 0.5315,
      "step": 13920
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1298288106918335,
      "learning_rate": 3.609271652267465e-07,
      "loss": 0.5213,
      "step": 13921
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1473060846328735,
      "learning_rate": 3.6054452444311493e-07,
      "loss": 0.5293,
      "step": 13922
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9427440762519836,
      "learning_rate": 3.601620790108135e-07,
      "loss": 0.4833,
      "step": 13923
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0919525623321533,
      "learning_rate": 3.597798289459464e-07,
      "loss": 0.5746,
      "step": 13924
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0796968936920166,
      "learning_rate": 3.593977742646088e-07,
      "loss": 0.4685,
      "step": 13925
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0801947116851807,
      "learning_rate": 3.5901591498288755e-07,
      "loss": 0.4749,
      "step": 13926
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0195661783218384,
      "learning_rate": 3.5863425111686e-07,
      "loss": 0.4622,
      "step": 13927
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1256657838821411,
      "learning_rate": 3.5825278268259987e-07,
      "loss": 0.5211,
      "step": 13928
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0019874572753906,
      "learning_rate": 3.5787150969616657e-07,
      "loss": 0.4866,
      "step": 13929
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0935301780700684,
      "learning_rate": 3.57490432173615e-07,
      "loss": 0.5172,
      "step": 13930
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0015305280685425,
      "learning_rate": 3.5710955013099233e-07,
      "loss": 0.4644,
      "step": 13931
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.057268500328064,
      "learning_rate": 3.5672886358433356e-07,
      "loss": 0.4947,
      "step": 13932
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.059360384941101,
      "learning_rate": 3.5634837254967023e-07,
      "loss": 0.4757,
      "step": 13933
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0322678089141846,
      "learning_rate": 3.559680770430235e-07,
      "loss": 0.4716,
      "step": 13934
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0753318071365356,
      "learning_rate": 3.555879770804049e-07,
      "loss": 0.5344,
      "step": 13935
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0942437648773193,
      "learning_rate": 3.5520807267782007e-07,
      "loss": 0.4606,
      "step": 13936
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1344181299209595,
      "learning_rate": 3.548283638512651e-07,
      "loss": 0.521,
      "step": 13937
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.02778959274292,
      "learning_rate": 3.544488506167282e-07,
      "loss": 0.5025,
      "step": 13938
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0618828535079956,
      "learning_rate": 3.5406953299019056e-07,
      "loss": 0.4692,
      "step": 13939
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0954022407531738,
      "learning_rate": 3.5369041098762103e-07,
      "loss": 0.512,
      "step": 13940
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1816658973693848,
      "learning_rate": 3.5331148462498635e-07,
      "loss": 0.5178,
      "step": 13941
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0592976808547974,
      "learning_rate": 3.529327539182403e-07,
      "loss": 0.4907,
      "step": 13942
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.006933569908142,
      "learning_rate": 3.5255421888332976e-07,
      "loss": 0.4825,
      "step": 13943
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9933514595031738,
      "learning_rate": 3.5217587953619404e-07,
      "loss": 0.475,
      "step": 13944
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.00538170337677,
      "learning_rate": 3.517977358927632e-07,
      "loss": 0.481,
      "step": 13945
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0523886680603027,
      "learning_rate": 3.514197879689596e-07,
      "loss": 0.5091,
      "step": 13946
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0707453489303589,
      "learning_rate": 3.5104203578069817e-07,
      "loss": 0.5074,
      "step": 13947
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0507975816726685,
      "learning_rate": 3.506644793438835e-07,
      "loss": 0.4826,
      "step": 13948
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1410112380981445,
      "learning_rate": 3.502871186744128e-07,
      "loss": 0.5147,
      "step": 13949
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0179779529571533,
      "learning_rate": 3.499099537881784e-07,
      "loss": 0.5329,
      "step": 13950
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.163852572441101,
      "learning_rate": 3.495329847010581e-07,
      "loss": 0.5019,
      "step": 13951
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.048890233039856,
      "learning_rate": 3.4915621142892595e-07,
      "loss": 0.5094,
      "step": 13952
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1120535135269165,
      "learning_rate": 3.48779633987647e-07,
      "loss": 0.5243,
      "step": 13953
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0523576736450195,
      "learning_rate": 3.4840325239307693e-07,
      "loss": 0.5379,
      "step": 13954
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0382822751998901,
      "learning_rate": 3.4802706666106525e-07,
      "loss": 0.4986,
      "step": 13955
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.141065239906311,
      "learning_rate": 3.476510768074498e-07,
      "loss": 0.4903,
      "step": 13956
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0527931451797485,
      "learning_rate": 3.4727528284806247e-07,
      "loss": 0.5229,
      "step": 13957
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.133631944656372,
      "learning_rate": 3.468996847987288e-07,
      "loss": 0.4652,
      "step": 13958
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1136630773544312,
      "learning_rate": 3.4652428267526184e-07,
      "loss": 0.577,
      "step": 13959
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0631787776947021,
      "learning_rate": 3.4614907649346884e-07,
      "loss": 0.4837,
      "step": 13960
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.162307858467102,
      "learning_rate": 3.4577406626914947e-07,
      "loss": 0.5281,
      "step": 13961
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.093318223953247,
      "learning_rate": 3.453992520180921e-07,
      "loss": 0.5125,
      "step": 13962
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0075652599334717,
      "learning_rate": 3.4502463375608143e-07,
      "loss": 0.4865,
      "step": 13963
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0205689668655396,
      "learning_rate": 3.446502114988892e-07,
      "loss": 0.4908,
      "step": 13964
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1372061967849731,
      "learning_rate": 3.442759852622812e-07,
      "loss": 0.5417,
      "step": 13965
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9880989193916321,
      "learning_rate": 3.4390195506201594e-07,
      "loss": 0.4487,
      "step": 13966
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.061226487159729,
      "learning_rate": 3.435281209138419e-07,
      "loss": 0.5025,
      "step": 13967
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.071082592010498,
      "learning_rate": 3.4315448283349985e-07,
      "loss": 0.5042,
      "step": 13968
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.08926260471344,
      "learning_rate": 3.4278104083672383e-07,
      "loss": 0.5171,
      "step": 13969
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0360966920852661,
      "learning_rate": 3.424077949392346e-07,
      "loss": 0.4877,
      "step": 13970
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0709553956985474,
      "learning_rate": 3.4203474515675293e-07,
      "loss": 0.5125,
      "step": 13971
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0262420177459717,
      "learning_rate": 3.4166189150498297e-07,
      "loss": 0.5018,
      "step": 13972
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0908358097076416,
      "learning_rate": 3.4128923399962543e-07,
      "loss": 0.4687,
      "step": 13973
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0744664669036865,
      "learning_rate": 3.4091677265637224e-07,
      "loss": 0.556,
      "step": 13974
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0527634620666504,
      "learning_rate": 3.405445074909053e-07,
      "loss": 0.4932,
      "step": 13975
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0401239395141602,
      "learning_rate": 3.401724385189009e-07,
      "loss": 0.462,
      "step": 13976
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0116753578186035,
      "learning_rate": 3.398005657560249e-07,
      "loss": 0.4909,
      "step": 13977
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0280588865280151,
      "learning_rate": 3.394288892179348e-07,
      "loss": 0.4832,
      "step": 13978
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.989622175693512,
      "learning_rate": 3.390574089202814e-07,
      "loss": 0.5134,
      "step": 13979
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0852323770523071,
      "learning_rate": 3.3868612487870657e-07,
      "loss": 0.5206,
      "step": 13980
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0556645393371582,
      "learning_rate": 3.3831503710884286e-07,
      "loss": 0.4992,
      "step": 13981
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.13899564743042,
      "learning_rate": 3.379441456263166e-07,
      "loss": 0.5275,
      "step": 13982
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9144753813743591,
      "learning_rate": 3.375734504467437e-07,
      "loss": 0.4646,
      "step": 13983
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9743185639381409,
      "learning_rate": 3.372029515857339e-07,
      "loss": 0.495,
      "step": 13984
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0322911739349365,
      "learning_rate": 3.368326490588875e-07,
      "loss": 0.4807,
      "step": 13985
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.016212821006775,
      "learning_rate": 3.364625428817958e-07,
      "loss": 0.4527,
      "step": 13986
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1167676448822021,
      "learning_rate": 3.360926330700431e-07,
      "loss": 0.505,
      "step": 13987
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0469144582748413,
      "learning_rate": 3.3572291963920536e-07,
      "loss": 0.4573,
      "step": 13988
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0507501363754272,
      "learning_rate": 3.353534026048494e-07,
      "loss": 0.5354,
      "step": 13989
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0410934686660767,
      "learning_rate": 3.3498408198253453e-07,
      "loss": 0.4821,
      "step": 13990
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9763138890266418,
      "learning_rate": 3.3461495778781104e-07,
      "loss": 0.4706,
      "step": 13991
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9764253497123718,
      "learning_rate": 3.342460300362227e-07,
      "loss": 0.4674,
      "step": 13992
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9839788675308228,
      "learning_rate": 3.3387729874330367e-07,
      "loss": 0.4861,
      "step": 13993
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0251315832138062,
      "learning_rate": 3.335087639245782e-07,
      "loss": 0.5231,
      "step": 13994
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0381351709365845,
      "learning_rate": 3.331404255955656e-07,
      "loss": 0.4944,
      "step": 13995
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0450767278671265,
      "learning_rate": 3.327722837717745e-07,
      "loss": 0.4844,
      "step": 13996
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0122859477996826,
      "learning_rate": 3.32404338468707e-07,
      "loss": 0.4894,
      "step": 13997
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.157048225402832,
      "learning_rate": 3.320365897018546e-07,
      "loss": 0.5038,
      "step": 13998
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1703745126724243,
      "learning_rate": 3.316690374867043e-07,
      "loss": 0.4965,
      "step": 13999
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1754573583602905,
      "learning_rate": 3.313016818387288e-07,
      "loss": 0.4803,
      "step": 14000
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0195411443710327,
      "learning_rate": 3.309345227734001e-07,
      "loss": 0.5153,
      "step": 14001
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.044917345046997,
      "learning_rate": 3.305675603061753e-07,
      "loss": 0.5025,
      "step": 14002
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0109405517578125,
      "learning_rate": 3.3020079445250655e-07,
      "loss": 0.5369,
      "step": 14003
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.014048457145691,
      "learning_rate": 3.2983422522783747e-07,
      "loss": 0.4931,
      "step": 14004
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.012646198272705,
      "learning_rate": 3.2946785264760305e-07,
      "loss": 0.4602,
      "step": 14005
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0836317539215088,
      "learning_rate": 3.291016767272298e-07,
      "loss": 0.535,
      "step": 14006
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0748432874679565,
      "learning_rate": 3.287356974821365e-07,
      "loss": 0.5124,
      "step": 14007
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0747649669647217,
      "learning_rate": 3.28369914927732e-07,
      "loss": 0.4709,
      "step": 14008
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0255682468414307,
      "learning_rate": 3.2800432907941935e-07,
      "loss": 0.4898,
      "step": 14009
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1123892068862915,
      "learning_rate": 3.276389399525914e-07,
      "loss": 0.4793,
      "step": 14010
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1055023670196533,
      "learning_rate": 3.272737475626342e-07,
      "loss": 0.5013,
      "step": 14011
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0478788614273071,
      "learning_rate": 3.269087519249242e-07,
      "loss": 0.4817,
      "step": 14012
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.009963870048523,
      "learning_rate": 3.2654395305482924e-07,
      "loss": 0.5239,
      "step": 14013
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1160995960235596,
      "learning_rate": 3.2617935096771137e-07,
      "loss": 0.4961,
      "step": 14014
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0429821014404297,
      "learning_rate": 3.258149456789228e-07,
      "loss": 0.4966,
      "step": 14015
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1230183839797974,
      "learning_rate": 3.2545073720380573e-07,
      "loss": 0.4588,
      "step": 14016
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0803278684616089,
      "learning_rate": 3.2508672555769617e-07,
      "loss": 0.4747,
      "step": 14017
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.160321831703186,
      "learning_rate": 3.2472291075592246e-07,
      "loss": 0.5333,
      "step": 14018
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9995227456092834,
      "learning_rate": 3.243592928138023e-07,
      "loss": 0.4823,
      "step": 14019
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0570666790008545,
      "learning_rate": 3.2399587174664794e-07,
      "loss": 0.5001,
      "step": 14020
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.076535701751709,
      "learning_rate": 3.236326475697593e-07,
      "loss": 0.49,
      "step": 14021
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0700603723526,
      "learning_rate": 3.232696202984326e-07,
      "loss": 0.4804,
      "step": 14022
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9756863713264465,
      "learning_rate": 3.2290678994795377e-07,
      "loss": 0.4687,
      "step": 14023
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0192291736602783,
      "learning_rate": 3.2254415653359906e-07,
      "loss": 0.5132,
      "step": 14024
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0446388721466064,
      "learning_rate": 3.2218172007063787e-07,
      "loss": 0.5352,
      "step": 14025
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0355427265167236,
      "learning_rate": 3.218194805743319e-07,
      "loss": 0.4519,
      "step": 14026
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.019925832748413,
      "learning_rate": 3.2145743805993334e-07,
      "loss": 0.4755,
      "step": 14027
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0217880010604858,
      "learning_rate": 3.210955925426873e-07,
      "loss": 0.4838,
      "step": 14028
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.120097041130066,
      "learning_rate": 3.2073394403782823e-07,
      "loss": 0.5432,
      "step": 14029
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.034629225730896,
      "learning_rate": 3.2037249256058445e-07,
      "loss": 0.5506,
      "step": 14030
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0962677001953125,
      "learning_rate": 3.2001123812617663e-07,
      "loss": 0.5269,
      "step": 14031
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1263171434402466,
      "learning_rate": 3.196501807498148e-07,
      "loss": 0.4973,
      "step": 14032
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0495620965957642,
      "learning_rate": 3.192893204467018e-07,
      "loss": 0.4721,
      "step": 14033
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.103755235671997,
      "learning_rate": 3.189286572320327e-07,
      "loss": 0.525,
      "step": 14034
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1351439952850342,
      "learning_rate": 3.185681911209937e-07,
      "loss": 0.4788,
      "step": 14035
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.162345290184021,
      "learning_rate": 3.1820792212876316e-07,
      "loss": 0.4734,
      "step": 14036
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1245495080947876,
      "learning_rate": 3.178478502705101e-07,
      "loss": 0.5024,
      "step": 14037
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1061865091323853,
      "learning_rate": 3.174879755613952e-07,
      "loss": 0.4776,
      "step": 14038
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1747066974639893,
      "learning_rate": 3.1712829801657294e-07,
      "loss": 0.4794,
      "step": 14039
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0927338600158691,
      "learning_rate": 3.167688176511874e-07,
      "loss": 0.5348,
      "step": 14040
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0735430717468262,
      "learning_rate": 3.1640953448037527e-07,
      "loss": 0.5223,
      "step": 14041
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.075239896774292,
      "learning_rate": 3.1605044851926504e-07,
      "loss": 0.487,
      "step": 14042
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1099472045898438,
      "learning_rate": 3.1569155978297463e-07,
      "loss": 0.4947,
      "step": 14043
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0738979578018188,
      "learning_rate": 3.1533286828661915e-07,
      "loss": 0.4845,
      "step": 14044
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0492703914642334,
      "learning_rate": 3.1497437404529875e-07,
      "loss": 0.5279,
      "step": 14045
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0727201700210571,
      "learning_rate": 3.1461607707410914e-07,
      "loss": 0.5097,
      "step": 14046
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.156922698020935,
      "learning_rate": 3.142579773881377e-07,
      "loss": 0.5034,
      "step": 14047
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.085688591003418,
      "learning_rate": 3.1390007500246236e-07,
      "loss": 0.5045,
      "step": 14048
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1152316331863403,
      "learning_rate": 3.135423699321527e-07,
      "loss": 0.5576,
      "step": 14049
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1113206148147583,
      "learning_rate": 3.131848621922717e-07,
      "loss": 0.5027,
      "step": 14050
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0150794982910156,
      "learning_rate": 3.128275517978707e-07,
      "loss": 0.4817,
      "step": 14051
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.154528260231018,
      "learning_rate": 3.124704387639976e-07,
      "loss": 0.5732,
      "step": 14052
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0682569742202759,
      "learning_rate": 3.1211352310568655e-07,
      "loss": 0.5297,
      "step": 14053
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0718752145767212,
      "learning_rate": 3.1175680483796713e-07,
      "loss": 0.5059,
      "step": 14054
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1030879020690918,
      "learning_rate": 3.1140028397585953e-07,
      "loss": 0.4824,
      "step": 14055
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9959520697593689,
      "learning_rate": 3.110439605343751e-07,
      "loss": 0.4686,
      "step": 14056
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1040695905685425,
      "learning_rate": 3.1068783452851856e-07,
      "loss": 0.5372,
      "step": 14057
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0181715488433838,
      "learning_rate": 3.1033190597328456e-07,
      "loss": 0.5207,
      "step": 14058
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0138585567474365,
      "learning_rate": 3.099761748836594e-07,
      "loss": 0.4886,
      "step": 14059
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0275546312332153,
      "learning_rate": 3.0962064127462167e-07,
      "loss": 0.4894,
      "step": 14060
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.049033761024475,
      "learning_rate": 3.092653051611427e-07,
      "loss": 0.4732,
      "step": 14061
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9946855902671814,
      "learning_rate": 3.089101665581834e-07,
      "loss": 0.5534,
      "step": 14062
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.047498106956482,
      "learning_rate": 3.085552254806978e-07,
      "loss": 0.4832,
      "step": 14063
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0959535837173462,
      "learning_rate": 3.0820048194363183e-07,
      "loss": 0.5209,
      "step": 14064
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0900388956069946,
      "learning_rate": 3.0784593596192123e-07,
      "loss": 0.4941,
      "step": 14065
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0469268560409546,
      "learning_rate": 3.074915875504969e-07,
      "loss": 0.4463,
      "step": 14066
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9177989959716797,
      "learning_rate": 3.0713743672427686e-07,
      "loss": 0.4102,
      "step": 14067
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9485307335853577,
      "learning_rate": 3.067834834981731e-07,
      "loss": 0.4613,
      "step": 14068
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.00711190700531,
      "learning_rate": 3.0642972788709203e-07,
      "loss": 0.4552,
      "step": 14069
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0489661693572998,
      "learning_rate": 3.060761699059267e-07,
      "loss": 0.4923,
      "step": 14070
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1041911840438843,
      "learning_rate": 3.057228095695647e-07,
      "loss": 0.5478,
      "step": 14071
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9792740941047668,
      "learning_rate": 3.053696468928857e-07,
      "loss": 0.4304,
      "step": 14072
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0784748792648315,
      "learning_rate": 3.0501668189075794e-07,
      "loss": 0.5076,
      "step": 14073
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0770944356918335,
      "learning_rate": 3.0466391457804666e-07,
      "loss": 0.4694,
      "step": 14074
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0410479307174683,
      "learning_rate": 3.0431134496960333e-07,
      "loss": 0.5219,
      "step": 14075
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.069656252861023,
      "learning_rate": 3.0395897308027443e-07,
      "loss": 0.4861,
      "step": 14076
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1559025049209595,
      "learning_rate": 3.0360679892489643e-07,
      "loss": 0.5164,
      "step": 14077
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0050629377365112,
      "learning_rate": 3.032548225182985e-07,
      "loss": 0.5105,
      "step": 14078
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0117554664611816,
      "learning_rate": 3.029030438753017e-07,
      "loss": 0.521,
      "step": 14079
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.005658745765686,
      "learning_rate": 3.025514630107179e-07,
      "loss": 0.4724,
      "step": 14080
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0892596244812012,
      "learning_rate": 3.0220007993934987e-07,
      "loss": 0.5437,
      "step": 14081
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0093494653701782,
      "learning_rate": 3.018488946759951e-07,
      "loss": 0.4781,
      "step": 14082
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0474176406860352,
      "learning_rate": 3.01497907235439e-07,
      "loss": 0.4923,
      "step": 14083
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9824331998825073,
      "learning_rate": 3.0114711763246096e-07,
      "loss": 0.468,
      "step": 14084
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0985101461410522,
      "learning_rate": 3.007965258818324e-07,
      "loss": 0.5218,
      "step": 14085
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.067883849143982,
      "learning_rate": 3.0044613199831373e-07,
      "loss": 0.5282,
      "step": 14086
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1233497858047485,
      "learning_rate": 3.0009593599666044e-07,
      "loss": 0.5522,
      "step": 14087
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1419447660446167,
      "learning_rate": 2.9974593789161843e-07,
      "loss": 0.5165,
      "step": 14088
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1232085227966309,
      "learning_rate": 2.9939613769792265e-07,
      "loss": 0.5118,
      "step": 14089
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0514224767684937,
      "learning_rate": 2.9904653543030406e-07,
      "loss": 0.5259,
      "step": 14090
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0916627645492554,
      "learning_rate": 2.98697131103482e-07,
      "loss": 0.5176,
      "step": 14091
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0547813177108765,
      "learning_rate": 2.983479247321691e-07,
      "loss": 0.5191,
      "step": 14092
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0756022930145264,
      "learning_rate": 2.979989163310704e-07,
      "loss": 0.5091,
      "step": 14093
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0705457925796509,
      "learning_rate": 2.976501059148779e-07,
      "loss": 0.5185,
      "step": 14094
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0069702863693237,
      "learning_rate": 2.9730149349828265e-07,
      "loss": 0.4211,
      "step": 14095
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0970326662063599,
      "learning_rate": 2.969530790959624e-07,
      "loss": 0.5021,
      "step": 14096
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1100468635559082,
      "learning_rate": 2.9660486272258703e-07,
      "loss": 0.4899,
      "step": 14097
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.123530387878418,
      "learning_rate": 2.9625684439281875e-07,
      "loss": 0.4867,
      "step": 14098
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0842902660369873,
      "learning_rate": 2.959090241213114e-07,
      "loss": 0.5289,
      "step": 14099
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0588605403900146,
      "learning_rate": 2.9556140192271045e-07,
      "loss": 0.529,
      "step": 14100
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.020761489868164,
      "learning_rate": 2.9521397781165475e-07,
      "loss": 0.508,
      "step": 14101
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0218114852905273,
      "learning_rate": 2.9486675180277035e-07,
      "loss": 0.5011,
      "step": 14102
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0822890996932983,
      "learning_rate": 2.9451972391067897e-07,
      "loss": 0.4697,
      "step": 14103
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0297285318374634,
      "learning_rate": 2.941728941499938e-07,
      "loss": 0.4652,
      "step": 14104
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1335946321487427,
      "learning_rate": 2.938262625353172e-07,
      "loss": 0.4826,
      "step": 14105
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.004652976989746,
      "learning_rate": 2.934798290812446e-07,
      "loss": 0.5152,
      "step": 14106
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.048423171043396,
      "learning_rate": 2.931335938023644e-07,
      "loss": 0.4854,
      "step": 14107
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0791021585464478,
      "learning_rate": 2.9278755671325377e-07,
      "loss": 0.4997,
      "step": 14108
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0030678510665894,
      "learning_rate": 2.924417178284855e-07,
      "loss": 0.5084,
      "step": 14109
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.008026361465454,
      "learning_rate": 2.9209607716261856e-07,
      "loss": 0.4792,
      "step": 14110
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0434467792510986,
      "learning_rate": 2.917506347302079e-07,
      "loss": 0.525,
      "step": 14111
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1228113174438477,
      "learning_rate": 2.9140539054580087e-07,
      "loss": 0.5111,
      "step": 14112
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0027477741241455,
      "learning_rate": 2.9106034462393187e-07,
      "loss": 0.4626,
      "step": 14113
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0428019762039185,
      "learning_rate": 2.9071549697913035e-07,
      "loss": 0.4866,
      "step": 14114
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0616719722747803,
      "learning_rate": 2.9037084762591704e-07,
      "loss": 0.4292,
      "step": 14115
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.103381872177124,
      "learning_rate": 2.900263965788036e-07,
      "loss": 0.5328,
      "step": 14116
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0784556865692139,
      "learning_rate": 2.8968214385229453e-07,
      "loss": 0.4891,
      "step": 14117
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0696626901626587,
      "learning_rate": 2.8933808946088383e-07,
      "loss": 0.4962,
      "step": 14118
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0749086141586304,
      "learning_rate": 2.889942334190593e-07,
      "loss": 0.5058,
      "step": 14119
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9835822582244873,
      "learning_rate": 2.8865057574129883e-07,
      "loss": 0.4921,
      "step": 14120
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0954842567443848,
      "learning_rate": 2.8830711644207257e-07,
      "loss": 0.4839,
      "step": 14121
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1231663227081299,
      "learning_rate": 2.8796385553584326e-07,
      "loss": 0.5203,
      "step": 14122
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0576690435409546,
      "learning_rate": 2.8762079303706505e-07,
      "loss": 0.5109,
      "step": 14123
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1051459312438965,
      "learning_rate": 2.8727792896018015e-07,
      "loss": 0.4861,
      "step": 14124
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.027230143547058,
      "learning_rate": 2.8693526331962875e-07,
      "loss": 0.4674,
      "step": 14125
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0249128341674805,
      "learning_rate": 2.865927961298376e-07,
      "loss": 0.4791,
      "step": 14126
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0711956024169922,
      "learning_rate": 2.8625052740522683e-07,
      "loss": 0.4984,
      "step": 14127
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1102118492126465,
      "learning_rate": 2.859084571602083e-07,
      "loss": 0.4992,
      "step": 14128
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.123199462890625,
      "learning_rate": 2.8556658540918603e-07,
      "loss": 0.4748,
      "step": 14129
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0180432796478271,
      "learning_rate": 2.8522491216655403e-07,
      "loss": 0.4835,
      "step": 14130
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9965765476226807,
      "learning_rate": 2.848834374467002e-07,
      "loss": 0.5006,
      "step": 14131
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1160887479782104,
      "learning_rate": 2.8454216126400146e-07,
      "loss": 0.5333,
      "step": 14132
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.216166377067566,
      "learning_rate": 2.842010836328274e-07,
      "loss": 0.5463,
      "step": 14133
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0514857769012451,
      "learning_rate": 2.838602045675426e-07,
      "loss": 0.4975,
      "step": 14134
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0855978727340698,
      "learning_rate": 2.8351952408249726e-07,
      "loss": 0.4765,
      "step": 14135
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0219852924346924,
      "learning_rate": 2.831790421920377e-07,
      "loss": 0.5216,
      "step": 14136
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0745776891708374,
      "learning_rate": 2.828387589104997e-07,
      "loss": 0.5266,
      "step": 14137
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0392996072769165,
      "learning_rate": 2.824986742522118e-07,
      "loss": 0.4988,
      "step": 14138
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0775796175003052,
      "learning_rate": 2.8215878823149466e-07,
      "loss": 0.5413,
      "step": 14139
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0642188787460327,
      "learning_rate": 2.818191008626581e-07,
      "loss": 0.4648,
      "step": 14140
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0208340883255005,
      "learning_rate": 2.8147961216000497e-07,
      "loss": 0.4709,
      "step": 14141
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.038403868675232,
      "learning_rate": 2.8114032213783226e-07,
      "loss": 0.4433,
      "step": 14142
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0203797817230225,
      "learning_rate": 2.808012308104241e-07,
      "loss": 0.5383,
      "step": 14143
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0287710428237915,
      "learning_rate": 2.80462338192059e-07,
      "loss": 0.5089,
      "step": 14144
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0058563947677612,
      "learning_rate": 2.801236442970073e-07,
      "loss": 0.4697,
      "step": 14145
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9962086081504822,
      "learning_rate": 2.797851491395293e-07,
      "loss": 0.5054,
      "step": 14146
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0989856719970703,
      "learning_rate": 2.79446852733879e-07,
      "loss": 0.5381,
      "step": 14147
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0387845039367676,
      "learning_rate": 2.791087550942995e-07,
      "loss": 0.4932,
      "step": 14148
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0991252660751343,
      "learning_rate": 2.7877085623502775e-07,
      "loss": 0.4984,
      "step": 14149
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1020551919937134,
      "learning_rate": 2.784331561702908e-07,
      "loss": 0.5014,
      "step": 14150
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0331794023513794,
      "learning_rate": 2.780956549143088e-07,
      "loss": 0.452,
      "step": 14151
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9821256399154663,
      "learning_rate": 2.7775835248129267e-07,
      "loss": 0.4704,
      "step": 14152
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1878122091293335,
      "learning_rate": 2.7742124888544497e-07,
      "loss": 0.5128,
      "step": 14153
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1046786308288574,
      "learning_rate": 2.7708434414095875e-07,
      "loss": 0.5183,
      "step": 14154
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0835120677947998,
      "learning_rate": 2.7674763826202265e-07,
      "loss": 0.5177,
      "step": 14155
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0790420770645142,
      "learning_rate": 2.764111312628115e-07,
      "loss": 0.5319,
      "step": 14156
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0519121885299683,
      "learning_rate": 2.7607482315749554e-07,
      "loss": 0.4714,
      "step": 14157
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.108740210533142,
      "learning_rate": 2.757387139602352e-07,
      "loss": 0.4883,
      "step": 14158
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0215269327163696,
      "learning_rate": 2.754028036851836e-07,
      "loss": 0.4484,
      "step": 14159
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.124632716178894,
      "learning_rate": 2.750670923464838e-07,
      "loss": 0.4925,
      "step": 14160
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0852049589157104,
      "learning_rate": 2.747315799582728e-07,
      "loss": 0.4534,
      "step": 14161
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0176403522491455,
      "learning_rate": 2.7439626653467555e-07,
      "loss": 0.5349,
      "step": 14162
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0927029848098755,
      "learning_rate": 2.7406115208981345e-07,
      "loss": 0.4894,
      "step": 14163
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9954310059547424,
      "learning_rate": 2.7372623663779575e-07,
      "loss": 0.4644,
      "step": 14164
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0536550283432007,
      "learning_rate": 2.733915201927245e-07,
      "loss": 0.4885,
      "step": 14165
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.105161190032959,
      "learning_rate": 2.7305700276869406e-07,
      "loss": 0.5142,
      "step": 14166
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0193461179733276,
      "learning_rate": 2.727226843797881e-07,
      "loss": 0.4866,
      "step": 14167
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0644636154174805,
      "learning_rate": 2.7238856504008594e-07,
      "loss": 0.5238,
      "step": 14168
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0790624618530273,
      "learning_rate": 2.7205464476365575e-07,
      "loss": 0.4718,
      "step": 14169
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1688655614852905,
      "learning_rate": 2.7172092356455626e-07,
      "loss": 0.5232,
      "step": 14170
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0596356391906738,
      "learning_rate": 2.7138740145684017e-07,
      "loss": 0.5308,
      "step": 14171
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1095932722091675,
      "learning_rate": 2.7105407845455124e-07,
      "loss": 0.5524,
      "step": 14172
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9973832964897156,
      "learning_rate": 2.707209545717238e-07,
      "loss": 0.4979,
      "step": 14173
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.084646224975586,
      "learning_rate": 2.70388029822386e-07,
      "loss": 0.4858,
      "step": 14174
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1202423572540283,
      "learning_rate": 2.700553042205539e-07,
      "loss": 0.4811,
      "step": 14175
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0021730661392212,
      "learning_rate": 2.6972277778023913e-07,
      "loss": 0.4513,
      "step": 14176
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0620533227920532,
      "learning_rate": 2.693904505154432e-07,
      "loss": 0.5149,
      "step": 14177
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1149684190750122,
      "learning_rate": 2.690583224401588e-07,
      "loss": 0.5114,
      "step": 14178
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.963422954082489,
      "learning_rate": 2.687263935683704e-07,
      "loss": 0.4722,
      "step": 14179
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1437015533447266,
      "learning_rate": 2.6839466391405444e-07,
      "loss": 0.5359,
      "step": 14180
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9957416653633118,
      "learning_rate": 2.680631334911793e-07,
      "loss": 0.4835,
      "step": 14181
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.138525128364563,
      "learning_rate": 2.677318023137049e-07,
      "loss": 0.5379,
      "step": 14182
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0956865549087524,
      "learning_rate": 2.674006703955817e-07,
      "loss": 0.5163,
      "step": 14183
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1254189014434814,
      "learning_rate": 2.670697377507514e-07,
      "loss": 0.4935,
      "step": 14184
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0349822044372559,
      "learning_rate": 2.667390043931517e-07,
      "loss": 0.493,
      "step": 14185
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0068690776824951,
      "learning_rate": 2.664084703367059e-07,
      "loss": 0.4698,
      "step": 14186
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0804328918457031,
      "learning_rate": 2.6607813559533236e-07,
      "loss": 0.4675,
      "step": 14187
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0260028839111328,
      "learning_rate": 2.6574800018294043e-07,
      "loss": 0.4944,
      "step": 14188
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0747073888778687,
      "learning_rate": 2.654180641134313e-07,
      "loss": 0.4736,
      "step": 14189
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0424423217773438,
      "learning_rate": 2.650883274006966e-07,
      "loss": 0.4994,
      "step": 14190
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0522016286849976,
      "learning_rate": 2.6475879005862183e-07,
      "loss": 0.5297,
      "step": 14191
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0279130935668945,
      "learning_rate": 2.644294521010804e-07,
      "loss": 0.4684,
      "step": 14192
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1287299394607544,
      "learning_rate": 2.6410031354194175e-07,
      "loss": 0.5351,
      "step": 14193
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.033314824104309,
      "learning_rate": 2.6377137439506373e-07,
      "loss": 0.4824,
      "step": 14194
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0730446577072144,
      "learning_rate": 2.634426346742969e-07,
      "loss": 0.496,
      "step": 14195
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1193658113479614,
      "learning_rate": 2.6311409439348403e-07,
      "loss": 0.496,
      "step": 14196
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0377310514450073,
      "learning_rate": 2.6278575356645687e-07,
      "loss": 0.5061,
      "step": 14197
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0643011331558228,
      "learning_rate": 2.624576122070427e-07,
      "loss": 0.5335,
      "step": 14198
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0411330461502075,
      "learning_rate": 2.621296703290588e-07,
      "loss": 0.4619,
      "step": 14199
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0853526592254639,
      "learning_rate": 2.6180192794631133e-07,
      "loss": 0.4932,
      "step": 14200
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0662851333618164,
      "learning_rate": 2.6147438507260205e-07,
      "loss": 0.5227,
      "step": 14201
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.104363203048706,
      "learning_rate": 2.611470417217227e-07,
      "loss": 0.4924,
      "step": 14202
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.99798983335495,
      "learning_rate": 2.6081989790745554e-07,
      "loss": 0.4812,
      "step": 14203
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0650646686553955,
      "learning_rate": 2.6049295364357684e-07,
      "loss": 0.5305,
      "step": 14204
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0008906126022339,
      "learning_rate": 2.6016620894385113e-07,
      "loss": 0.4693,
      "step": 14205
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0338935852050781,
      "learning_rate": 2.59839663822038e-07,
      "loss": 0.4815,
      "step": 14206
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0857926607131958,
      "learning_rate": 2.5951331829188797e-07,
      "loss": 0.4963,
      "step": 14207
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0653680562973022,
      "learning_rate": 2.591871723671402e-07,
      "loss": 0.4684,
      "step": 14208
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9513919949531555,
      "learning_rate": 2.5886122606152866e-07,
      "loss": 0.5128,
      "step": 14209
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0318009853363037,
      "learning_rate": 2.585354793887779e-07,
      "loss": 0.4508,
      "step": 14210
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0974669456481934,
      "learning_rate": 2.5820993236260305e-07,
      "loss": 0.4683,
      "step": 14211
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9730995297431946,
      "learning_rate": 2.5788458499671376e-07,
      "loss": 0.5013,
      "step": 14212
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1871789693832397,
      "learning_rate": 2.5755943730480735e-07,
      "loss": 0.509,
      "step": 14213
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1025487184524536,
      "learning_rate": 2.5723448930057405e-07,
      "loss": 0.549,
      "step": 14214
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0638960599899292,
      "learning_rate": 2.569097409976995e-07,
      "loss": 0.5135,
      "step": 14215
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.033321499824524,
      "learning_rate": 2.5658519240985444e-07,
      "loss": 0.5202,
      "step": 14216
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0302057266235352,
      "learning_rate": 2.5626084355070634e-07,
      "loss": 0.5184,
      "step": 14217
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0556660890579224,
      "learning_rate": 2.5593669443391145e-07,
      "loss": 0.545,
      "step": 14218
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1177645921707153,
      "learning_rate": 2.556127450731194e-07,
      "loss": 0.5428,
      "step": 14219
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.101194977760315,
      "learning_rate": 2.552889954819704e-07,
      "loss": 0.4978,
      "step": 14220
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.108578085899353,
      "learning_rate": 2.5496544567409577e-07,
      "loss": 0.5178,
      "step": 14221
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0729098320007324,
      "learning_rate": 2.5464209566311847e-07,
      "loss": 0.4939,
      "step": 14222
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.151045560836792,
      "learning_rate": 2.5431894546265654e-07,
      "loss": 0.5319,
      "step": 14223
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.021720051765442,
      "learning_rate": 2.5399599508631356e-07,
      "loss": 0.4726,
      "step": 14224
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.077812671661377,
      "learning_rate": 2.5367324454768916e-07,
      "loss": 0.5378,
      "step": 14225
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0677251815795898,
      "learning_rate": 2.5335069386037414e-07,
      "loss": 0.5248,
      "step": 14226
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0210872888565063,
      "learning_rate": 2.530283430379471e-07,
      "loss": 0.5225,
      "step": 14227
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0425536632537842,
      "learning_rate": 2.5270619209398497e-07,
      "loss": 0.4455,
      "step": 14228
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1434992551803589,
      "learning_rate": 2.523842410420496e-07,
      "loss": 0.5393,
      "step": 14229
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0491610765457153,
      "learning_rate": 2.5206248989569803e-07,
      "loss": 0.4876,
      "step": 14230
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1168839931488037,
      "learning_rate": 2.5174093866847826e-07,
      "loss": 0.4623,
      "step": 14231
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0826624631881714,
      "learning_rate": 2.5141958737392947e-07,
      "loss": 0.5193,
      "step": 14232
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0712814331054688,
      "learning_rate": 2.5109843602558247e-07,
      "loss": 0.5055,
      "step": 14233
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9830718040466309,
      "learning_rate": 2.507774846369615e-07,
      "loss": 0.4724,
      "step": 14234
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0501651763916016,
      "learning_rate": 2.5045673322157735e-07,
      "loss": 0.5073,
      "step": 14235
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0189639329910278,
      "learning_rate": 2.501361817929393e-07,
      "loss": 0.45,
      "step": 14236
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0900280475616455,
      "learning_rate": 2.4981583036454203e-07,
      "loss": 0.4828,
      "step": 14237
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.06129789352417,
      "learning_rate": 2.494956789498759e-07,
      "loss": 0.4955,
      "step": 14238
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1390588283538818,
      "learning_rate": 2.491757275624207e-07,
      "loss": 0.4832,
      "step": 14239
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.138748288154602,
      "learning_rate": 2.4885597621564896e-07,
      "loss": 0.5358,
      "step": 14240
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0612497329711914,
      "learning_rate": 2.485364249230238e-07,
      "loss": 0.4911,
      "step": 14241
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9879231452941895,
      "learning_rate": 2.4821707369800163e-07,
      "loss": 0.4806,
      "step": 14242
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.046484112739563,
      "learning_rate": 2.478979225540268e-07,
      "loss": 0.5028,
      "step": 14243
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0972697734832764,
      "learning_rate": 2.475789715045401e-07,
      "loss": 0.5167,
      "step": 14244
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0879430770874023,
      "learning_rate": 2.472602205629698e-07,
      "loss": 0.4745,
      "step": 14245
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1586531400680542,
      "learning_rate": 2.469416697427379e-07,
      "loss": 0.5199,
      "step": 14246
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.147218108177185,
      "learning_rate": 2.466233190572581e-07,
      "loss": 0.495,
      "step": 14247
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0244981050491333,
      "learning_rate": 2.463051685199341e-07,
      "loss": 0.5247,
      "step": 14248
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0598788261413574,
      "learning_rate": 2.4598721814416306e-07,
      "loss": 0.4921,
      "step": 14249
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9894571304321289,
      "learning_rate": 2.4566946794333247e-07,
      "loss": 0.5022,
      "step": 14250
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0936572551727295,
      "learning_rate": 2.4535191793082116e-07,
      "loss": 0.5097,
      "step": 14251
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.021803855895996,
      "learning_rate": 2.4503456812e-07,
      "loss": 0.4551,
      "step": 14252
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1300513744354248,
      "learning_rate": 2.447174185242324e-07,
      "loss": 0.5114,
      "step": 14253
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0158747434616089,
      "learning_rate": 2.4440046915687135e-07,
      "loss": 0.4517,
      "step": 14254
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0202269554138184,
      "learning_rate": 2.4408372003126345e-07,
      "loss": 0.5171,
      "step": 14255
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0510153770446777,
      "learning_rate": 2.4376717116074533e-07,
      "loss": 0.4848,
      "step": 14256
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9678918719291687,
      "learning_rate": 2.434508225586457e-07,
      "loss": 0.4999,
      "step": 14257
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0264215469360352,
      "learning_rate": 2.431346742382856e-07,
      "loss": 0.4809,
      "step": 14258
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0267912149429321,
      "learning_rate": 2.428187262129761e-07,
      "loss": 0.4719,
      "step": 14259
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.011001706123352,
      "learning_rate": 2.4250297849602145e-07,
      "loss": 0.5374,
      "step": 14260
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0350229740142822,
      "learning_rate": 2.421874311007155e-07,
      "loss": 0.4745,
      "step": 14261
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.153439998626709,
      "learning_rate": 2.41872084040346e-07,
      "loss": 0.4898,
      "step": 14262
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0444737672805786,
      "learning_rate": 2.4155693732819065e-07,
      "loss": 0.4752,
      "step": 14263
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1378803253173828,
      "learning_rate": 2.412419909775199e-07,
      "loss": 0.4922,
      "step": 14264
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.020803689956665,
      "learning_rate": 2.4092724500159315e-07,
      "loss": 0.521,
      "step": 14265
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.044555902481079,
      "learning_rate": 2.406126994136654e-07,
      "loss": 0.4783,
      "step": 14266
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0339492559432983,
      "learning_rate": 2.402983542269799e-07,
      "loss": 0.4759,
      "step": 14267
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0984539985656738,
      "learning_rate": 2.3998420945477276e-07,
      "loss": 0.5041,
      "step": 14268
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0510377883911133,
      "learning_rate": 2.3967026511027224e-07,
      "loss": 0.5406,
      "step": 14269
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0136500597000122,
      "learning_rate": 2.393565212066962e-07,
      "loss": 0.4782,
      "step": 14270
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0046253204345703,
      "learning_rate": 2.3904297775725614e-07,
      "loss": 0.4767,
      "step": 14271
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0714184045791626,
      "learning_rate": 2.3872963477515497e-07,
      "loss": 0.4912,
      "step": 14272
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0916383266448975,
      "learning_rate": 2.3841649227358489e-07,
      "loss": 0.4851,
      "step": 14273
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0295509099960327,
      "learning_rate": 2.3810355026573195e-07,
      "loss": 0.4955,
      "step": 14274
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0655426979064941,
      "learning_rate": 2.377908087647729e-07,
      "loss": 0.5101,
      "step": 14275
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0833792686462402,
      "learning_rate": 2.374782677838766e-07,
      "loss": 0.4631,
      "step": 14276
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0340633392333984,
      "learning_rate": 2.3716592733620315e-07,
      "loss": 0.5448,
      "step": 14277
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.169402837753296,
      "learning_rate": 2.3685378743490306e-07,
      "loss": 0.4724,
      "step": 14278
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0679082870483398,
      "learning_rate": 2.3654184809312032e-07,
      "loss": 0.4914,
      "step": 14279
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9339148998260498,
      "learning_rate": 2.362301093239905e-07,
      "loss": 0.448,
      "step": 14280
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0225436687469482,
      "learning_rate": 2.359185711406381e-07,
      "loss": 0.48,
      "step": 14281
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9594351649284363,
      "learning_rate": 2.3560723355618152e-07,
      "loss": 0.5045,
      "step": 14282
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9956153035163879,
      "learning_rate": 2.3529609658373032e-07,
      "loss": 0.5083,
      "step": 14283
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.073124885559082,
      "learning_rate": 2.3498516023638562e-07,
      "loss": 0.5129,
      "step": 14284
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0852166414260864,
      "learning_rate": 2.3467442452723976e-07,
      "loss": 0.4635,
      "step": 14285
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1746938228607178,
      "learning_rate": 2.3436388946937504e-07,
      "loss": 0.4606,
      "step": 14286
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0940499305725098,
      "learning_rate": 2.3405355507586992e-07,
      "loss": 0.5258,
      "step": 14287
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.974661111831665,
      "learning_rate": 2.3374342135979e-07,
      "loss": 0.5102,
      "step": 14288
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0678569078445435,
      "learning_rate": 2.3343348833419377e-07,
      "loss": 0.5254,
      "step": 14289
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.104823350906372,
      "learning_rate": 2.3312375601213134e-07,
      "loss": 0.4705,
      "step": 14290
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1579174995422363,
      "learning_rate": 2.3281422440664503e-07,
      "loss": 0.5125,
      "step": 14291
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1120131015777588,
      "learning_rate": 2.3250489353076777e-07,
      "loss": 0.5045,
      "step": 14292
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1061304807662964,
      "learning_rate": 2.3219576339752525e-07,
      "loss": 0.5066,
      "step": 14293
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0827875137329102,
      "learning_rate": 2.3188683401993261e-07,
      "loss": 0.5188,
      "step": 14294
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0205931663513184,
      "learning_rate": 2.3157810541099724e-07,
      "loss": 0.4545,
      "step": 14295
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0359200239181519,
      "learning_rate": 2.3126957758372149e-07,
      "loss": 0.4723,
      "step": 14296
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0978033542633057,
      "learning_rate": 2.3096125055109386e-07,
      "loss": 0.4738,
      "step": 14297
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0552581548690796,
      "learning_rate": 2.3065312432609788e-07,
      "loss": 0.484,
      "step": 14298
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0160704851150513,
      "learning_rate": 2.3034519892170705e-07,
      "loss": 0.4968,
      "step": 14299
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.022432804107666,
      "learning_rate": 2.3003747435088764e-07,
      "loss": 0.4674,
      "step": 14300
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0523029565811157,
      "learning_rate": 2.2972995062659764e-07,
      "loss": 0.5172,
      "step": 14301
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0756049156188965,
      "learning_rate": 2.2942262776178392e-07,
      "loss": 0.5255,
      "step": 14302
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9936608672142029,
      "learning_rate": 2.291155057693878e-07,
      "loss": 0.503,
      "step": 14303
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1403104066848755,
      "learning_rate": 2.2880858466234114e-07,
      "loss": 0.505,
      "step": 14304
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0118169784545898,
      "learning_rate": 2.2850186445356693e-07,
      "loss": 0.5185,
      "step": 14305
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9938391447067261,
      "learning_rate": 2.281953451559804e-07,
      "loss": 0.4619,
      "step": 14306
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.087198257446289,
      "learning_rate": 2.2788902678248904e-07,
      "loss": 0.5247,
      "step": 14307
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.039725661277771,
      "learning_rate": 2.2758290934598805e-07,
      "loss": 0.5071,
      "step": 14308
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.034071445465088,
      "learning_rate": 2.2727699285937043e-07,
      "loss": 0.4666,
      "step": 14309
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.218772530555725,
      "learning_rate": 2.2697127733551483e-07,
      "loss": 0.5297,
      "step": 14310
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0864248275756836,
      "learning_rate": 2.2666576278729424e-07,
      "loss": 0.5342,
      "step": 14311
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0638132095336914,
      "learning_rate": 2.2636044922757339e-07,
      "loss": 0.5038,
      "step": 14312
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1192634105682373,
      "learning_rate": 2.2605533666920753e-07,
      "loss": 0.5212,
      "step": 14313
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1108735799789429,
      "learning_rate": 2.257504251250442e-07,
      "loss": 0.4626,
      "step": 14314
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0333267450332642,
      "learning_rate": 2.2544571460792308e-07,
      "loss": 0.5094,
      "step": 14315
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0197890996932983,
      "learning_rate": 2.251412051306717e-07,
      "loss": 0.4993,
      "step": 14316
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.036181092262268,
      "learning_rate": 2.2483689670611542e-07,
      "loss": 0.5418,
      "step": 14317
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1110544204711914,
      "learning_rate": 2.2453278934706446e-07,
      "loss": 0.5396,
      "step": 14318
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0619498491287231,
      "learning_rate": 2.2422888306632584e-07,
      "loss": 0.4909,
      "step": 14319
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.028778314590454,
      "learning_rate": 2.2392517787669487e-07,
      "loss": 0.4499,
      "step": 14320
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1261348724365234,
      "learning_rate": 2.2362167379096023e-07,
      "loss": 0.5234,
      "step": 14321
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.000189185142517,
      "learning_rate": 2.2331837082190056e-07,
      "loss": 0.4514,
      "step": 14322
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0539196729660034,
      "learning_rate": 2.2301526898228842e-07,
      "loss": 0.4784,
      "step": 14323
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0466829538345337,
      "learning_rate": 2.2271236828488474e-07,
      "loss": 0.5106,
      "step": 14324
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0722671747207642,
      "learning_rate": 2.224096687424443e-07,
      "loss": 0.498,
      "step": 14325
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0605310201644897,
      "learning_rate": 2.2210717036771246e-07,
      "loss": 0.4761,
      "step": 14326
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1165287494659424,
      "learning_rate": 2.218048731734268e-07,
      "loss": 0.5357,
      "step": 14327
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9855444431304932,
      "learning_rate": 2.215027771723155e-07,
      "loss": 0.4909,
      "step": 14328
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0161094665527344,
      "learning_rate": 2.2120088237709946e-07,
      "loss": 0.4939,
      "step": 14329
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1264501810073853,
      "learning_rate": 2.2089918880049023e-07,
      "loss": 0.4765,
      "step": 14330
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0361504554748535,
      "learning_rate": 2.205976964551909e-07,
      "loss": 0.4931,
      "step": 14331
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9984549880027771,
      "learning_rate": 2.2029640535389586e-07,
      "loss": 0.4754,
      "step": 14332
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0490080118179321,
      "learning_rate": 2.1999531550929098e-07,
      "loss": 0.5058,
      "step": 14333
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9620938897132874,
      "learning_rate": 2.1969442693405673e-07,
      "loss": 0.4717,
      "step": 14334
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0211260318756104,
      "learning_rate": 2.1939373964085964e-07,
      "loss": 0.5389,
      "step": 14335
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0039058923721313,
      "learning_rate": 2.190932536423618e-07,
      "loss": 0.487,
      "step": 14336
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1592278480529785,
      "learning_rate": 2.1879296895121637e-07,
      "loss": 0.5299,
      "step": 14337
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0102602243423462,
      "learning_rate": 2.1849288558006442e-07,
      "loss": 0.502,
      "step": 14338
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.131880521774292,
      "learning_rate": 2.1819300354154526e-07,
      "loss": 0.5316,
      "step": 14339
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.104852557182312,
      "learning_rate": 2.1789332284828323e-07,
      "loss": 0.5145,
      "step": 14340
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9619296193122864,
      "learning_rate": 2.175938435128977e-07,
      "loss": 0.4365,
      "step": 14341
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0772216320037842,
      "learning_rate": 2.1729456554799855e-07,
      "loss": 0.5193,
      "step": 14342
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1998577117919922,
      "learning_rate": 2.1699548896618795e-07,
      "loss": 0.5393,
      "step": 14343
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.083191990852356,
      "learning_rate": 2.1669661378005802e-07,
      "loss": 0.4686,
      "step": 14344
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0519863367080688,
      "learning_rate": 2.1639794000219426e-07,
      "loss": 0.5003,
      "step": 14345
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0952211618423462,
      "learning_rate": 2.1609946764517108e-07,
      "loss": 0.5063,
      "step": 14346
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.121744155883789,
      "learning_rate": 2.1580119672155898e-07,
      "loss": 0.5417,
      "step": 14347
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0885792970657349,
      "learning_rate": 2.1550312724391452e-07,
      "loss": 0.5188,
      "step": 14348
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0304995775222778,
      "learning_rate": 2.152052592247894e-07,
      "loss": 0.4922,
      "step": 14349
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0206528902053833,
      "learning_rate": 2.1490759267672634e-07,
      "loss": 0.498,
      "step": 14350
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1047496795654297,
      "learning_rate": 2.1461012761225696e-07,
      "loss": 0.5174,
      "step": 14351
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0037637948989868,
      "learning_rate": 2.143128640439085e-07,
      "loss": 0.475,
      "step": 14352
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0277127027511597,
      "learning_rate": 2.1401580198419812e-07,
      "loss": 0.475,
      "step": 14353
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0374289751052856,
      "learning_rate": 2.1371894144563254e-07,
      "loss": 0.4978,
      "step": 14354
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0003501176834106,
      "learning_rate": 2.1342228244071173e-07,
      "loss": 0.5156,
      "step": 14355
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0034006834030151,
      "learning_rate": 2.1312582498192792e-07,
      "loss": 0.4977,
      "step": 14356
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0414868593215942,
      "learning_rate": 2.1282956908176277e-07,
      "loss": 0.5104,
      "step": 14357
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0152901411056519,
      "learning_rate": 2.125335147526919e-07,
      "loss": 0.4949,
      "step": 14358
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0467476844787598,
      "learning_rate": 2.122376620071792e-07,
      "loss": 0.531,
      "step": 14359
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0745909214019775,
      "learning_rate": 2.1194201085768363e-07,
      "loss": 0.478,
      "step": 14360
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0887025594711304,
      "learning_rate": 2.1164656131665407e-07,
      "loss": 0.4818,
      "step": 14361
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.028891921043396,
      "learning_rate": 2.1135131339652947e-07,
      "loss": 0.5066,
      "step": 14362
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0102150440216064,
      "learning_rate": 2.1105626710974325e-07,
      "loss": 0.4772,
      "step": 14363
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.108974814414978,
      "learning_rate": 2.1076142246871766e-07,
      "loss": 0.4554,
      "step": 14364
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0432991981506348,
      "learning_rate": 2.1046677948586836e-07,
      "loss": 0.4985,
      "step": 14365
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0474931001663208,
      "learning_rate": 2.1017233817360149e-07,
      "loss": 0.499,
      "step": 14366
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1366848945617676,
      "learning_rate": 2.098780985443144e-07,
      "loss": 0.5547,
      "step": 14367
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0178462266921997,
      "learning_rate": 2.095840606103966e-07,
      "loss": 0.4721,
      "step": 14368
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0264699459075928,
      "learning_rate": 2.092902243842304e-07,
      "loss": 0.4803,
      "step": 14369
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2258989810943604,
      "learning_rate": 2.0899658987818705e-07,
      "loss": 0.496,
      "step": 14370
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1892192363739014,
      "learning_rate": 2.0870315710462996e-07,
      "loss": 0.5231,
      "step": 14371
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0772793292999268,
      "learning_rate": 2.0840992607591593e-07,
      "loss": 0.4884,
      "step": 14372
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1337677240371704,
      "learning_rate": 2.081168968043906e-07,
      "loss": 0.5228,
      "step": 14373
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0992305278778076,
      "learning_rate": 2.0782406930239363e-07,
      "loss": 0.5369,
      "step": 14374
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0421539545059204,
      "learning_rate": 2.0753144358225397e-07,
      "loss": 0.5005,
      "step": 14375
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1078789234161377,
      "learning_rate": 2.072390196562929e-07,
      "loss": 0.5082,
      "step": 14376
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0897104740142822,
      "learning_rate": 2.0694679753682445e-07,
      "loss": 0.4868,
      "step": 14377
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0101878643035889,
      "learning_rate": 2.0665477723615268e-07,
      "loss": 0.4782,
      "step": 14378
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0218583345413208,
      "learning_rate": 2.063629587665733e-07,
      "loss": 0.4867,
      "step": 14379
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0644055604934692,
      "learning_rate": 2.0607134214037373e-07,
      "loss": 0.4456,
      "step": 14380
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9680347442626953,
      "learning_rate": 2.05779927369833e-07,
      "loss": 0.4666,
      "step": 14381
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0135297775268555,
      "learning_rate": 2.054887144672224e-07,
      "loss": 0.4552,
      "step": 14382
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.971634566783905,
      "learning_rate": 2.0519770344480272e-07,
      "loss": 0.4987,
      "step": 14383
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0475090742111206,
      "learning_rate": 2.0490689431482746e-07,
      "loss": 0.4633,
      "step": 14384
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0539919137954712,
      "learning_rate": 2.0461628708954183e-07,
      "loss": 0.5522,
      "step": 14385
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0730996131896973,
      "learning_rate": 2.0432588178118274e-07,
      "loss": 0.5039,
      "step": 14386
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9802207350730896,
      "learning_rate": 2.0403567840197813e-07,
      "loss": 0.4511,
      "step": 14387
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1255625486373901,
      "learning_rate": 2.0374567696414716e-07,
      "loss": 0.4906,
      "step": 14388
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1176317930221558,
      "learning_rate": 2.0345587747990004e-07,
      "loss": 0.5227,
      "step": 14389
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0412148237228394,
      "learning_rate": 2.0316627996144035e-07,
      "loss": 0.4668,
      "step": 14390
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1022497415542603,
      "learning_rate": 2.028768844209622e-07,
      "loss": 0.5007,
      "step": 14391
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1177451610565186,
      "learning_rate": 2.0258769087065034e-07,
      "loss": 0.5445,
      "step": 14392
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0330328941345215,
      "learning_rate": 2.022986993226811e-07,
      "loss": 0.5068,
      "step": 14393
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9648503661155701,
      "learning_rate": 2.020099097892242e-07,
      "loss": 0.4734,
      "step": 14394
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0981762409210205,
      "learning_rate": 2.0172132228243878e-07,
      "loss": 0.5329,
      "step": 14395
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.131181001663208,
      "learning_rate": 2.014329368144774e-07,
      "loss": 0.479,
      "step": 14396
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1691778898239136,
      "learning_rate": 2.0114475339748085e-07,
      "loss": 0.527,
      "step": 14397
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0281850099563599,
      "learning_rate": 2.0085677204358445e-07,
      "loss": 0.4742,
      "step": 14398
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0565160512924194,
      "learning_rate": 2.005689927649157e-07,
      "loss": 0.5312,
      "step": 14399
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9973809123039246,
      "learning_rate": 2.0028141557358992e-07,
      "loss": 0.5035,
      "step": 14400
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0344796180725098,
      "learning_rate": 1.999940404817169e-07,
      "loss": 0.5028,
      "step": 14401
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1034725904464722,
      "learning_rate": 1.9970686750139633e-07,
      "loss": 0.4856,
      "step": 14402
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.039278268814087,
      "learning_rate": 1.994198966447214e-07,
      "loss": 0.5301,
      "step": 14403
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0402063131332397,
      "learning_rate": 1.991331279237746e-07,
      "loss": 0.5286,
      "step": 14404
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0492223501205444,
      "learning_rate": 1.988465613506302e-07,
      "loss": 0.4573,
      "step": 14405
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.086842656135559,
      "learning_rate": 1.9856019693735463e-07,
      "loss": 0.5222,
      "step": 14406
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0863054990768433,
      "learning_rate": 1.982740346960077e-07,
      "loss": 0.5254,
      "step": 14407
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1995936632156372,
      "learning_rate": 1.9798807463863589e-07,
      "loss": 0.5319,
      "step": 14408
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0130900144577026,
      "learning_rate": 1.977023167772818e-07,
      "loss": 0.4829,
      "step": 14409
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0801007747650146,
      "learning_rate": 1.9741676112397688e-07,
      "loss": 0.4629,
      "step": 14410
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0296639204025269,
      "learning_rate": 1.9713140769074546e-07,
      "loss": 0.5006,
      "step": 14411
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0868396759033203,
      "learning_rate": 1.9684625648960287e-07,
      "loss": 0.4896,
      "step": 14412
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0036476850509644,
      "learning_rate": 1.965613075325551e-07,
      "loss": 0.4749,
      "step": 14413
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0925575494766235,
      "learning_rate": 1.9627656083160085e-07,
      "loss": 0.526,
      "step": 14414
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0652610063552856,
      "learning_rate": 1.9599201639872943e-07,
      "loss": 0.4753,
      "step": 14415
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0112617015838623,
      "learning_rate": 1.9570767424592186e-07,
      "loss": 0.456,
      "step": 14416
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0837966203689575,
      "learning_rate": 1.9542353438515183e-07,
      "loss": 0.4816,
      "step": 14417
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0132328271865845,
      "learning_rate": 1.9513959682838314e-07,
      "loss": 0.5037,
      "step": 14418
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0372248888015747,
      "learning_rate": 1.9485586158757009e-07,
      "loss": 0.5072,
      "step": 14419
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2160942554473877,
      "learning_rate": 1.9457232867466204e-07,
      "loss": 0.535,
      "step": 14420
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1435277462005615,
      "learning_rate": 1.9428899810159606e-07,
      "loss": 0.5158,
      "step": 14421
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1938700675964355,
      "learning_rate": 1.9400586988030212e-07,
      "loss": 0.5031,
      "step": 14422
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0440253019332886,
      "learning_rate": 1.937229440227023e-07,
      "loss": 0.5007,
      "step": 14423
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9880737662315369,
      "learning_rate": 1.9344022054070933e-07,
      "loss": 0.4593,
      "step": 14424
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9918844699859619,
      "learning_rate": 1.9315769944622808e-07,
      "loss": 0.4999,
      "step": 14425
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0597563982009888,
      "learning_rate": 1.9287538075115463e-07,
      "loss": 0.5231,
      "step": 14426
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0294430255889893,
      "learning_rate": 1.9259326446737503e-07,
      "loss": 0.4994,
      "step": 14427
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0238746404647827,
      "learning_rate": 1.9231135060677087e-07,
      "loss": 0.5056,
      "step": 14428
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1582118272781372,
      "learning_rate": 1.9202963918120988e-07,
      "loss": 0.5307,
      "step": 14429
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9513401389122009,
      "learning_rate": 1.9174813020255533e-07,
      "loss": 0.5188,
      "step": 14430
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9851300120353699,
      "learning_rate": 1.9146682368266112e-07,
      "loss": 0.4879,
      "step": 14431
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0757297277450562,
      "learning_rate": 1.9118571963336996e-07,
      "loss": 0.5066,
      "step": 14432
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.122973084449768,
      "learning_rate": 1.9090481806652017e-07,
      "loss": 0.5234,
      "step": 14433
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0094428062438965,
      "learning_rate": 1.9062411899393896e-07,
      "loss": 0.4785,
      "step": 14434
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0200306177139282,
      "learning_rate": 1.9034362242744576e-07,
      "loss": 0.4915,
      "step": 14435
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0503755807876587,
      "learning_rate": 1.9006332837885054e-07,
      "loss": 0.5085,
      "step": 14436
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1125425100326538,
      "learning_rate": 1.8978323685995558e-07,
      "loss": 0.5781,
      "step": 14437
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0734832286834717,
      "learning_rate": 1.8950334788255586e-07,
      "loss": 0.4952,
      "step": 14438
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0181453227996826,
      "learning_rate": 1.8922366145843585e-07,
      "loss": 0.4679,
      "step": 14439
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.951262354850769,
      "learning_rate": 1.8894417759937055e-07,
      "loss": 0.4909,
      "step": 14440
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0891555547714233,
      "learning_rate": 1.886648963171306e-07,
      "loss": 0.4835,
      "step": 14441
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.052882194519043,
      "learning_rate": 1.8838581762347485e-07,
      "loss": 0.4943,
      "step": 14442
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1120226383209229,
      "learning_rate": 1.881069415301534e-07,
      "loss": 0.4715,
      "step": 14443
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1016628742218018,
      "learning_rate": 1.8782826804890908e-07,
      "loss": 0.5361,
      "step": 14444
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0507054328918457,
      "learning_rate": 1.875497971914758e-07,
      "loss": 0.5833,
      "step": 14445
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0540279150009155,
      "learning_rate": 1.872715289695798e-07,
      "loss": 0.4896,
      "step": 14446
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9839321970939636,
      "learning_rate": 1.8699346339493774e-07,
      "loss": 0.5175,
      "step": 14447
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0354623794555664,
      "learning_rate": 1.867156004792575e-07,
      "loss": 0.505,
      "step": 14448
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.145330786705017,
      "learning_rate": 1.864379402342381e-07,
      "loss": 0.4834,
      "step": 14449
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0762995481491089,
      "learning_rate": 1.8616048267157348e-07,
      "loss": 0.5029,
      "step": 14450
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.158921480178833,
      "learning_rate": 1.8588322780294377e-07,
      "loss": 0.4992,
      "step": 14451
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.077999472618103,
      "learning_rate": 1.8560617564002458e-07,
      "loss": 0.4996,
      "step": 14452
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1388030052185059,
      "learning_rate": 1.8532932619448106e-07,
      "loss": 0.4947,
      "step": 14453
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0563603639602661,
      "learning_rate": 1.8505267947797056e-07,
      "loss": 0.4851,
      "step": 14454
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1018091440200806,
      "learning_rate": 1.847762355021421e-07,
      "loss": 0.4919,
      "step": 14455
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.133117914199829,
      "learning_rate": 1.8449999427863575e-07,
      "loss": 0.4967,
      "step": 14456
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0800427198410034,
      "learning_rate": 1.842239558190817e-07,
      "loss": 0.5022,
      "step": 14457
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0592983961105347,
      "learning_rate": 1.839481201351051e-07,
      "loss": 0.5063,
      "step": 14458
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0915387868881226,
      "learning_rate": 1.8367248723831889e-07,
      "loss": 0.5533,
      "step": 14459
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.020050048828125,
      "learning_rate": 1.833970571403293e-07,
      "loss": 0.475,
      "step": 14460
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0217317342758179,
      "learning_rate": 1.831218298527343e-07,
      "loss": 0.496,
      "step": 14461
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.089906096458435,
      "learning_rate": 1.828468053871213e-07,
      "loss": 0.5353,
      "step": 14462
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0170221328735352,
      "learning_rate": 1.825719837550727e-07,
      "loss": 0.4988,
      "step": 14463
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0263185501098633,
      "learning_rate": 1.822973649681592e-07,
      "loss": 0.4907,
      "step": 14464
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.040390133857727,
      "learning_rate": 1.820229490379438e-07,
      "loss": 0.4174,
      "step": 14465
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.092755675315857,
      "learning_rate": 1.8174873597598176e-07,
      "loss": 0.4812,
      "step": 14466
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0939713716506958,
      "learning_rate": 1.814747257938182e-07,
      "loss": 0.5367,
      "step": 14467
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0125434398651123,
      "learning_rate": 1.8120091850299225e-07,
      "loss": 0.5075,
      "step": 14468
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0190476179122925,
      "learning_rate": 1.809273141150325e-07,
      "loss": 0.4582,
      "step": 14469
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0586004257202148,
      "learning_rate": 1.8065391264145805e-07,
      "loss": 0.4978,
      "step": 14470
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0722798109054565,
      "learning_rate": 1.8038071409378299e-07,
      "loss": 0.5564,
      "step": 14471
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0449334383010864,
      "learning_rate": 1.8010771848350983e-07,
      "loss": 0.5336,
      "step": 14472
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0054998397827148,
      "learning_rate": 1.7983492582213324e-07,
      "loss": 0.502,
      "step": 14473
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1076698303222656,
      "learning_rate": 1.7956233612114017e-07,
      "loss": 0.4899,
      "step": 14474
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.04656183719635,
      "learning_rate": 1.792899493920075e-07,
      "loss": 0.4579,
      "step": 14475
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1003092527389526,
      "learning_rate": 1.79017765646205e-07,
      "loss": 0.5395,
      "step": 14476
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9659669399261475,
      "learning_rate": 1.78745784895194e-07,
      "loss": 0.4773,
      "step": 14477
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1086894273757935,
      "learning_rate": 1.7847400715042594e-07,
      "loss": 0.5042,
      "step": 14478
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.077738881111145,
      "learning_rate": 1.7820243242334334e-07,
      "loss": 0.5238,
      "step": 14479
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0929759740829468,
      "learning_rate": 1.7793106072538423e-07,
      "loss": 0.496,
      "step": 14480
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0933939218521118,
      "learning_rate": 1.7765989206797285e-07,
      "loss": 0.525,
      "step": 14481
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2027020454406738,
      "learning_rate": 1.7738892646252726e-07,
      "loss": 0.5044,
      "step": 14482
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0866278409957886,
      "learning_rate": 1.7711816392045778e-07,
      "loss": 0.51,
      "step": 14483
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0649842023849487,
      "learning_rate": 1.7684760445316418e-07,
      "loss": 0.4774,
      "step": 14484
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0627365112304688,
      "learning_rate": 1.765772480720407e-07,
      "loss": 0.4744,
      "step": 14485
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0996586084365845,
      "learning_rate": 1.763070947884693e-07,
      "loss": 0.5037,
      "step": 14486
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0557044744491577,
      "learning_rate": 1.7603714461382481e-07,
      "loss": 0.5296,
      "step": 14487
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9884737133979797,
      "learning_rate": 1.7576739755947593e-07,
      "loss": 0.4505,
      "step": 14488
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9897318482398987,
      "learning_rate": 1.7549785363677906e-07,
      "loss": 0.4991,
      "step": 14489
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0085302591323853,
      "learning_rate": 1.7522851285708465e-07,
      "loss": 0.5155,
      "step": 14490
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0740585327148438,
      "learning_rate": 1.7495937523173356e-07,
      "loss": 0.4794,
      "step": 14491
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0635310411453247,
      "learning_rate": 1.7469044077205732e-07,
      "loss": 0.4955,
      "step": 14492
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.020662546157837,
      "learning_rate": 1.744217094893813e-07,
      "loss": 0.5138,
      "step": 14493
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0481855869293213,
      "learning_rate": 1.7415318139502036e-07,
      "loss": 0.4705,
      "step": 14494
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.105790376663208,
      "learning_rate": 1.7388485650028043e-07,
      "loss": 0.5045,
      "step": 14495
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1326957941055298,
      "learning_rate": 1.7361673481646025e-07,
      "loss": 0.5019,
      "step": 14496
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2094613313674927,
      "learning_rate": 1.7334881635485023e-07,
      "loss": 0.492,
      "step": 14497
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1281813383102417,
      "learning_rate": 1.7308110112673027e-07,
      "loss": 0.5219,
      "step": 14498
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0920004844665527,
      "learning_rate": 1.7281358914337408e-07,
      "loss": 0.5141,
      "step": 14499
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1044858694076538,
      "learning_rate": 1.7254628041604437e-07,
      "loss": 0.5035,
      "step": 14500
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0102866888046265,
      "learning_rate": 1.7227917495599823e-07,
      "loss": 0.5135,
      "step": 14501
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2572340965270996,
      "learning_rate": 1.7201227277448108e-07,
      "loss": 0.6066,
      "step": 14502
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1052078008651733,
      "learning_rate": 1.7174557388273173e-07,
      "loss": 0.5146,
      "step": 14503
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1050742864608765,
      "learning_rate": 1.7147907829198008e-07,
      "loss": 0.4964,
      "step": 14504
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1257394552230835,
      "learning_rate": 1.7121278601344715e-07,
      "loss": 0.5344,
      "step": 14505
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.069999098777771,
      "learning_rate": 1.7094669705834566e-07,
      "loss": 0.501,
      "step": 14506
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.023105502128601,
      "learning_rate": 1.706808114378805e-07,
      "loss": 0.4674,
      "step": 14507
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0694199800491333,
      "learning_rate": 1.7041512916324554e-07,
      "loss": 0.4259,
      "step": 14508
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9969576001167297,
      "learning_rate": 1.7014965024562846e-07,
      "loss": 0.4873,
      "step": 14509
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0783843994140625,
      "learning_rate": 1.698843746962081e-07,
      "loss": 0.5172,
      "step": 14510
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0928618907928467,
      "learning_rate": 1.6961930252615388e-07,
      "loss": 0.4765,
      "step": 14511
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0058009624481201,
      "learning_rate": 1.6935443374662741e-07,
      "loss": 0.4716,
      "step": 14512
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.999502956867218,
      "learning_rate": 1.6908976836878088e-07,
      "loss": 0.4922,
      "step": 14513
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1754869222640991,
      "learning_rate": 1.6882530640375872e-07,
      "loss": 0.4827,
      "step": 14514
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1096594333648682,
      "learning_rate": 1.68561047862697e-07,
      "loss": 0.4981,
      "step": 14515
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6716039180755615,
      "learning_rate": 1.6829699275672186e-07,
      "loss": 0.5013,
      "step": 14516
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1831146478652954,
      "learning_rate": 1.6803314109695157e-07,
      "loss": 0.5198,
      "step": 14517
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.009282112121582,
      "learning_rate": 1.677694928944973e-07,
      "loss": 0.5035,
      "step": 14518
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0758283138275146,
      "learning_rate": 1.6750604816045902e-07,
      "loss": 0.4773,
      "step": 14519
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0731827020645142,
      "learning_rate": 1.6724280690593008e-07,
      "loss": 0.506,
      "step": 14520
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0044021606445312,
      "learning_rate": 1.6697976914199497e-07,
      "loss": 0.4654,
      "step": 14521
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0994043350219727,
      "learning_rate": 1.6671693487972818e-07,
      "loss": 0.4649,
      "step": 14522
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.109891653060913,
      "learning_rate": 1.6645430413019858e-07,
      "loss": 0.5406,
      "step": 14523
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.03861403465271,
      "learning_rate": 1.6619187690446293e-07,
      "loss": 0.5204,
      "step": 14524
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0034364461898804,
      "learning_rate": 1.659296532135718e-07,
      "loss": 0.4742,
      "step": 14525
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1274032592773438,
      "learning_rate": 1.6566763306856638e-07,
      "loss": 0.5234,
      "step": 14526
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0752556324005127,
      "learning_rate": 1.6540581648048003e-07,
      "loss": 0.5089,
      "step": 14527
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0327403545379639,
      "learning_rate": 1.6514420346033565e-07,
      "loss": 0.5078,
      "step": 14528
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0728938579559326,
      "learning_rate": 1.6488279401915052e-07,
      "loss": 0.5332,
      "step": 14529
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0218722820281982,
      "learning_rate": 1.6462158816792973e-07,
      "loss": 0.5185,
      "step": 14530
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.051649808883667,
      "learning_rate": 1.643605859176739e-07,
      "loss": 0.4676,
      "step": 14531
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9655593633651733,
      "learning_rate": 1.6409978727937094e-07,
      "loss": 0.4628,
      "step": 14532
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1105687618255615,
      "learning_rate": 1.6383919226400368e-07,
      "loss": 0.5038,
      "step": 14533
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0202934741973877,
      "learning_rate": 1.6357880088254396e-07,
      "loss": 0.4905,
      "step": 14534
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1168105602264404,
      "learning_rate": 1.633186131459563e-07,
      "loss": 0.4577,
      "step": 14535
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1130887269973755,
      "learning_rate": 1.6305862906519587e-07,
      "loss": 0.4963,
      "step": 14536
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.187774658203125,
      "learning_rate": 1.6279884865121108e-07,
      "loss": 0.5196,
      "step": 14537
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1031478643417358,
      "learning_rate": 1.6253927191493879e-07,
      "loss": 0.5342,
      "step": 14538
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9925063848495483,
      "learning_rate": 1.622798988673091e-07,
      "loss": 0.4971,
      "step": 14539
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1163438558578491,
      "learning_rate": 1.6202072951924386e-07,
      "loss": 0.4807,
      "step": 14540
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0186446905136108,
      "learning_rate": 1.6176176388165598e-07,
      "loss": 0.5142,
      "step": 14541
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9980571866035461,
      "learning_rate": 1.6150300196544955e-07,
      "loss": 0.4519,
      "step": 14542
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1521648168563843,
      "learning_rate": 1.612444437815186e-07,
      "loss": 0.4882,
      "step": 14543
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9618294835090637,
      "learning_rate": 1.6098608934075166e-07,
      "loss": 0.4419,
      "step": 14544
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0831882953643799,
      "learning_rate": 1.607279386540278e-07,
      "loss": 0.531,
      "step": 14545
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0838994979858398,
      "learning_rate": 1.60469991732215e-07,
      "loss": 0.4938,
      "step": 14546
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1143907308578491,
      "learning_rate": 1.6021224858617513e-07,
      "loss": 0.4969,
      "step": 14547
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9926535487174988,
      "learning_rate": 1.5995470922676116e-07,
      "loss": 0.5519,
      "step": 14548
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0101168155670166,
      "learning_rate": 1.5969737366481774e-07,
      "loss": 0.4652,
      "step": 14549
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1262658834457397,
      "learning_rate": 1.5944024191117958e-07,
      "loss": 0.5355,
      "step": 14550
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.187333106994629,
      "learning_rate": 1.5918331397667298e-07,
      "loss": 0.4975,
      "step": 14551
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0564457178115845,
      "learning_rate": 1.589265898721176e-07,
      "loss": 0.4909,
      "step": 14552
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0846610069274902,
      "learning_rate": 1.586700696083232e-07,
      "loss": 0.5091,
      "step": 14553
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1003252267837524,
      "learning_rate": 1.5841375319608943e-07,
      "loss": 0.4942,
      "step": 14554
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0836139917373657,
      "learning_rate": 1.5815764064621043e-07,
      "loss": 0.4812,
      "step": 14555
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1072503328323364,
      "learning_rate": 1.5790173196946924e-07,
      "loss": 0.4757,
      "step": 14556
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.140049695968628,
      "learning_rate": 1.5764602717664224e-07,
      "loss": 0.4893,
      "step": 14557
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0675603151321411,
      "learning_rate": 1.5739052627849581e-07,
      "loss": 0.4685,
      "step": 14558
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1174269914627075,
      "learning_rate": 1.571352292857875e-07,
      "loss": 0.4534,
      "step": 14559
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1924841403961182,
      "learning_rate": 1.5688013620926757e-07,
      "loss": 0.51,
      "step": 14560
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0370993614196777,
      "learning_rate": 1.566252470596774e-07,
      "loss": 0.4771,
      "step": 14561
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0299807786941528,
      "learning_rate": 1.5637056184774958e-07,
      "loss": 0.5161,
      "step": 14562
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.042910099029541,
      "learning_rate": 1.5611608058420714e-07,
      "loss": 0.4819,
      "step": 14563
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.017249584197998,
      "learning_rate": 1.5586180327976598e-07,
      "loss": 0.4749,
      "step": 14564
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.109641194343567,
      "learning_rate": 1.5560772994513251e-07,
      "loss": 0.508,
      "step": 14565
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9985325932502747,
      "learning_rate": 1.55353860591006e-07,
      "loss": 0.4372,
      "step": 14566
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0511524677276611,
      "learning_rate": 1.5510019522807397e-07,
      "loss": 0.4739,
      "step": 14567
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0950795412063599,
      "learning_rate": 1.5484673386701953e-07,
      "loss": 0.5365,
      "step": 14568
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0569123029708862,
      "learning_rate": 1.545934765185131e-07,
      "loss": 0.4913,
      "step": 14569
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0184240341186523,
      "learning_rate": 1.5434042319321996e-07,
      "loss": 0.4783,
      "step": 14570
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0293238162994385,
      "learning_rate": 1.5408757390179496e-07,
      "loss": 0.4696,
      "step": 14571
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0004197359085083,
      "learning_rate": 1.5383492865488459e-07,
      "loss": 0.498,
      "step": 14572
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0610783100128174,
      "learning_rate": 1.5358248746312588e-07,
      "loss": 0.517,
      "step": 14573
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0828512907028198,
      "learning_rate": 1.533302503371503e-07,
      "loss": 0.4697,
      "step": 14574
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1009697914123535,
      "learning_rate": 1.5307821728757722e-07,
      "loss": 0.5079,
      "step": 14575
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1081169843673706,
      "learning_rate": 1.5282638832501917e-07,
      "loss": 0.5489,
      "step": 14576
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.023945689201355,
      "learning_rate": 1.5257476346007938e-07,
      "loss": 0.4622,
      "step": 14577
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0600881576538086,
      "learning_rate": 1.523233427033538e-07,
      "loss": 0.487,
      "step": 14578
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0859774351119995,
      "learning_rate": 1.5207212606542786e-07,
      "loss": 0.5268,
      "step": 14579
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9725646376609802,
      "learning_rate": 1.518211135568809e-07,
      "loss": 0.4515,
      "step": 14580
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0840494632720947,
      "learning_rate": 1.5157030518828054e-07,
      "loss": 0.5228,
      "step": 14581
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0530489683151245,
      "learning_rate": 1.513197009701889e-07,
      "loss": 0.4621,
      "step": 14582
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0261887311935425,
      "learning_rate": 1.510693009131564e-07,
      "loss": 0.5065,
      "step": 14583
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0902878046035767,
      "learning_rate": 1.50819105027728e-07,
      "loss": 0.4306,
      "step": 14584
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.032993197441101,
      "learning_rate": 1.5056911332443801e-07,
      "loss": 0.5454,
      "step": 14585
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0643889904022217,
      "learning_rate": 1.5031932581381247e-07,
      "loss": 0.524,
      "step": 14586
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0280283689498901,
      "learning_rate": 1.5006974250636906e-07,
      "loss": 0.5156,
      "step": 14587
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.021931767463684,
      "learning_rate": 1.498203634126183e-07,
      "loss": 0.4612,
      "step": 14588
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.01459801197052,
      "learning_rate": 1.4957118854305842e-07,
      "loss": 0.5029,
      "step": 14589
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0916061401367188,
      "learning_rate": 1.4932221790818268e-07,
      "loss": 0.5149,
      "step": 14590
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.045670509338379,
      "learning_rate": 1.4907345151847387e-07,
      "loss": 0.511,
      "step": 14591
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0131471157073975,
      "learning_rate": 1.4882488938440688e-07,
      "loss": 0.4715,
      "step": 14592
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.168458104133606,
      "learning_rate": 1.485765315164478e-07,
      "loss": 0.5041,
      "step": 14593
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0689053535461426,
      "learning_rate": 1.483283779250544e-07,
      "loss": 0.4845,
      "step": 14594
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2584795951843262,
      "learning_rate": 1.4808042862067496e-07,
      "loss": 0.5169,
      "step": 14595
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0628129243850708,
      "learning_rate": 1.4783268361375058e-07,
      "loss": 0.5561,
      "step": 14596
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0348163843154907,
      "learning_rate": 1.4758514291471238e-07,
      "loss": 0.4674,
      "step": 14597
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0075467824935913,
      "learning_rate": 1.4733780653398254e-07,
      "loss": 0.4785,
      "step": 14598
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1389700174331665,
      "learning_rate": 1.4709067448197722e-07,
      "loss": 0.5509,
      "step": 14599
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0486923456192017,
      "learning_rate": 1.4684374676910197e-07,
      "loss": 0.5141,
      "step": 14600
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.038835883140564,
      "learning_rate": 1.4659702340575287e-07,
      "loss": 0.4996,
      "step": 14601
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9914584159851074,
      "learning_rate": 1.4635050440232002e-07,
      "loss": 0.461,
      "step": 14602
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0763351917266846,
      "learning_rate": 1.4610418976918172e-07,
      "loss": 0.5136,
      "step": 14603
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1267943382263184,
      "learning_rate": 1.4585807951671194e-07,
      "loss": 0.5065,
      "step": 14604
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0288209915161133,
      "learning_rate": 1.4561217365527124e-07,
      "loss": 0.5359,
      "step": 14605
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1225999593734741,
      "learning_rate": 1.453664721952147e-07,
      "loss": 0.4836,
      "step": 14606
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0499444007873535,
      "learning_rate": 1.451209751468885e-07,
      "loss": 0.5059,
      "step": 14607
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0559086799621582,
      "learning_rate": 1.448756825206288e-07,
      "loss": 0.4934,
      "step": 14608
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1914482116699219,
      "learning_rate": 1.4463059432676395e-07,
      "loss": 0.5341,
      "step": 14609
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.074084997177124,
      "learning_rate": 1.4438571057561523e-07,
      "loss": 0.4574,
      "step": 14610
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0202916860580444,
      "learning_rate": 1.4414103127749157e-07,
      "loss": 0.4697,
      "step": 14611
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0890121459960938,
      "learning_rate": 1.4389655644269752e-07,
      "loss": 0.5635,
      "step": 14612
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1268235445022583,
      "learning_rate": 1.4365228608152647e-07,
      "loss": 0.5436,
      "step": 14613
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.097791075706482,
      "learning_rate": 1.4340822020426304e-07,
      "loss": 0.5227,
      "step": 14614
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1117488145828247,
      "learning_rate": 1.4316435882118563e-07,
      "loss": 0.5534,
      "step": 14615
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0937881469726562,
      "learning_rate": 1.429207019425599e-07,
      "loss": 0.5099,
      "step": 14616
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0312176942825317,
      "learning_rate": 1.426772495786477e-07,
      "loss": 0.4859,
      "step": 14617
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1102279424667358,
      "learning_rate": 1.4243400173969968e-07,
      "loss": 0.4672,
      "step": 14618
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0215271711349487,
      "learning_rate": 1.4219095843595654e-07,
      "loss": 0.4931,
      "step": 14619
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.036686658859253,
      "learning_rate": 1.4194811967765344e-07,
      "loss": 0.4831,
      "step": 14620
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.077612042427063,
      "learning_rate": 1.417054854750155e-07,
      "loss": 0.4795,
      "step": 14621
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0578806400299072,
      "learning_rate": 1.414630558382579e-07,
      "loss": 0.4379,
      "step": 14622
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1002401113510132,
      "learning_rate": 1.4122083077759087e-07,
      "loss": 0.4855,
      "step": 14623
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0033938884735107,
      "learning_rate": 1.409788103032106e-07,
      "loss": 0.4872,
      "step": 14624
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0382969379425049,
      "learning_rate": 1.4073699442531007e-07,
      "loss": 0.4959,
      "step": 14625
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0288439989089966,
      "learning_rate": 1.4049538315407064e-07,
      "loss": 0.4967,
      "step": 14626
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0696370601654053,
      "learning_rate": 1.4025397649966577e-07,
      "loss": 0.4914,
      "step": 14627
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0011566877365112,
      "learning_rate": 1.400127744722596e-07,
      "loss": 0.501,
      "step": 14628
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.134408950805664,
      "learning_rate": 1.3977177708200896e-07,
      "loss": 0.5081,
      "step": 14629
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0760116577148438,
      "learning_rate": 1.395309843390613e-07,
      "loss": 0.506,
      "step": 14630
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0447229146957397,
      "learning_rate": 1.3929039625355633e-07,
      "loss": 0.5096,
      "step": 14631
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1114329099655151,
      "learning_rate": 1.3905001283562257e-07,
      "loss": 0.5185,
      "step": 14632
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0503195524215698,
      "learning_rate": 1.3880983409538252e-07,
      "loss": 0.4906,
      "step": 14633
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1024737358093262,
      "learning_rate": 1.3856986004295082e-07,
      "loss": 0.4867,
      "step": 14634
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0304983854293823,
      "learning_rate": 1.3833009068842995e-07,
      "loss": 0.4899,
      "step": 14635
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0985244512557983,
      "learning_rate": 1.3809052604191632e-07,
      "loss": 0.5125,
      "step": 14636
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1270508766174316,
      "learning_rate": 1.3785116611349736e-07,
      "loss": 0.5057,
      "step": 14637
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1066454648971558,
      "learning_rate": 1.3761201091325172e-07,
      "loss": 0.5184,
      "step": 14638
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0407905578613281,
      "learning_rate": 1.3737306045124966e-07,
      "loss": 0.5327,
      "step": 14639
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1179550886154175,
      "learning_rate": 1.3713431473755147e-07,
      "loss": 0.4996,
      "step": 14640
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1324025392532349,
      "learning_rate": 1.3689577378221019e-07,
      "loss": 0.5664,
      "step": 14641
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9954050183296204,
      "learning_rate": 1.3665743759527173e-07,
      "loss": 0.449,
      "step": 14642
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1661484241485596,
      "learning_rate": 1.3641930618676912e-07,
      "loss": 0.5231,
      "step": 14643
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1205106973648071,
      "learning_rate": 1.3618137956673105e-07,
      "loss": 0.4912,
      "step": 14644
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1276148557662964,
      "learning_rate": 1.3594365774517447e-07,
      "loss": 0.5189,
      "step": 14645
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1023553609848022,
      "learning_rate": 1.357061407321103e-07,
      "loss": 0.477,
      "step": 14646
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1168313026428223,
      "learning_rate": 1.3546882853753885e-07,
      "loss": 0.5251,
      "step": 14647
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9755619764328003,
      "learning_rate": 1.3523172117145212e-07,
      "loss": 0.5231,
      "step": 14648
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0576835870742798,
      "learning_rate": 1.349948186438349e-07,
      "loss": 0.498,
      "step": 14649
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0488373041152954,
      "learning_rate": 1.347581209646609e-07,
      "loss": 0.5153,
      "step": 14650
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.083781361579895,
      "learning_rate": 1.3452162814389824e-07,
      "loss": 0.4722,
      "step": 14651
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0556163787841797,
      "learning_rate": 1.342853401915034e-07,
      "loss": 0.5482,
      "step": 14652
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.03798246383667,
      "learning_rate": 1.3404925711742734e-07,
      "loss": 0.539,
      "step": 14653
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.024129033088684,
      "learning_rate": 1.3381337893160818e-07,
      "loss": 0.4782,
      "step": 14654
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0649831295013428,
      "learning_rate": 1.3357770564398075e-07,
      "loss": 0.4895,
      "step": 14655
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0626680850982666,
      "learning_rate": 1.333422372644666e-07,
      "loss": 0.5219,
      "step": 14656
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.059069275856018,
      "learning_rate": 1.331069738029811e-07,
      "loss": 0.4972,
      "step": 14657
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0212222337722778,
      "learning_rate": 1.3287191526942968e-07,
      "loss": 0.5299,
      "step": 14658
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0332025289535522,
      "learning_rate": 1.3263706167371104e-07,
      "loss": 0.4738,
      "step": 14659
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.107440710067749,
      "learning_rate": 1.324024130257129e-07,
      "loss": 0.534,
      "step": 14660
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0443707704544067,
      "learning_rate": 1.3216796933531672e-07,
      "loss": 0.5026,
      "step": 14661
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9763154983520508,
      "learning_rate": 1.31933730612393e-07,
      "loss": 0.5172,
      "step": 14662
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0752264261245728,
      "learning_rate": 1.316996968668044e-07,
      "loss": 0.5152,
      "step": 14663
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.047065258026123,
      "learning_rate": 1.3146586810840745e-07,
      "loss": 0.4737,
      "step": 14664
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1606717109680176,
      "learning_rate": 1.312322443470454e-07,
      "loss": 0.5027,
      "step": 14665
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1153333187103271,
      "learning_rate": 1.309988255925565e-07,
      "loss": 0.5118,
      "step": 14666
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1025995016098022,
      "learning_rate": 1.30765611854769e-07,
      "loss": 0.5324,
      "step": 14667
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.097686529159546,
      "learning_rate": 1.305326031435028e-07,
      "loss": 0.5022,
      "step": 14668
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1420540809631348,
      "learning_rate": 1.3029979946856953e-07,
      "loss": 0.5704,
      "step": 14669
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.013661503791809,
      "learning_rate": 1.3006720083977076e-07,
      "loss": 0.4919,
      "step": 14670
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.01823091506958,
      "learning_rate": 1.2983480726690033e-07,
      "loss": 0.4721,
      "step": 14671
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.14744234085083,
      "learning_rate": 1.296026187597449e-07,
      "loss": 0.4951,
      "step": 14672
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.062508225440979,
      "learning_rate": 1.2937063532807992e-07,
      "loss": 0.4741,
      "step": 14673
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0481140613555908,
      "learning_rate": 1.2913885698167427e-07,
      "loss": 0.4931,
      "step": 14674
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1365739107131958,
      "learning_rate": 1.2890728373028626e-07,
      "loss": 0.4872,
      "step": 14675
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0365864038467407,
      "learning_rate": 1.2867591558366755e-07,
      "loss": 0.4568,
      "step": 14676
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0156171321868896,
      "learning_rate": 1.2844475255156087e-07,
      "loss": 0.4704,
      "step": 14677
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.044424295425415,
      "learning_rate": 1.2821379464369732e-07,
      "loss": 0.4823,
      "step": 14678
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0783231258392334,
      "learning_rate": 1.2798304186980358e-07,
      "loss": 0.5153,
      "step": 14679
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0163437128067017,
      "learning_rate": 1.277524942395958e-07,
      "loss": 0.4605,
      "step": 14680
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.115587592124939,
      "learning_rate": 1.275221517627806e-07,
      "loss": 0.5101,
      "step": 14681
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9934950470924377,
      "learning_rate": 1.2729201444905803e-07,
      "loss": 0.4888,
      "step": 14682
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0094565153121948,
      "learning_rate": 1.2706208230811812e-07,
      "loss": 0.4648,
      "step": 14683
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1033930778503418,
      "learning_rate": 1.2683235534964088e-07,
      "loss": 0.5416,
      "step": 14684
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0976402759552002,
      "learning_rate": 1.2660283358330195e-07,
      "loss": 0.4732,
      "step": 14685
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0931764841079712,
      "learning_rate": 1.263735170187641e-07,
      "loss": 0.5057,
      "step": 14686
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.050828456878662,
      "learning_rate": 1.26144405665683e-07,
      "loss": 0.4758,
      "step": 14687
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0917552709579468,
      "learning_rate": 1.2591549953370586e-07,
      "loss": 0.5101,
      "step": 14688
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0817075967788696,
      "learning_rate": 1.2568679863247168e-07,
      "loss": 0.5084,
      "step": 14689
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0188014507293701,
      "learning_rate": 1.2545830297161e-07,
      "loss": 0.4613,
      "step": 14690
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0029340982437134,
      "learning_rate": 1.2523001256074196e-07,
      "loss": 0.4473,
      "step": 14691
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.103075385093689,
      "learning_rate": 1.2500192740947936e-07,
      "loss": 0.49,
      "step": 14692
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0108256340026855,
      "learning_rate": 1.2477404752742784e-07,
      "loss": 0.4927,
      "step": 14693
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1069060564041138,
      "learning_rate": 1.2454637292418082e-07,
      "loss": 0.4983,
      "step": 14694
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1655943393707275,
      "learning_rate": 1.2431890360932507e-07,
      "loss": 0.5047,
      "step": 14695
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1198543310165405,
      "learning_rate": 1.2409163959244019e-07,
      "loss": 0.5091,
      "step": 14696
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.043695092201233,
      "learning_rate": 1.2386458088309296e-07,
      "loss": 0.4854,
      "step": 14697
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0993931293487549,
      "learning_rate": 1.2363772749084625e-07,
      "loss": 0.4677,
      "step": 14698
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0852168798446655,
      "learning_rate": 1.2341107942525132e-07,
      "loss": 0.4745,
      "step": 14699
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9771943092346191,
      "learning_rate": 1.2318463669585112e-07,
      "loss": 0.4683,
      "step": 14700
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.085769772529602,
      "learning_rate": 1.229583993121808e-07,
      "loss": 0.495,
      "step": 14701
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0052050352096558,
      "learning_rate": 1.2273236728376604e-07,
      "loss": 0.4935,
      "step": 14702
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0405563116073608,
      "learning_rate": 1.2250654062012478e-07,
      "loss": 0.4673,
      "step": 14703
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0291260480880737,
      "learning_rate": 1.2228091933076613e-07,
      "loss": 0.4609,
      "step": 14704
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9860481023788452,
      "learning_rate": 1.2205550342518803e-07,
      "loss": 0.4391,
      "step": 14705
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0832188129425049,
      "learning_rate": 1.2183029291288452e-07,
      "loss": 0.5244,
      "step": 14706
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0049821138381958,
      "learning_rate": 1.2160528780333803e-07,
      "loss": 0.4529,
      "step": 14707
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0232901573181152,
      "learning_rate": 1.2138048810602154e-07,
      "loss": 0.5217,
      "step": 14708
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0310624837875366,
      "learning_rate": 1.2115589383040083e-07,
      "loss": 0.5262,
      "step": 14709
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1270962953567505,
      "learning_rate": 1.2093150498593387e-07,
      "loss": 0.4924,
      "step": 14710
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0199156999588013,
      "learning_rate": 1.2070732158206754e-07,
      "loss": 0.5297,
      "step": 14711
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.102827548980713,
      "learning_rate": 1.2048334362824265e-07,
      "loss": 0.5266,
      "step": 14712
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.082411766052246,
      "learning_rate": 1.202595711338894e-07,
      "loss": 0.5163,
      "step": 14713
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9880638122558594,
      "learning_rate": 1.2003600410842974e-07,
      "loss": 0.4991,
      "step": 14714
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1178537607192993,
      "learning_rate": 1.1981264256127832e-07,
      "loss": 0.5123,
      "step": 14715
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0149600505828857,
      "learning_rate": 1.1958948650183988e-07,
      "loss": 0.516,
      "step": 14716
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.02154541015625,
      "learning_rate": 1.1936653593950964e-07,
      "loss": 0.4847,
      "step": 14717
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0899893045425415,
      "learning_rate": 1.1914379088367677e-07,
      "loss": 0.5178,
      "step": 14718
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0712817907333374,
      "learning_rate": 1.1892125134371935e-07,
      "loss": 0.4968,
      "step": 14719
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1291003227233887,
      "learning_rate": 1.1869891732900762e-07,
      "loss": 0.5253,
      "step": 14720
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0928550958633423,
      "learning_rate": 1.1847678884890467e-07,
      "loss": 0.4624,
      "step": 14721
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9802485108375549,
      "learning_rate": 1.1825486591276136e-07,
      "loss": 0.4553,
      "step": 14722
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1071562767028809,
      "learning_rate": 1.1803314852992409e-07,
      "loss": 0.5389,
      "step": 14723
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.029962182044983,
      "learning_rate": 1.1781163670972762e-07,
      "loss": 0.5059,
      "step": 14724
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1084882020950317,
      "learning_rate": 1.1759033046149948e-07,
      "loss": 0.5179,
      "step": 14725
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0458074808120728,
      "learning_rate": 1.1736922979455778e-07,
      "loss": 0.5318,
      "step": 14726
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1072486639022827,
      "learning_rate": 1.1714833471821175e-07,
      "loss": 0.5415,
      "step": 14727
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0224193334579468,
      "learning_rate": 1.1692764524176337e-07,
      "loss": 0.5223,
      "step": 14728
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.950425922870636,
      "learning_rate": 1.1670716137450577e-07,
      "loss": 0.4669,
      "step": 14729
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0135602951049805,
      "learning_rate": 1.1648688312572099e-07,
      "loss": 0.4715,
      "step": 14730
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1646565198898315,
      "learning_rate": 1.1626681050468492e-07,
      "loss": 0.5031,
      "step": 14731
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1194438934326172,
      "learning_rate": 1.1604694352066459e-07,
      "loss": 0.5512,
      "step": 14732
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9765666127204895,
      "learning_rate": 1.1582728218291761e-07,
      "loss": 0.4943,
      "step": 14733
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0588070154190063,
      "learning_rate": 1.1560782650069269e-07,
      "loss": 0.4678,
      "step": 14734
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0291880369186401,
      "learning_rate": 1.153885764832302e-07,
      "loss": 0.5237,
      "step": 14735
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0827213525772095,
      "learning_rate": 1.1516953213976278e-07,
      "loss": 0.5302,
      "step": 14736
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0477744340896606,
      "learning_rate": 1.1495069347951416e-07,
      "loss": 0.4948,
      "step": 14737
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0322061777114868,
      "learning_rate": 1.1473206051169694e-07,
      "loss": 0.5209,
      "step": 14738
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0588160753250122,
      "learning_rate": 1.1451363324551822e-07,
      "loss": 0.4953,
      "step": 14739
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0796961784362793,
      "learning_rate": 1.1429541169017511e-07,
      "loss": 0.4789,
      "step": 14740
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0964117050170898,
      "learning_rate": 1.1407739585485633e-07,
      "loss": 0.5103,
      "step": 14741
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1128816604614258,
      "learning_rate": 1.1385958574874178e-07,
      "loss": 0.5043,
      "step": 14742
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.173375129699707,
      "learning_rate": 1.1364198138100191e-07,
      "loss": 0.5545,
      "step": 14743
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2093064785003662,
      "learning_rate": 1.1342458276079937e-07,
      "loss": 0.5485,
      "step": 14744
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0547919273376465,
      "learning_rate": 1.1320738989728963e-07,
      "loss": 0.436,
      "step": 14745
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9337969422340393,
      "learning_rate": 1.1299040279961593e-07,
      "loss": 0.4476,
      "step": 14746
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.009351134300232,
      "learning_rate": 1.1277362147691595e-07,
      "loss": 0.4925,
      "step": 14747
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0779575109481812,
      "learning_rate": 1.125570459383174e-07,
      "loss": 0.5148,
      "step": 14748
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0640534162521362,
      "learning_rate": 1.1234067619293909e-07,
      "loss": 0.4839,
      "step": 14749
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1556848287582397,
      "learning_rate": 1.1212451224989262e-07,
      "loss": 0.4944,
      "step": 14750
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1237772703170776,
      "learning_rate": 1.1190855411827906e-07,
      "loss": 0.5018,
      "step": 14751
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.168222427368164,
      "learning_rate": 1.1169280180719111e-07,
      "loss": 0.4884,
      "step": 14752
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9854783415794373,
      "learning_rate": 1.114772553257154e-07,
      "loss": 0.4873,
      "step": 14753
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0684829950332642,
      "learning_rate": 1.1126191468292579e-07,
      "loss": 0.4655,
      "step": 14754
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0761710405349731,
      "learning_rate": 1.1104677988789004e-07,
      "loss": 0.5349,
      "step": 14755
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9973959922790527,
      "learning_rate": 1.1083185094966753e-07,
      "loss": 0.4533,
      "step": 14756
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.116524577140808,
      "learning_rate": 1.1061712787730716e-07,
      "loss": 0.5009,
      "step": 14757
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0338749885559082,
      "learning_rate": 1.1040261067985114e-07,
      "loss": 0.4979,
      "step": 14758
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0604186058044434,
      "learning_rate": 1.1018829936633113e-07,
      "loss": 0.5396,
      "step": 14759
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.136382818222046,
      "learning_rate": 1.0997419394577158e-07,
      "loss": 0.535,
      "step": 14760
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0905015468597412,
      "learning_rate": 1.0976029442718694e-07,
      "loss": 0.504,
      "step": 14761
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0957517623901367,
      "learning_rate": 1.0954660081958502e-07,
      "loss": 0.4676,
      "step": 14762
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0968356132507324,
      "learning_rate": 1.0933311313196304e-07,
      "loss": 0.5497,
      "step": 14763
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.05763840675354,
      "learning_rate": 1.091198313733105e-07,
      "loss": 0.4728,
      "step": 14764
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0490801334381104,
      "learning_rate": 1.0890675555260688e-07,
      "loss": 0.4904,
      "step": 14765
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0433591604232788,
      "learning_rate": 1.086938856788261e-07,
      "loss": 0.4838,
      "step": 14766
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1754895448684692,
      "learning_rate": 1.0848122176092935e-07,
      "loss": 0.4764,
      "step": 14767
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0383951663970947,
      "learning_rate": 1.0826876380787221e-07,
      "loss": 0.5165,
      "step": 14768
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.01365327835083,
      "learning_rate": 1.0805651182860033e-07,
      "loss": 0.4913,
      "step": 14769
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0312130451202393,
      "learning_rate": 1.0784446583205099e-07,
      "loss": 0.4835,
      "step": 14770
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.113490343093872,
      "learning_rate": 1.0763262582715206e-07,
      "loss": 0.5329,
      "step": 14771
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0257731676101685,
      "learning_rate": 1.0742099182282529e-07,
      "loss": 0.4546,
      "step": 14772
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0378912687301636,
      "learning_rate": 1.0720956382797965e-07,
      "loss": 0.4755,
      "step": 14773
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0855647325515747,
      "learning_rate": 1.0699834185151802e-07,
      "loss": 0.4862,
      "step": 14774
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.115774154663086,
      "learning_rate": 1.0678732590233553e-07,
      "loss": 0.5119,
      "step": 14775
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0004231929779053,
      "learning_rate": 1.0657651598931563e-07,
      "loss": 0.4612,
      "step": 14776
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0334473848342896,
      "learning_rate": 1.0636591212133673e-07,
      "loss": 0.4967,
      "step": 14777
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.09772527217865,
      "learning_rate": 1.0615551430726456e-07,
      "loss": 0.5046,
      "step": 14778
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9907475113868713,
      "learning_rate": 1.0594532255595979e-07,
      "loss": 0.4728,
      "step": 14779
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.056544542312622,
      "learning_rate": 1.0573533687627258e-07,
      "loss": 0.4913,
      "step": 14780
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9403951168060303,
      "learning_rate": 1.0552555727704417e-07,
      "loss": 0.4794,
      "step": 14781
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1103562116622925,
      "learning_rate": 1.053159837671075e-07,
      "loss": 0.5223,
      "step": 14782
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9928891062736511,
      "learning_rate": 1.0510661635528774e-07,
      "loss": 0.4647,
      "step": 14783
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.085282325744629,
      "learning_rate": 1.0489745505040006e-07,
      "loss": 0.4453,
      "step": 14784
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0276105403900146,
      "learning_rate": 1.0468849986125185e-07,
      "loss": 0.4977,
      "step": 14785
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0931756496429443,
      "learning_rate": 1.0447975079664163e-07,
      "loss": 0.4814,
      "step": 14786
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0745348930358887,
      "learning_rate": 1.042712078653585e-07,
      "loss": 0.5049,
      "step": 14787
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0469303131103516,
      "learning_rate": 1.0406287107618429e-07,
      "loss": 0.5248,
      "step": 14788
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0048182010650635,
      "learning_rate": 1.0385474043789034e-07,
      "loss": 0.4649,
      "step": 14789
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0827223062515259,
      "learning_rate": 1.0364681595924131e-07,
      "loss": 0.4582,
      "step": 14790
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9856858253479004,
      "learning_rate": 1.034390976489913e-07,
      "loss": 0.4624,
      "step": 14791
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9975650906562805,
      "learning_rate": 1.0323158551588663e-07,
      "loss": 0.4744,
      "step": 14792
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0474202632904053,
      "learning_rate": 1.030242795686659e-07,
      "loss": 0.4667,
      "step": 14793
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.112815022468567,
      "learning_rate": 1.0281717981605765e-07,
      "loss": 0.5151,
      "step": 14794
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0142254829406738,
      "learning_rate": 1.0261028626678104e-07,
      "loss": 0.4269,
      "step": 14795
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0286717414855957,
      "learning_rate": 1.024035989295491e-07,
      "loss": 0.5014,
      "step": 14796
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.137579083442688,
      "learning_rate": 1.0219711781306374e-07,
      "loss": 0.5317,
      "step": 14797
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1120980978012085,
      "learning_rate": 1.0199084292602024e-07,
      "loss": 0.4991,
      "step": 14798
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0727988481521606,
      "learning_rate": 1.0178477427710276e-07,
      "loss": 0.5524,
      "step": 14799
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0857231616973877,
      "learning_rate": 1.015789118749888e-07,
      "loss": 0.5149,
      "step": 14800
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.041313648223877,
      "learning_rate": 1.0137325572834644e-07,
      "loss": 0.4561,
      "step": 14801
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1011313199996948,
      "learning_rate": 1.0116780584583596e-07,
      "loss": 0.5172,
      "step": 14802
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0425204038619995,
      "learning_rate": 1.0096256223610657e-07,
      "loss": 0.4757,
      "step": 14803
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1371713876724243,
      "learning_rate": 1.0075752490780133e-07,
      "loss": 0.53,
      "step": 14804
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0720175504684448,
      "learning_rate": 1.0055269386955391e-07,
      "loss": 0.5051,
      "step": 14805
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.046457290649414,
      "learning_rate": 1.0034806912998796e-07,
      "loss": 0.5115,
      "step": 14806
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0416407585144043,
      "learning_rate": 1.0014365069772102e-07,
      "loss": 0.518,
      "step": 14807
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0310873985290527,
      "learning_rate": 9.993943858135846e-08,
      "loss": 0.5512,
      "step": 14808
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1090185642242432,
      "learning_rate": 9.973543278950115e-08,
      "loss": 0.4912,
      "step": 14809
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0710328817367554,
      "learning_rate": 9.953163333073779e-08,
      "loss": 0.5639,
      "step": 14810
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.053472876548767,
      "learning_rate": 9.932804021364928e-08,
      "loss": 0.5128,
      "step": 14811
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1337611675262451,
      "learning_rate": 9.912465344680933e-08,
      "loss": 0.51,
      "step": 14812
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9499417543411255,
      "learning_rate": 9.892147303878108e-08,
      "loss": 0.4073,
      "step": 14813
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0529181957244873,
      "learning_rate": 9.871849899811991e-08,
      "loss": 0.4939,
      "step": 14814
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0673210620880127,
      "learning_rate": 9.851573133337288e-08,
      "loss": 0.5408,
      "step": 14815
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0431410074234009,
      "learning_rate": 9.83131700530765e-08,
      "loss": 0.5126,
      "step": 14816
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1376043558120728,
      "learning_rate": 9.81108151657617e-08,
      "loss": 0.4993,
      "step": 14817
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1424436569213867,
      "learning_rate": 9.790866667994781e-08,
      "loss": 0.5119,
      "step": 14818
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.170620083808899,
      "learning_rate": 9.770672460414688e-08,
      "loss": 0.5726,
      "step": 14819
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0776851177215576,
      "learning_rate": 9.750498894686156e-08,
      "loss": 0.5107,
      "step": 14820
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0550538301467896,
      "learning_rate": 9.730345971658728e-08,
      "loss": 0.4677,
      "step": 14821
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0215625762939453,
      "learning_rate": 9.710213692180836e-08,
      "loss": 0.5126,
      "step": 14822
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1082737445831299,
      "learning_rate": 9.690102057100304e-08,
      "loss": 0.5123,
      "step": 14823
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.137060284614563,
      "learning_rate": 9.670011067263896e-08,
      "loss": 0.4662,
      "step": 14824
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0273690223693848,
      "learning_rate": 9.649940723517549e-08,
      "loss": 0.4731,
      "step": 14825
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1005918979644775,
      "learning_rate": 9.629891026706472e-08,
      "loss": 0.455,
      "step": 14826
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0610016584396362,
      "learning_rate": 9.609861977674773e-08,
      "loss": 0.4533,
      "step": 14827
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9681522846221924,
      "learning_rate": 9.589853577265829e-08,
      "loss": 0.4696,
      "step": 14828
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0362496376037598,
      "learning_rate": 9.569865826322133e-08,
      "loss": 0.5161,
      "step": 14829
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0674338340759277,
      "learning_rate": 9.549898725685291e-08,
      "loss": 0.4986,
      "step": 14830
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0913596153259277,
      "learning_rate": 9.52995227619613e-08,
      "loss": 0.4844,
      "step": 14831
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0585317611694336,
      "learning_rate": 9.510026478694423e-08,
      "loss": 0.4917,
      "step": 14832
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9431830048561096,
      "learning_rate": 9.49012133401922e-08,
      "loss": 0.5156,
      "step": 14833
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1033228635787964,
      "learning_rate": 9.47023684300863e-08,
      "loss": 0.5279,
      "step": 14834
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1913155317306519,
      "learning_rate": 9.450373006499924e-08,
      "loss": 0.4868,
      "step": 14835
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1376891136169434,
      "learning_rate": 9.430529825329492e-08,
      "loss": 0.4777,
      "step": 14836
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0231413841247559,
      "learning_rate": 9.410707300333e-08,
      "loss": 0.5451,
      "step": 14837
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0114684104919434,
      "learning_rate": 9.390905432344833e-08,
      "loss": 0.4916,
      "step": 14838
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1515475511550903,
      "learning_rate": 9.371124222199046e-08,
      "loss": 0.4936,
      "step": 14839
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1124534606933594,
      "learning_rate": 9.35136367072842e-08,
      "loss": 0.4832,
      "step": 14840
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0438929796218872,
      "learning_rate": 9.331623778765009e-08,
      "loss": 0.4578,
      "step": 14841
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1023262739181519,
      "learning_rate": 9.311904547139982e-08,
      "loss": 0.4794,
      "step": 14842
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.084631085395813,
      "learning_rate": 9.292205976683733e-08,
      "loss": 0.5164,
      "step": 14843
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.036280870437622,
      "learning_rate": 9.272528068225595e-08,
      "loss": 0.5072,
      "step": 14844
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.027374029159546,
      "learning_rate": 9.252870822594239e-08,
      "loss": 0.5063,
      "step": 14845
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0010735988616943,
      "learning_rate": 9.233234240617228e-08,
      "loss": 0.4835,
      "step": 14846
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0823222398757935,
      "learning_rate": 9.213618323121564e-08,
      "loss": 0.5167,
      "step": 14847
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0411672592163086,
      "learning_rate": 9.19402307093309e-08,
      "loss": 0.4888,
      "step": 14848
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0619031190872192,
      "learning_rate": 9.174448484876864e-08,
      "loss": 0.4686,
      "step": 14849
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9982632398605347,
      "learning_rate": 9.154894565777173e-08,
      "loss": 0.4258,
      "step": 14850
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.087734580039978,
      "learning_rate": 9.135361314457358e-08,
      "loss": 0.5276,
      "step": 14851
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9884198307991028,
      "learning_rate": 9.115848731739874e-08,
      "loss": 0.502,
      "step": 14852
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1553099155426025,
      "learning_rate": 9.096356818446395e-08,
      "loss": 0.5425,
      "step": 14853
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0436712503433228,
      "learning_rate": 9.076885575397543e-08,
      "loss": 0.4802,
      "step": 14854
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1476726531982422,
      "learning_rate": 9.057435003413273e-08,
      "loss": 0.5413,
      "step": 14855
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.073279619216919,
      "learning_rate": 9.038005103312486e-08,
      "loss": 0.5065,
      "step": 14856
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0361818075180054,
      "learning_rate": 9.018595875913416e-08,
      "loss": 0.475,
      "step": 14857
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9552186727523804,
      "learning_rate": 8.999207322033299e-08,
      "loss": 0.4886,
      "step": 14858
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0540318489074707,
      "learning_rate": 8.979839442488425e-08,
      "loss": 0.5102,
      "step": 14859
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0041334629058838,
      "learning_rate": 8.960492238094421e-08,
      "loss": 0.4447,
      "step": 14860
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0236811637878418,
      "learning_rate": 8.941165709665966e-08,
      "loss": 0.5233,
      "step": 14861
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0665013790130615,
      "learning_rate": 8.921859858016635e-08,
      "loss": 0.4775,
      "step": 14862
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0712215900421143,
      "learning_rate": 8.902574683959442e-08,
      "loss": 0.4774,
      "step": 14863
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0452264547348022,
      "learning_rate": 8.883310188306515e-08,
      "loss": 0.4736,
      "step": 14864
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2021039724349976,
      "learning_rate": 8.864066371868873e-08,
      "loss": 0.5335,
      "step": 14865
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1182342767715454,
      "learning_rate": 8.844843235456868e-08,
      "loss": 0.5135,
      "step": 14866
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0686935186386108,
      "learning_rate": 8.825640779879962e-08,
      "loss": 0.446,
      "step": 14867
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.956581711769104,
      "learning_rate": 8.806459005946565e-08,
      "loss": 0.5015,
      "step": 14868
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1133993864059448,
      "learning_rate": 8.787297914464533e-08,
      "loss": 0.4799,
      "step": 14869
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.090657114982605,
      "learning_rate": 8.768157506240494e-08,
      "loss": 0.5349,
      "step": 14870
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0605127811431885,
      "learning_rate": 8.749037782080528e-08,
      "loss": 0.5265,
      "step": 14871
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0305068492889404,
      "learning_rate": 8.729938742789601e-08,
      "loss": 0.4894,
      "step": 14872
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0121204853057861,
      "learning_rate": 8.71086038917196e-08,
      "loss": 0.5261,
      "step": 14873
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0988103151321411,
      "learning_rate": 8.691802722030906e-08,
      "loss": 0.5069,
      "step": 14874
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0493741035461426,
      "learning_rate": 8.672765742168964e-08,
      "loss": 0.4722,
      "step": 14875
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1380290985107422,
      "learning_rate": 8.65374945038755e-08,
      "loss": 0.492,
      "step": 14876
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.029023289680481,
      "learning_rate": 8.634753847487575e-08,
      "loss": 0.5584,
      "step": 14877
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0613927841186523,
      "learning_rate": 8.615778934268793e-08,
      "loss": 0.5209,
      "step": 14878
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0069340467453003,
      "learning_rate": 8.59682471153006e-08,
      "loss": 0.4645,
      "step": 14879
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9931952953338623,
      "learning_rate": 8.577891180069687e-08,
      "loss": 0.4564,
      "step": 14880
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0470446348190308,
      "learning_rate": 8.558978340684642e-08,
      "loss": 0.4986,
      "step": 14881
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.008948802947998,
      "learning_rate": 8.540086194171515e-08,
      "loss": 0.5117,
      "step": 14882
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0242713689804077,
      "learning_rate": 8.521214741325722e-08,
      "loss": 0.5012,
      "step": 14883
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0157408714294434,
      "learning_rate": 8.502363982941797e-08,
      "loss": 0.454,
      "step": 14884
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.059425711631775,
      "learning_rate": 8.483533919813546e-08,
      "loss": 0.5074,
      "step": 14885
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1020406484603882,
      "learning_rate": 8.464724552733782e-08,
      "loss": 0.5256,
      "step": 14886
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0872770547866821,
      "learning_rate": 8.445935882494593e-08,
      "loss": 0.484,
      "step": 14887
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1084372997283936,
      "learning_rate": 8.427167909887069e-08,
      "loss": 0.4639,
      "step": 14888
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.215596318244934,
      "learning_rate": 8.408420635701353e-08,
      "loss": 0.4957,
      "step": 14889
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0778475999832153,
      "learning_rate": 8.389694060726927e-08,
      "loss": 0.5289,
      "step": 14890
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1347565650939941,
      "learning_rate": 8.370988185752383e-08,
      "loss": 0.551,
      "step": 14891
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0359888076782227,
      "learning_rate": 8.352303011565254e-08,
      "loss": 0.5024,
      "step": 14892
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0887922048568726,
      "learning_rate": 8.333638538952305e-08,
      "loss": 0.5033,
      "step": 14893
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0436344146728516,
      "learning_rate": 8.314994768699458e-08,
      "loss": 0.511,
      "step": 14894
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0825071334838867,
      "learning_rate": 8.296371701591699e-08,
      "loss": 0.4679,
      "step": 14895
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.039556860923767,
      "learning_rate": 8.277769338413288e-08,
      "loss": 0.5195,
      "step": 14896
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0870271921157837,
      "learning_rate": 8.259187679947434e-08,
      "loss": 0.489,
      "step": 14897
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.037808895111084,
      "learning_rate": 8.240626726976453e-08,
      "loss": 0.493,
      "step": 14898
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1069194078445435,
      "learning_rate": 8.222086480282054e-08,
      "loss": 0.4903,
      "step": 14899
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0563931465148926,
      "learning_rate": 8.20356694064478e-08,
      "loss": 0.5428,
      "step": 14900
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9862633347511292,
      "learning_rate": 8.185068108844507e-08,
      "loss": 0.4719,
      "step": 14901
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0177010297775269,
      "learning_rate": 8.166589985660056e-08,
      "loss": 0.4688,
      "step": 14902
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1310821771621704,
      "learning_rate": 8.148132571869582e-08,
      "loss": 0.534,
      "step": 14903
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.032629132270813,
      "learning_rate": 8.129695868250242e-08,
      "loss": 0.4774,
      "step": 14904
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9880489706993103,
      "learning_rate": 8.111279875578304e-08,
      "loss": 0.4906,
      "step": 14905
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0270191431045532,
      "learning_rate": 8.092884594629147e-08,
      "loss": 0.5082,
      "step": 14906
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0651283264160156,
      "learning_rate": 8.074510026177485e-08,
      "loss": 0.4889,
      "step": 14907
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0554499626159668,
      "learning_rate": 8.056156170996866e-08,
      "loss": 0.5256,
      "step": 14908
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.075369119644165,
      "learning_rate": 8.03782302986017e-08,
      "loss": 0.5045,
      "step": 14909
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0443814992904663,
      "learning_rate": 8.019510603539338e-08,
      "loss": 0.4852,
      "step": 14910
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0223075151443481,
      "learning_rate": 8.001218892805474e-08,
      "loss": 0.4904,
      "step": 14911
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.062279224395752,
      "learning_rate": 7.982947898428739e-08,
      "loss": 0.5036,
      "step": 14912
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0690451860427856,
      "learning_rate": 7.964697621178463e-08,
      "loss": 0.5454,
      "step": 14913
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0020755529403687,
      "learning_rate": 7.946468061823031e-08,
      "loss": 0.4802,
      "step": 14914
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.057814598083496,
      "learning_rate": 7.928259221130163e-08,
      "loss": 0.5025,
      "step": 14915
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0814919471740723,
      "learning_rate": 7.910071099866523e-08,
      "loss": 0.5347,
      "step": 14916
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.983623206615448,
      "learning_rate": 7.891903698797886e-08,
      "loss": 0.4386,
      "step": 14917
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0692455768585205,
      "learning_rate": 7.87375701868931e-08,
      "loss": 0.5177,
      "step": 14918
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0529448986053467,
      "learning_rate": 7.855631060304792e-08,
      "loss": 0.4429,
      "step": 14919
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0599701404571533,
      "learning_rate": 7.837525824407665e-08,
      "loss": 0.4916,
      "step": 14920
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0690699815750122,
      "learning_rate": 7.819441311760156e-08,
      "loss": 0.4819,
      "step": 14921
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1062662601470947,
      "learning_rate": 7.801377523123877e-08,
      "loss": 0.5503,
      "step": 14922
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9787145256996155,
      "learning_rate": 7.783334459259273e-08,
      "loss": 0.4679,
      "step": 14923
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0887858867645264,
      "learning_rate": 7.765312120926182e-08,
      "loss": 0.5464,
      "step": 14924
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9913561940193176,
      "learning_rate": 7.747310508883444e-08,
      "loss": 0.4786,
      "step": 14925
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0134284496307373,
      "learning_rate": 7.729329623889114e-08,
      "loss": 0.4498,
      "step": 14926
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.087054967880249,
      "learning_rate": 7.711369466700147e-08,
      "loss": 0.561,
      "step": 14927
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9631887674331665,
      "learning_rate": 7.693430038072824e-08,
      "loss": 0.4651,
      "step": 14928
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9987837672233582,
      "learning_rate": 7.675511338762654e-08,
      "loss": 0.4178,
      "step": 14929
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.06411874294281,
      "learning_rate": 7.657613369523975e-08,
      "loss": 0.4956,
      "step": 14930
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0640369653701782,
      "learning_rate": 7.639736131110465e-08,
      "loss": 0.5093,
      "step": 14931
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.065610647201538,
      "learning_rate": 7.621879624274853e-08,
      "loss": 0.4942,
      "step": 14932
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.056787133216858,
      "learning_rate": 7.604043849769094e-08,
      "loss": 0.4742,
      "step": 14933
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0905635356903076,
      "learning_rate": 7.586228808344087e-08,
      "loss": 0.5321,
      "step": 14934
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0646344423294067,
      "learning_rate": 7.56843450075001e-08,
      "loss": 0.5134,
      "step": 14935
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0673986673355103,
      "learning_rate": 7.550660927736042e-08,
      "loss": 0.4523,
      "step": 14936
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1001416444778442,
      "learning_rate": 7.53290809005075e-08,
      "loss": 0.5033,
      "step": 14937
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0971667766571045,
      "learning_rate": 7.515175988441481e-08,
      "loss": 0.4623,
      "step": 14938
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0954697132110596,
      "learning_rate": 7.497464623654915e-08,
      "loss": 0.4978,
      "step": 14939
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0560624599456787,
      "learning_rate": 7.479773996436845e-08,
      "loss": 0.4591,
      "step": 14940
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.057520866394043,
      "learning_rate": 7.46210410753212e-08,
      "loss": 0.4754,
      "step": 14941
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0726072788238525,
      "learning_rate": 7.44445495768481e-08,
      "loss": 0.5061,
      "step": 14942
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.061622142791748,
      "learning_rate": 7.426826547637989e-08,
      "loss": 0.5456,
      "step": 14943
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.933109700679779,
      "learning_rate": 7.40921887813395e-08,
      "loss": 0.4892,
      "step": 14944
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0784633159637451,
      "learning_rate": 7.391631949914102e-08,
      "loss": 0.4997,
      "step": 14945
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0952481031417847,
      "learning_rate": 7.374065763719018e-08,
      "loss": 0.4948,
      "step": 14946
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0322444438934326,
      "learning_rate": 7.356520320288274e-08,
      "loss": 0.5358,
      "step": 14947
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0545867681503296,
      "learning_rate": 7.338995620360722e-08,
      "loss": 0.4536,
      "step": 14948
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0081437826156616,
      "learning_rate": 7.321491664674163e-08,
      "loss": 0.4412,
      "step": 14949
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1125178337097168,
      "learning_rate": 7.304008453965727e-08,
      "loss": 0.5431,
      "step": 14950
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0762580633163452,
      "learning_rate": 7.286545988971495e-08,
      "loss": 0.5061,
      "step": 14951
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0408401489257812,
      "learning_rate": 7.269104270426818e-08,
      "loss": 0.4921,
      "step": 14952
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1675944328308105,
      "learning_rate": 7.251683299066059e-08,
      "loss": 0.5155,
      "step": 14953
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1432479619979858,
      "learning_rate": 7.23428307562274e-08,
      "loss": 0.519,
      "step": 14954
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0275051593780518,
      "learning_rate": 7.216903600829605e-08,
      "loss": 0.4785,
      "step": 14955
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1260793209075928,
      "learning_rate": 7.199544875418407e-08,
      "loss": 0.5439,
      "step": 14956
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1154240369796753,
      "learning_rate": 7.182206900119948e-08,
      "loss": 0.5083,
      "step": 14957
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.080336570739746,
      "learning_rate": 7.164889675664477e-08,
      "loss": 0.5349,
      "step": 14958
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1574434041976929,
      "learning_rate": 7.147593202781022e-08,
      "loss": 0.4962,
      "step": 14959
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1089222431182861,
      "learning_rate": 7.13031748219789e-08,
      "loss": 0.4882,
      "step": 14960
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0524308681488037,
      "learning_rate": 7.113062514642555e-08,
      "loss": 0.4888,
      "step": 14961
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0922738313674927,
      "learning_rate": 7.095828300841435e-08,
      "loss": 0.4505,
      "step": 14962
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0481584072113037,
      "learning_rate": 7.078614841520392e-08,
      "loss": 0.5466,
      "step": 14963
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9718726873397827,
      "learning_rate": 7.061422137404129e-08,
      "loss": 0.5016,
      "step": 14964
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.035399317741394,
      "learning_rate": 7.044250189216561e-08,
      "loss": 0.4608,
      "step": 14965
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0872936248779297,
      "learning_rate": 7.027098997680726e-08,
      "loss": 0.5362,
      "step": 14966
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.107075572013855,
      "learning_rate": 7.00996856351882e-08,
      "loss": 0.5019,
      "step": 14967
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.14595627784729,
      "learning_rate": 6.992858887452158e-08,
      "loss": 0.4828,
      "step": 14968
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0169750452041626,
      "learning_rate": 6.975769970201163e-08,
      "loss": 0.5016,
      "step": 14969
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9879432320594788,
      "learning_rate": 6.958701812485369e-08,
      "loss": 0.5074,
      "step": 14970
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.06552255153656,
      "learning_rate": 6.94165441502348e-08,
      "loss": 0.498,
      "step": 14971
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1178659200668335,
      "learning_rate": 6.924627778533366e-08,
      "loss": 0.4977,
      "step": 14972
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.096285343170166,
      "learning_rate": 6.907621903731842e-08,
      "loss": 0.5299,
      "step": 14973
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0517348051071167,
      "learning_rate": 6.890636791335003e-08,
      "loss": 0.4543,
      "step": 14974
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9691824316978455,
      "learning_rate": 6.873672442058054e-08,
      "loss": 0.4705,
      "step": 14975
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0984865427017212,
      "learning_rate": 6.856728856615314e-08,
      "loss": 0.4788,
      "step": 14976
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0343778133392334,
      "learning_rate": 6.839806035720209e-08,
      "loss": 0.5031,
      "step": 14977
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.026618242263794,
      "learning_rate": 6.822903980085282e-08,
      "loss": 0.5167,
      "step": 14978
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0796104669570923,
      "learning_rate": 6.806022690422187e-08,
      "loss": 0.5074,
      "step": 14979
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0644936561584473,
      "learning_rate": 6.789162167441798e-08,
      "loss": 0.5252,
      "step": 14980
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0769859552383423,
      "learning_rate": 6.772322411854048e-08,
      "loss": 0.5469,
      "step": 14981
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1746636629104614,
      "learning_rate": 6.755503424368037e-08,
      "loss": 0.5216,
      "step": 14982
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1133933067321777,
      "learning_rate": 6.73870520569181e-08,
      "loss": 0.5592,
      "step": 14983
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0979746580123901,
      "learning_rate": 6.721927756532853e-08,
      "loss": 0.4748,
      "step": 14984
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0335559844970703,
      "learning_rate": 6.705171077597495e-08,
      "loss": 0.4971,
      "step": 14985
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.084326148033142,
      "learning_rate": 6.68843516959139e-08,
      "loss": 0.5545,
      "step": 14986
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1023401021957397,
      "learning_rate": 6.67172003321903e-08,
      "loss": 0.5284,
      "step": 14987
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1031067371368408,
      "learning_rate": 6.655025669184522e-08,
      "loss": 0.5324,
      "step": 14988
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0138829946517944,
      "learning_rate": 6.638352078190636e-08,
      "loss": 0.4523,
      "step": 14989
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0137693881988525,
      "learning_rate": 6.621699260939418e-08,
      "loss": 0.4951,
      "step": 14990
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0743118524551392,
      "learning_rate": 6.605067218132145e-08,
      "loss": 0.5367,
      "step": 14991
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0681713819503784,
      "learning_rate": 6.58845595046903e-08,
      "loss": 0.562,
      "step": 14992
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1324034929275513,
      "learning_rate": 6.571865458649629e-08,
      "loss": 0.5746,
      "step": 14993
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0964475870132446,
      "learning_rate": 6.555295743372492e-08,
      "loss": 0.49,
      "step": 14994
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9967697262763977,
      "learning_rate": 6.538746805335284e-08,
      "loss": 0.4638,
      "step": 14995
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.097658395767212,
      "learning_rate": 6.52221864523478e-08,
      "loss": 0.5465,
      "step": 14996
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0048773288726807,
      "learning_rate": 6.505711263766978e-08,
      "loss": 0.4931,
      "step": 14997
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0225428342819214,
      "learning_rate": 6.48922466162688e-08,
      "loss": 0.5089,
      "step": 14998
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.98487389087677,
      "learning_rate": 6.472758839508819e-08,
      "loss": 0.4879,
      "step": 14999
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.986153244972229,
      "learning_rate": 6.456313798105962e-08,
      "loss": 0.4959,
      "step": 15000
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.085741639137268,
      "learning_rate": 6.439889538110867e-08,
      "loss": 0.5745,
      "step": 15001
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0911871194839478,
      "learning_rate": 6.423486060215034e-08,
      "loss": 0.5103,
      "step": 15002
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0940908193588257,
      "learning_rate": 6.40710336510919e-08,
      "loss": 0.4667,
      "step": 15003
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0498441457748413,
      "learning_rate": 6.390741453483119e-08,
      "loss": 0.4825,
      "step": 15004
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.019486665725708,
      "learning_rate": 6.374400326025765e-08,
      "loss": 0.5004,
      "step": 15005
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1773452758789062,
      "learning_rate": 6.358079983425247e-08,
      "loss": 0.4969,
      "step": 15006
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.051240086555481,
      "learning_rate": 6.341780426368737e-08,
      "loss": 0.5094,
      "step": 15007
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0501857995986938,
      "learning_rate": 6.32550165554252e-08,
      "loss": 0.5171,
      "step": 15008
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1003665924072266,
      "learning_rate": 6.309243671632048e-08,
      "loss": 0.541,
      "step": 15009
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1350167989730835,
      "learning_rate": 6.293006475321939e-08,
      "loss": 0.5066,
      "step": 15010
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1172188520431519,
      "learning_rate": 6.276790067295813e-08,
      "loss": 0.5128,
      "step": 15011
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0877690315246582,
      "learning_rate": 6.260594448236513e-08,
      "loss": 0.4797,
      "step": 15012
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0320991277694702,
      "learning_rate": 6.244419618825992e-08,
      "loss": 0.4801,
      "step": 15013
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0631580352783203,
      "learning_rate": 6.228265579745318e-08,
      "loss": 0.4995,
      "step": 15014
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0299152135849,
      "learning_rate": 6.212132331674725e-08,
      "loss": 0.5023,
      "step": 15015
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.146296739578247,
      "learning_rate": 6.196019875293391e-08,
      "loss": 0.5539,
      "step": 15016
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.012778401374817,
      "learning_rate": 6.179928211279884e-08,
      "loss": 0.4976,
      "step": 15017
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0477205514907837,
      "learning_rate": 6.163857340311718e-08,
      "loss": 0.4565,
      "step": 15018
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0732078552246094,
      "learning_rate": 6.147807263065575e-08,
      "loss": 0.4543,
      "step": 15019
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1308003664016724,
      "learning_rate": 6.131777980217302e-08,
      "loss": 0.5284,
      "step": 15020
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0013229846954346,
      "learning_rate": 6.115769492441859e-08,
      "loss": 0.5011,
      "step": 15021
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1048378944396973,
      "learning_rate": 6.099781800413151e-08,
      "loss": 0.5263,
      "step": 15022
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2660025358200073,
      "learning_rate": 6.083814904804586e-08,
      "loss": 0.5616,
      "step": 15023
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.045714259147644,
      "learning_rate": 6.067868806288346e-08,
      "loss": 0.5137,
      "step": 15024
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9999276399612427,
      "learning_rate": 6.05194350553584e-08,
      "loss": 0.4884,
      "step": 15025
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1267025470733643,
      "learning_rate": 6.036039003217697e-08,
      "loss": 0.5067,
      "step": 15026
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0568078756332397,
      "learning_rate": 6.02015530000355e-08,
      "loss": 0.5517,
      "step": 15027
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0691008567810059,
      "learning_rate": 6.00429239656225e-08,
      "loss": 0.4856,
      "step": 15028
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0379536151885986,
      "learning_rate": 5.988450293561765e-08,
      "loss": 0.471,
      "step": 15029
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0408005714416504,
      "learning_rate": 5.972628991669006e-08,
      "loss": 0.4636,
      "step": 15030
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1032767295837402,
      "learning_rate": 5.956828491550326e-08,
      "loss": 0.5221,
      "step": 15031
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0133955478668213,
      "learning_rate": 5.941048793870918e-08,
      "loss": 0.4752,
      "step": 15032
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.028398871421814,
      "learning_rate": 5.92528989929525e-08,
      "loss": 0.5115,
      "step": 15033
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9963531494140625,
      "learning_rate": 5.9095518084868467e-08,
      "loss": 0.5157,
      "step": 15034
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0203803777694702,
      "learning_rate": 5.893834522108399e-08,
      "loss": 0.5161,
      "step": 15035
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.105560064315796,
      "learning_rate": 5.8781380408217124e-08,
      "loss": 0.4944,
      "step": 15036
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0529372692108154,
      "learning_rate": 5.862462365287702e-08,
      "loss": 0.4984,
      "step": 15037
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.101141333580017,
      "learning_rate": 5.846807496166451e-08,
      "loss": 0.5171,
      "step": 15038
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.069178581237793,
      "learning_rate": 5.831173434117043e-08,
      "loss": 0.4868,
      "step": 15039
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0065348148345947,
      "learning_rate": 5.815560179797897e-08,
      "loss": 0.4754,
      "step": 15040
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.010067343711853,
      "learning_rate": 5.7999677338663184e-08,
      "loss": 0.475,
      "step": 15041
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0668708086013794,
      "learning_rate": 5.7843960969790056e-08,
      "loss": 0.5041,
      "step": 15042
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0093073844909668,
      "learning_rate": 5.768845269791379e-08,
      "loss": 0.5062,
      "step": 15043
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9874602556228638,
      "learning_rate": 5.7533152529584135e-08,
      "loss": 0.4237,
      "step": 15044
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1246216297149658,
      "learning_rate": 5.7378060471340866e-08,
      "loss": 0.4612,
      "step": 15045
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1335710287094116,
      "learning_rate": 5.7223176529712097e-08,
      "loss": 0.5429,
      "step": 15046
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9604591727256775,
      "learning_rate": 5.70685007112215e-08,
      "loss": 0.4539,
      "step": 15047
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0197806358337402,
      "learning_rate": 5.691403302238052e-08,
      "loss": 0.4636,
      "step": 15048
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0385576486587524,
      "learning_rate": 5.6759773469694523e-08,
      "loss": 0.5175,
      "step": 15049
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9743496179580688,
      "learning_rate": 5.660572205965775e-08,
      "loss": 0.4715,
      "step": 15050
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.082789659500122,
      "learning_rate": 5.645187879875724e-08,
      "loss": 0.4957,
      "step": 15051
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9587162137031555,
      "learning_rate": 5.6298243693470586e-08,
      "loss": 0.5016,
      "step": 15052
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2008246183395386,
      "learning_rate": 5.614481675026762e-08,
      "loss": 0.4949,
      "step": 15053
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0328693389892578,
      "learning_rate": 5.59915979756076e-08,
      "loss": 0.4821,
      "step": 15054
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.09230375289917,
      "learning_rate": 5.583858737594205e-08,
      "loss": 0.486,
      "step": 15055
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1106675863265991,
      "learning_rate": 5.5685784957714707e-08,
      "loss": 0.4895,
      "step": 15056
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0798262357711792,
      "learning_rate": 5.5533190727358745e-08,
      "loss": 0.5254,
      "step": 15057
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0557674169540405,
      "learning_rate": 5.538080469129958e-08,
      "loss": 0.5157,
      "step": 15058
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9619843363761902,
      "learning_rate": 5.522862685595376e-08,
      "loss": 0.4913,
      "step": 15059
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0052016973495483,
      "learning_rate": 5.507665722772837e-08,
      "loss": 0.4738,
      "step": 15060
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.033711314201355,
      "learning_rate": 5.492489581302329e-08,
      "loss": 0.4792,
      "step": 15061
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0828896760940552,
      "learning_rate": 5.477334261822842e-08,
      "loss": 0.503,
      "step": 15062
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0456746816635132,
      "learning_rate": 5.46219976497242e-08,
      "loss": 0.4859,
      "step": 15063
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9850556254386902,
      "learning_rate": 5.447086091388443e-08,
      "loss": 0.4761,
      "step": 15064
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0214520692825317,
      "learning_rate": 5.4319932417072344e-08,
      "loss": 0.5008,
      "step": 15065
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0863326787948608,
      "learning_rate": 5.416921216564286e-08,
      "loss": 0.5266,
      "step": 15066
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0874569416046143,
      "learning_rate": 5.401870016594313e-08,
      "loss": 0.507,
      "step": 15067
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1380571126937866,
      "learning_rate": 5.38683964243103e-08,
      "loss": 0.5038,
      "step": 15068
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0237140655517578,
      "learning_rate": 5.3718300947072086e-08,
      "loss": 0.4931,
      "step": 15069
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.051347255706787,
      "learning_rate": 5.356841374055011e-08,
      "loss": 0.5361,
      "step": 15070
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0510364770889282,
      "learning_rate": 5.341873481105431e-08,
      "loss": 0.5426,
      "step": 15071
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.096970796585083,
      "learning_rate": 5.3269264164887977e-08,
      "loss": 0.554,
      "step": 15072
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1007144451141357,
      "learning_rate": 5.3120001808344425e-08,
      "loss": 0.4805,
      "step": 15073
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.994536280632019,
      "learning_rate": 5.297094774770861e-08,
      "loss": 0.4438,
      "step": 15074
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0439180135726929,
      "learning_rate": 5.282210198925664e-08,
      "loss": 0.5223,
      "step": 15075
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0629830360412598,
      "learning_rate": 5.267346453925626e-08,
      "loss": 0.5244,
      "step": 15076
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2122411727905273,
      "learning_rate": 5.2525035403965805e-08,
      "loss": 0.4977,
      "step": 15077
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.081017017364502,
      "learning_rate": 5.237681458963473e-08,
      "loss": 0.5228,
      "step": 15078
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.982690691947937,
      "learning_rate": 5.222880210250469e-08,
      "loss": 0.4755,
      "step": 15079
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0423614978790283,
      "learning_rate": 5.2080997948807944e-08,
      "loss": 0.5246,
      "step": 15080
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.084811806678772,
      "learning_rate": 5.193340213476727e-08,
      "loss": 0.4849,
      "step": 15081
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1122099161148071,
      "learning_rate": 5.178601466659827e-08,
      "loss": 0.5022,
      "step": 15082
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1095976829528809,
      "learning_rate": 5.163883555050708e-08,
      "loss": 0.5244,
      "step": 15083
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0307388305664062,
      "learning_rate": 5.149186479268986e-08,
      "loss": 0.4971,
      "step": 15084
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1129001379013062,
      "learning_rate": 5.134510239933554e-08,
      "loss": 0.5413,
      "step": 15085
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0419707298278809,
      "learning_rate": 5.119854837662419e-08,
      "loss": 0.4838,
      "step": 15086
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1126266717910767,
      "learning_rate": 5.1052202730725865e-08,
      "loss": 0.4948,
      "step": 15087
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0315604209899902,
      "learning_rate": 5.0906065467803965e-08,
      "loss": 0.4857,
      "step": 15088
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0968666076660156,
      "learning_rate": 5.0760136594010246e-08,
      "loss": 0.4508,
      "step": 15089
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.015987753868103,
      "learning_rate": 5.061441611549034e-08,
      "loss": 0.4891,
      "step": 15090
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.024927020072937,
      "learning_rate": 5.046890403837989e-08,
      "loss": 0.5044,
      "step": 15091
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.001304268836975,
      "learning_rate": 5.032360036880568e-08,
      "loss": 0.49,
      "step": 15092
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0158631801605225,
      "learning_rate": 5.0178505112885576e-08,
      "loss": 0.4882,
      "step": 15093
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1382091045379639,
      "learning_rate": 5.00336182767297e-08,
      "loss": 0.4925,
      "step": 15094
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.147162914276123,
      "learning_rate": 4.988893986643817e-08,
      "loss": 0.4986,
      "step": 15095
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9799484014511108,
      "learning_rate": 4.9744469888103887e-08,
      "loss": 0.5284,
      "step": 15096
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0316082239151,
      "learning_rate": 4.9600208347809206e-08,
      "loss": 0.4583,
      "step": 15097
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1625584363937378,
      "learning_rate": 4.945615525162761e-08,
      "loss": 0.527,
      "step": 15098
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.015092134475708,
      "learning_rate": 4.931231060562702e-08,
      "loss": 0.4586,
      "step": 15099
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.123383641242981,
      "learning_rate": 4.916867441586204e-08,
      "loss": 0.5561,
      "step": 15100
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.009797215461731,
      "learning_rate": 4.902524668838116e-08,
      "loss": 0.4791,
      "step": 15101
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.050193190574646,
      "learning_rate": 4.88820274292251e-08,
      "loss": 0.4855,
      "step": 15102
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0637693405151367,
      "learning_rate": 4.873901664442182e-08,
      "loss": 0.4757,
      "step": 15103
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0080116987228394,
      "learning_rate": 4.8596214339995395e-08,
      "loss": 0.4764,
      "step": 15104
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1039217710494995,
      "learning_rate": 4.8453620521957124e-08,
      "loss": 0.503,
      "step": 15105
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.090170979499817,
      "learning_rate": 4.83112351963122e-08,
      "loss": 0.465,
      "step": 15106
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1621887683868408,
      "learning_rate": 4.816905836905528e-08,
      "loss": 0.5391,
      "step": 15107
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0159717798233032,
      "learning_rate": 4.802709004617267e-08,
      "loss": 0.493,
      "step": 15108
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0658420324325562,
      "learning_rate": 4.788533023364295e-08,
      "loss": 0.4565,
      "step": 15109
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1132019758224487,
      "learning_rate": 4.77437789374352e-08,
      "loss": 0.5395,
      "step": 15110
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1035053730010986,
      "learning_rate": 4.760243616350913e-08,
      "loss": 0.5191,
      "step": 15111
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1152589321136475,
      "learning_rate": 4.746130191781606e-08,
      "loss": 0.499,
      "step": 15112
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0198644399642944,
      "learning_rate": 4.7320376206299034e-08,
      "loss": 0.5292,
      "step": 15113
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0909252166748047,
      "learning_rate": 4.717965903489219e-08,
      "loss": 0.5047,
      "step": 15114
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.043396234512329,
      "learning_rate": 4.7039150409519674e-08,
      "loss": 0.4878,
      "step": 15115
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0830494165420532,
      "learning_rate": 4.6898850336098975e-08,
      "loss": 0.489,
      "step": 15116
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.072623610496521,
      "learning_rate": 4.675875882053704e-08,
      "loss": 0.5088,
      "step": 15117
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9616676568984985,
      "learning_rate": 4.6618875868733037e-08,
      "loss": 0.4776,
      "step": 15118
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0668658018112183,
      "learning_rate": 4.6479201486575585e-08,
      "loss": 0.4682,
      "step": 15119
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0576627254486084,
      "learning_rate": 4.633973567994776e-08,
      "loss": 0.5172,
      "step": 15120
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0642863512039185,
      "learning_rate": 4.620047845472098e-08,
      "loss": 0.5077,
      "step": 15121
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0122225284576416,
      "learning_rate": 4.606142981675887e-08,
      "loss": 0.5101,
      "step": 15122
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.069036602973938,
      "learning_rate": 4.592258977191622e-08,
      "loss": 0.5149,
      "step": 15123
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.05073881149292,
      "learning_rate": 4.578395832603999e-08,
      "loss": 0.4921,
      "step": 15124
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.090564489364624,
      "learning_rate": 4.5645535484966085e-08,
      "loss": 0.4889,
      "step": 15125
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0945969820022583,
      "learning_rate": 4.5507321254524287e-08,
      "loss": 0.4914,
      "step": 15126
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1259173154830933,
      "learning_rate": 4.536931564053382e-08,
      "loss": 0.51,
      "step": 15127
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1442131996154785,
      "learning_rate": 4.523151864880504e-08,
      "loss": 0.4917,
      "step": 15128
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.194004774093628,
      "learning_rate": 4.5093930285141086e-08,
      "loss": 0.5238,
      "step": 15129
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0926456451416016,
      "learning_rate": 4.4956550555334546e-08,
      "loss": 0.5154,
      "step": 15130
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0433238744735718,
      "learning_rate": 4.4819379465170785e-08,
      "loss": 0.4799,
      "step": 15131
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1860097646713257,
      "learning_rate": 4.4682417020425194e-08,
      "loss": 0.5124,
      "step": 15132
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1235885620117188,
      "learning_rate": 4.454566322686371e-08,
      "loss": 0.5486,
      "step": 15133
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1014682054519653,
      "learning_rate": 4.440911809024673e-08,
      "loss": 0.5076,
      "step": 15134
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.122446060180664,
      "learning_rate": 4.427278161632187e-08,
      "loss": 0.4887,
      "step": 15135
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0771417617797852,
      "learning_rate": 4.413665381083065e-08,
      "loss": 0.433,
      "step": 15136
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.126584768295288,
      "learning_rate": 4.4000734679504606e-08,
      "loss": 0.4713,
      "step": 15137
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0992733240127563,
      "learning_rate": 4.386502422806749e-08,
      "loss": 0.5091,
      "step": 15138
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.031760573387146,
      "learning_rate": 4.37295224622325e-08,
      "loss": 0.4908,
      "step": 15139
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9821397066116333,
      "learning_rate": 4.359422938770619e-08,
      "loss": 0.5034,
      "step": 15140
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1158956289291382,
      "learning_rate": 4.3459145010184e-08,
      "loss": 0.4501,
      "step": 15141
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0483918190002441,
      "learning_rate": 4.3324269335355274e-08,
      "loss": 0.5242,
      "step": 15142
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1130990982055664,
      "learning_rate": 4.31896023688988e-08,
      "loss": 0.5054,
      "step": 15143
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1633292436599731,
      "learning_rate": 4.305514411648393e-08,
      "loss": 0.4917,
      "step": 15144
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.010609745979309,
      "learning_rate": 4.2920894583773906e-08,
      "loss": 0.4604,
      "step": 15145
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.957053005695343,
      "learning_rate": 4.278685377641978e-08,
      "loss": 0.4582,
      "step": 15146
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0870672464370728,
      "learning_rate": 4.2653021700066466e-08,
      "loss": 0.4749,
      "step": 15147
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.087636113166809,
      "learning_rate": 4.251939836034946e-08,
      "loss": 0.4881,
      "step": 15148
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.094960331916809,
      "learning_rate": 4.238598376289482e-08,
      "loss": 0.4742,
      "step": 15149
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0509110689163208,
      "learning_rate": 4.225277791331972e-08,
      "loss": 0.4919,
      "step": 15150
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0320489406585693,
      "learning_rate": 4.211978081723356e-08,
      "loss": 0.492,
      "step": 15151
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0644983053207397,
      "learning_rate": 4.19869924802363e-08,
      "loss": 0.4932,
      "step": 15152
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9970551133155823,
      "learning_rate": 4.185441290791903e-08,
      "loss": 0.4491,
      "step": 15153
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0908452272415161,
      "learning_rate": 4.1722042105863946e-08,
      "loss": 0.5192,
      "step": 15154
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.086653232574463,
      "learning_rate": 4.158988007964548e-08,
      "loss": 0.5056,
      "step": 15155
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.088154911994934,
      "learning_rate": 4.145792683482808e-08,
      "loss": 0.5084,
      "step": 15156
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0358635187149048,
      "learning_rate": 4.132618237696784e-08,
      "loss": 0.5237,
      "step": 15157
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0679439306259155,
      "learning_rate": 4.1194646711612555e-08,
      "loss": 0.5357,
      "step": 15158
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0736949443817139,
      "learning_rate": 4.1063319844299454e-08,
      "loss": 0.4725,
      "step": 15159
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0397822856903076,
      "learning_rate": 4.0932201780559674e-08,
      "loss": 0.5,
      "step": 15160
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1251829862594604,
      "learning_rate": 4.080129252591325e-08,
      "loss": 0.4841,
      "step": 15161
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0683003664016724,
      "learning_rate": 4.0670592085872984e-08,
      "loss": 0.5109,
      "step": 15162
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0046395063400269,
      "learning_rate": 4.054010046594115e-08,
      "loss": 0.4897,
      "step": 15163
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0667266845703125,
      "learning_rate": 4.040981767161334e-08,
      "loss": 0.5108,
      "step": 15164
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0909169912338257,
      "learning_rate": 4.027974370837518e-08,
      "loss": 0.5137,
      "step": 15165
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.038557529449463,
      "learning_rate": 4.014987858170283e-08,
      "loss": 0.5008,
      "step": 15166
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.090004324913025,
      "learning_rate": 4.0020222297065256e-08,
      "loss": 0.4983,
      "step": 15167
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1984957456588745,
      "learning_rate": 3.9890774859921987e-08,
      "loss": 0.5201,
      "step": 15168
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1261996030807495,
      "learning_rate": 3.97615362757231e-08,
      "loss": 0.5366,
      "step": 15169
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1402543783187866,
      "learning_rate": 3.9632506549910356e-08,
      "loss": 0.5212,
      "step": 15170
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0591979026794434,
      "learning_rate": 3.9503685687916627e-08,
      "loss": 0.46,
      "step": 15171
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0578498840332031,
      "learning_rate": 3.937507369516702e-08,
      "loss": 0.5166,
      "step": 15172
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1559275388717651,
      "learning_rate": 3.92466705770761e-08,
      "loss": 0.4912,
      "step": 15173
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0619618892669678,
      "learning_rate": 3.911847633905008e-08,
      "loss": 0.5137,
      "step": 15174
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.071740746498108,
      "learning_rate": 3.899049098648799e-08,
      "loss": 0.4929,
      "step": 15175
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1121670007705688,
      "learning_rate": 3.88627145247783e-08,
      "loss": 0.5078,
      "step": 15176
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.100979208946228,
      "learning_rate": 3.873514695930114e-08,
      "loss": 0.5421,
      "step": 15177
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0536065101623535,
      "learning_rate": 3.860778829542777e-08,
      "loss": 0.5298,
      "step": 15178
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0702991485595703,
      "learning_rate": 3.848063853852113e-08,
      "loss": 0.4964,
      "step": 15179
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0179272890090942,
      "learning_rate": 3.835369769393471e-08,
      "loss": 0.4662,
      "step": 15180
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0654373168945312,
      "learning_rate": 3.822696576701368e-08,
      "loss": 0.5076,
      "step": 15181
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0972926616668701,
      "learning_rate": 3.8100442763094324e-08,
      "loss": 0.4772,
      "step": 15182
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0048757791519165,
      "learning_rate": 3.797412868750461e-08,
      "loss": 0.4918,
      "step": 15183
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0761852264404297,
      "learning_rate": 3.784802354556249e-08,
      "loss": 0.4763,
      "step": 15184
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1550503969192505,
      "learning_rate": 3.7722127342578183e-08,
      "loss": 0.4978,
      "step": 15185
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.063727855682373,
      "learning_rate": 3.759644008385244e-08,
      "loss": 0.4868,
      "step": 15186
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0755116939544678,
      "learning_rate": 3.747096177467768e-08,
      "loss": 0.4823,
      "step": 15187
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.024055004119873,
      "learning_rate": 3.7345692420337476e-08,
      "loss": 0.46,
      "step": 15188
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0906200408935547,
      "learning_rate": 3.722063202610593e-08,
      "loss": 0.5097,
      "step": 15189
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0795869827270508,
      "learning_rate": 3.709578059724939e-08,
      "loss": 0.4971,
      "step": 15190
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0026204586029053,
      "learning_rate": 3.697113813902531e-08,
      "loss": 0.4939,
      "step": 15191
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0426870584487915,
      "learning_rate": 3.684670465668116e-08,
      "loss": 0.5157,
      "step": 15192
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0251553058624268,
      "learning_rate": 3.6722480155456655e-08,
      "loss": 0.4769,
      "step": 15193
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.051120400428772,
      "learning_rate": 3.6598464640582586e-08,
      "loss": 0.5039,
      "step": 15194
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0711596012115479,
      "learning_rate": 3.64746581172809e-08,
      "loss": 0.5578,
      "step": 15195
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0845519304275513,
      "learning_rate": 3.6351060590764656e-08,
      "loss": 0.4918,
      "step": 15196
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9789648056030273,
      "learning_rate": 3.6227672066237454e-08,
      "loss": 0.4764,
      "step": 15197
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0158416032791138,
      "learning_rate": 3.6104492548895695e-08,
      "loss": 0.5349,
      "step": 15198
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1074073314666748,
      "learning_rate": 3.5981522043925796e-08,
      "loss": 0.4843,
      "step": 15199
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0457102060317993,
      "learning_rate": 3.585876055650528e-08,
      "loss": 0.4954,
      "step": 15200
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0220361948013306,
      "learning_rate": 3.5736208091802784e-08,
      "loss": 0.4796,
      "step": 15201
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1319571733474731,
      "learning_rate": 3.5613864654979734e-08,
      "loss": 0.5557,
      "step": 15202
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1150519847869873,
      "learning_rate": 3.5491730251187016e-08,
      "loss": 0.5144,
      "step": 15203
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0242811441421509,
      "learning_rate": 3.5369804885567185e-08,
      "loss": 0.4811,
      "step": 15204
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.001131534576416,
      "learning_rate": 3.52480885632539e-08,
      "loss": 0.4878,
      "step": 15205
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0957980155944824,
      "learning_rate": 3.512658128937252e-08,
      "loss": 0.5016,
      "step": 15206
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0731688737869263,
      "learning_rate": 3.500528306904005e-08,
      "loss": 0.5034,
      "step": 15207
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.158174753189087,
      "learning_rate": 3.488419390736242e-08,
      "loss": 0.4819,
      "step": 15208
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.128722071647644,
      "learning_rate": 3.476331380943887e-08,
      "loss": 0.555,
      "step": 15209
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.090911865234375,
      "learning_rate": 3.464264278035978e-08,
      "loss": 0.5168,
      "step": 15210
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9718244671821594,
      "learning_rate": 3.452218082520553e-08,
      "loss": 0.4606,
      "step": 15211
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0440447330474854,
      "learning_rate": 3.440192794904873e-08,
      "loss": 0.4768,
      "step": 15212
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0195415019989014,
      "learning_rate": 3.4281884156953106e-08,
      "loss": 0.503,
      "step": 15213
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0665068626403809,
      "learning_rate": 3.416204945397239e-08,
      "loss": 0.4998,
      "step": 15214
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9613544344902039,
      "learning_rate": 3.4042423845153104e-08,
      "loss": 0.4924,
      "step": 15215
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0670095682144165,
      "learning_rate": 3.392300733553178e-08,
      "loss": 0.4948,
      "step": 15216
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1226885318756104,
      "learning_rate": 3.380379993013716e-08,
      "loss": 0.5421,
      "step": 15217
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1233232021331787,
      "learning_rate": 3.368480163398802e-08,
      "loss": 0.4939,
      "step": 15218
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0569380521774292,
      "learning_rate": 3.356601245209534e-08,
      "loss": 0.5018,
      "step": 15219
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0649720430374146,
      "learning_rate": 3.344743238946124e-08,
      "loss": 0.4755,
      "step": 15220
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0704591274261475,
      "learning_rate": 3.332906145107839e-08,
      "loss": 0.4771,
      "step": 15221
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0231249332427979,
      "learning_rate": 3.3210899641930586e-08,
      "loss": 0.4562,
      "step": 15222
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1724458932876587,
      "learning_rate": 3.3092946966994385e-08,
      "loss": 0.5385,
      "step": 15223
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0665589570999146,
      "learning_rate": 3.297520343123473e-08,
      "loss": 0.5129,
      "step": 15224
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0072379112243652,
      "learning_rate": 3.285766903961096e-08,
      "loss": 0.4644,
      "step": 15225
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1121455430984497,
      "learning_rate": 3.274034379707081e-08,
      "loss": 0.5327,
      "step": 15226
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.982637882232666,
      "learning_rate": 3.262322770855475e-08,
      "loss": 0.5118,
      "step": 15227
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1324604749679565,
      "learning_rate": 3.250632077899496e-08,
      "loss": 0.5422,
      "step": 15228
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1338701248168945,
      "learning_rate": 3.238962301331305e-08,
      "loss": 0.5434,
      "step": 15229
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.043928861618042,
      "learning_rate": 3.227313441642288e-08,
      "loss": 0.5306,
      "step": 15230
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0605419874191284,
      "learning_rate": 3.2156854993229955e-08,
      "loss": 0.4544,
      "step": 15231
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1065747737884521,
      "learning_rate": 3.2040784748629814e-08,
      "loss": 0.4977,
      "step": 15232
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1279810667037964,
      "learning_rate": 3.192492368750966e-08,
      "loss": 0.4958,
      "step": 15233
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0252561569213867,
      "learning_rate": 3.180927181474891e-08,
      "loss": 0.5373,
      "step": 15234
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1200501918792725,
      "learning_rate": 3.16938291352159e-08,
      "loss": 0.5131,
      "step": 15235
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9503628611564636,
      "learning_rate": 3.157859565377286e-08,
      "loss": 0.4604,
      "step": 15236
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0107611417770386,
      "learning_rate": 3.146357137527145e-08,
      "loss": 0.4503,
      "step": 15237
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0233523845672607,
      "learning_rate": 3.1348756304554475e-08,
      "loss": 0.5112,
      "step": 15238
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.056778907775879,
      "learning_rate": 3.123415044645639e-08,
      "loss": 0.5001,
      "step": 15239
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0508291721343994,
      "learning_rate": 3.111975380580334e-08,
      "loss": 0.4796,
      "step": 15240
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0572813749313354,
      "learning_rate": 3.100556638741203e-08,
      "loss": 0.53,
      "step": 15241
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0045342445373535,
      "learning_rate": 3.089158819609084e-08,
      "loss": 0.4893,
      "step": 15242
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0140489339828491,
      "learning_rate": 3.077781923663814e-08,
      "loss": 0.5306,
      "step": 15243
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0265076160430908,
      "learning_rate": 3.066425951384455e-08,
      "loss": 0.4269,
      "step": 15244
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0639636516571045,
      "learning_rate": 3.055090903249236e-08,
      "loss": 0.5385,
      "step": 15245
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0341408252716064,
      "learning_rate": 3.0437767797353856e-08,
      "loss": 0.477,
      "step": 15246
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0694355964660645,
      "learning_rate": 3.032483581319301e-08,
      "loss": 0.5101,
      "step": 15247
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0831514596939087,
      "learning_rate": 3.021211308476546e-08,
      "loss": 0.497,
      "step": 15248
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.054814100265503,
      "learning_rate": 3.0099599616816856e-08,
      "loss": 0.4966,
      "step": 15249
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.145595669746399,
      "learning_rate": 2.998729541408507e-08,
      "loss": 0.5221,
      "step": 15250
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0958560705184937,
      "learning_rate": 2.987520048129911e-08,
      "loss": 0.5108,
      "step": 15251
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0394368171691895,
      "learning_rate": 2.976331482317796e-08,
      "loss": 0.4922,
      "step": 15252
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0591411590576172,
      "learning_rate": 2.9651638444434528e-08,
      "loss": 0.5281,
      "step": 15253
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0195757150650024,
      "learning_rate": 2.9540171349769497e-08,
      "loss": 0.5176,
      "step": 15254
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0487700700759888,
      "learning_rate": 2.942891354387689e-08,
      "loss": 0.49,
      "step": 15255
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0553514957427979,
      "learning_rate": 2.9317865031441295e-08,
      "loss": 0.5202,
      "step": 15256
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1601183414459229,
      "learning_rate": 2.920702581713841e-08,
      "loss": 0.4933,
      "step": 15257
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.051039457321167,
      "learning_rate": 2.909639590563562e-08,
      "loss": 0.5315,
      "step": 15258
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0642770528793335,
      "learning_rate": 2.8985975301591975e-08,
      "loss": 0.4957,
      "step": 15259
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.054998517036438,
      "learning_rate": 2.887576400965486e-08,
      "loss": 0.4657,
      "step": 15260
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9867322444915771,
      "learning_rate": 2.8765762034466682e-08,
      "loss": 0.4807,
      "step": 15261
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.082873821258545,
      "learning_rate": 2.8655969380658177e-08,
      "loss": 0.4793,
      "step": 15262
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.129400372505188,
      "learning_rate": 2.8546386052853427e-08,
      "loss": 0.4728,
      "step": 15263
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0983253717422485,
      "learning_rate": 2.8437012055665403e-08,
      "loss": 0.557,
      "step": 15264
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0699577331542969,
      "learning_rate": 2.832784739369987e-08,
      "loss": 0.5156,
      "step": 15265
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0261139869689941,
      "learning_rate": 2.8218892071553705e-08,
      "loss": 0.4538,
      "step": 15266
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.04954993724823,
      "learning_rate": 2.8110146093814906e-08,
      "loss": 0.4877,
      "step": 15267
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0849642753601074,
      "learning_rate": 2.8001609465061474e-08,
      "loss": 0.508,
      "step": 15268
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.011203408241272,
      "learning_rate": 2.7893282189863647e-08,
      "loss": 0.497,
      "step": 15269
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0393261909484863,
      "learning_rate": 2.7785164272783327e-08,
      "loss": 0.4831,
      "step": 15270
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1217502355575562,
      "learning_rate": 2.7677255718372986e-08,
      "loss": 0.4889,
      "step": 15271
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.070006012916565,
      "learning_rate": 2.7569556531175657e-08,
      "loss": 0.5044,
      "step": 15272
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.097985863685608,
      "learning_rate": 2.7462066715726045e-08,
      "loss": 0.5416,
      "step": 15273
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1098628044128418,
      "learning_rate": 2.7354786276551083e-08,
      "loss": 0.5338,
      "step": 15274
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2087064981460571,
      "learning_rate": 2.7247715218167714e-08,
      "loss": 0.526,
      "step": 15275
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1480826139450073,
      "learning_rate": 2.7140853545083447e-08,
      "loss": 0.4955,
      "step": 15276
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.115615963935852,
      "learning_rate": 2.703420126179912e-08,
      "loss": 0.5082,
      "step": 15277
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1329855918884277,
      "learning_rate": 2.692775837280448e-08,
      "loss": 0.5264,
      "step": 15278
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0670102834701538,
      "learning_rate": 2.682152488258205e-08,
      "loss": 0.5167,
      "step": 15279
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.010777473449707,
      "learning_rate": 2.671550079560492e-08,
      "loss": 0.4553,
      "step": 15280
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0586128234863281,
      "learning_rate": 2.6609686116337296e-08,
      "loss": 0.4824,
      "step": 15281
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1230065822601318,
      "learning_rate": 2.6504080849234504e-08,
      "loss": 0.5393,
      "step": 15282
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9813792109489441,
      "learning_rate": 2.6398684998742986e-08,
      "loss": 0.4788,
      "step": 15283
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0060399770736694,
      "learning_rate": 2.629349856930141e-08,
      "loss": 0.4565,
      "step": 15284
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.101292610168457,
      "learning_rate": 2.6188521565338466e-08,
      "loss": 0.4982,
      "step": 15285
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0829226970672607,
      "learning_rate": 2.608375399127394e-08,
      "loss": 0.4897,
      "step": 15286
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9973019957542419,
      "learning_rate": 2.5979195851519313e-08,
      "loss": 0.459,
      "step": 15287
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0220048427581787,
      "learning_rate": 2.5874847150477722e-08,
      "loss": 0.5127,
      "step": 15288
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.011262059211731,
      "learning_rate": 2.5770707892542878e-08,
      "loss": 0.4793,
      "step": 15289
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0880519151687622,
      "learning_rate": 2.5666778082099052e-08,
      "loss": 0.4742,
      "step": 15290
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.17592191696167,
      "learning_rate": 2.5563057723522742e-08,
      "loss": 0.5033,
      "step": 15291
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1462591886520386,
      "learning_rate": 2.5459546821181014e-08,
      "loss": 0.5171,
      "step": 15292
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9982767701148987,
      "learning_rate": 2.5356245379433154e-08,
      "loss": 0.4854,
      "step": 15293
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.009433627128601,
      "learning_rate": 2.525315340262846e-08,
      "loss": 0.4707,
      "step": 15294
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0816377401351929,
      "learning_rate": 2.51502708951068e-08,
      "loss": 0.5397,
      "step": 15295
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0993826389312744,
      "learning_rate": 2.5047597861201368e-08,
      "loss": 0.504,
      "step": 15296
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.122497320175171,
      "learning_rate": 2.494513430523482e-08,
      "loss": 0.4931,
      "step": 15297
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1604719161987305,
      "learning_rate": 2.4842880231522038e-08,
      "loss": 0.4961,
      "step": 15298
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0055816173553467,
      "learning_rate": 2.474083564436791e-08,
      "loss": 0.4649,
      "step": 15299
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0866940021514893,
      "learning_rate": 2.4639000548070114e-08,
      "loss": 0.5276,
      "step": 15300
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1128740310668945,
      "learning_rate": 2.4537374946915215e-08,
      "loss": 0.4593,
      "step": 15301
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0929911136627197,
      "learning_rate": 2.4435958845183684e-08,
      "loss": 0.4883,
      "step": 15302
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1067675352096558,
      "learning_rate": 2.4334752247145433e-08,
      "loss": 0.4879,
      "step": 15303
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9623573422431946,
      "learning_rate": 2.4233755157060944e-08,
      "loss": 0.5029,
      "step": 15304
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0398600101470947,
      "learning_rate": 2.413296757918404e-08,
      "loss": 0.4967,
      "step": 15305
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0774849653244019,
      "learning_rate": 2.4032389517757993e-08,
      "loss": 0.5044,
      "step": 15306
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9469539523124695,
      "learning_rate": 2.3932020977017745e-08,
      "loss": 0.4557,
      "step": 15307
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9823768734931946,
      "learning_rate": 2.3831861961189917e-08,
      "loss": 0.4771,
      "step": 15308
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0540803670883179,
      "learning_rate": 2.3731912474491137e-08,
      "loss": 0.4739,
      "step": 15309
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9758304953575134,
      "learning_rate": 2.3632172521130815e-08,
      "loss": 0.4468,
      "step": 15310
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9877710342407227,
      "learning_rate": 2.3532642105307813e-08,
      "loss": 0.4662,
      "step": 15311
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0869003534317017,
      "learning_rate": 2.3433321231213778e-08,
      "loss": 0.495,
      "step": 15312
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1220903396606445,
      "learning_rate": 2.3334209903029815e-08,
      "loss": 0.5294,
      "step": 15313
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.099942684173584,
      "learning_rate": 2.3235308124930357e-08,
      "loss": 0.5059,
      "step": 15314
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0976247787475586,
      "learning_rate": 2.3136615901078742e-08,
      "loss": 0.5256,
      "step": 15315
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.060356616973877,
      "learning_rate": 2.303813323563109e-08,
      "loss": 0.5001,
      "step": 15316
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.061465859413147,
      "learning_rate": 2.2939860132734084e-08,
      "loss": 0.475,
      "step": 15317
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9800360202789307,
      "learning_rate": 2.2841796596525522e-08,
      "loss": 0.5049,
      "step": 15318
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.152793049812317,
      "learning_rate": 2.2743942631134886e-08,
      "loss": 0.4839,
      "step": 15319
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2103699445724487,
      "learning_rate": 2.264629824068165e-08,
      "loss": 0.5802,
      "step": 15320
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1103157997131348,
      "learning_rate": 2.2548863429278645e-08,
      "loss": 0.5396,
      "step": 15321
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.095953106880188,
      "learning_rate": 2.2451638201027026e-08,
      "loss": 0.5135,
      "step": 15322
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1200344562530518,
      "learning_rate": 2.2354622560021854e-08,
      "loss": 0.4938,
      "step": 15323
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0356738567352295,
      "learning_rate": 2.2257816510347086e-08,
      "loss": 0.4789,
      "step": 15324
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0751638412475586,
      "learning_rate": 2.2161220056079457e-08,
      "loss": 0.4692,
      "step": 15325
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0484387874603271,
      "learning_rate": 2.2064833201286827e-08,
      "loss": 0.4921,
      "step": 15326
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0133472681045532,
      "learning_rate": 2.1968655950026508e-08,
      "loss": 0.5265,
      "step": 15327
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0596994161605835,
      "learning_rate": 2.187268830634859e-08,
      "loss": 0.4993,
      "step": 15328
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0807329416275024,
      "learning_rate": 2.1776930274294283e-08,
      "loss": 0.4878,
      "step": 15329
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0925438404083252,
      "learning_rate": 2.1681381857895923e-08,
      "loss": 0.4969,
      "step": 15330
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0323103666305542,
      "learning_rate": 2.1586043061175842e-08,
      "loss": 0.4788,
      "step": 15331
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0427535772323608,
      "learning_rate": 2.1490913888149166e-08,
      "loss": 0.533,
      "step": 15332
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1524258852005005,
      "learning_rate": 2.139599434282047e-08,
      "loss": 0.5057,
      "step": 15333
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1330947875976562,
      "learning_rate": 2.130128442918766e-08,
      "loss": 0.5622,
      "step": 15334
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1543328762054443,
      "learning_rate": 2.1206784151238113e-08,
      "loss": 0.5027,
      "step": 15335
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.051921010017395,
      "learning_rate": 2.111249351295086e-08,
      "loss": 0.543,
      "step": 15336
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0333747863769531,
      "learning_rate": 2.1018412518296617e-08,
      "loss": 0.508,
      "step": 15337
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9350318908691406,
      "learning_rate": 2.0924541171235545e-08,
      "loss": 0.4936,
      "step": 15338
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.206451416015625,
      "learning_rate": 2.083087947572171e-08,
      "loss": 0.4999,
      "step": 15339
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.131523847579956,
      "learning_rate": 2.073742743569862e-08,
      "loss": 0.5002,
      "step": 15340
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0815268754959106,
      "learning_rate": 2.0644185055100352e-08,
      "loss": 0.5543,
      "step": 15341
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.994455873966217,
      "learning_rate": 2.0551152337853208e-08,
      "loss": 0.4812,
      "step": 15342
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.13554847240448,
      "learning_rate": 2.0458329287875168e-08,
      "loss": 0.5011,
      "step": 15343
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1019952297210693,
      "learning_rate": 2.0365715909074213e-08,
      "loss": 0.5145,
      "step": 15344
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0579437017440796,
      "learning_rate": 2.027331220535056e-08,
      "loss": 0.5206,
      "step": 15345
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0751632452011108,
      "learning_rate": 2.018111818059387e-08,
      "loss": 0.4207,
      "step": 15346
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1553031206130981,
      "learning_rate": 2.008913383868716e-08,
      "loss": 0.5421,
      "step": 15347
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.113647222518921,
      "learning_rate": 1.999735918350343e-08,
      "loss": 0.5173,
      "step": 15348
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0515217781066895,
      "learning_rate": 1.990579421890626e-08,
      "loss": 0.5192,
      "step": 15349
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0645197629928589,
      "learning_rate": 1.9814438948751458e-08,
      "loss": 0.5053,
      "step": 15350
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0292072296142578,
      "learning_rate": 1.9723293376886497e-08,
      "loss": 0.4778,
      "step": 15351
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.040724277496338,
      "learning_rate": 1.963235750714776e-08,
      "loss": 0.5397,
      "step": 15352
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0659323930740356,
      "learning_rate": 1.9541631343365507e-08,
      "loss": 0.5223,
      "step": 15353
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2221729755401611,
      "learning_rate": 1.9451114889359468e-08,
      "loss": 0.5599,
      "step": 15354
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0946619510650635,
      "learning_rate": 1.936080814894048e-08,
      "loss": 0.499,
      "step": 15355
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0898137092590332,
      "learning_rate": 1.9270711125912167e-08,
      "loss": 0.4925,
      "step": 15356
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0585594177246094,
      "learning_rate": 1.9180823824067053e-08,
      "loss": 0.493,
      "step": 15357
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9604061245918274,
      "learning_rate": 1.909114624719044e-08,
      "loss": 0.5012,
      "step": 15358
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1087369918823242,
      "learning_rate": 1.90016783990582e-08,
      "loss": 0.4827,
      "step": 15359
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0932306051254272,
      "learning_rate": 1.891242028343787e-08,
      "loss": 0.5323,
      "step": 15360
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0271295309066772,
      "learning_rate": 1.8823371904087563e-08,
      "loss": 0.4937,
      "step": 15361
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0557682514190674,
      "learning_rate": 1.8734533264757047e-08,
      "loss": 0.4682,
      "step": 15362
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1083639860153198,
      "learning_rate": 1.864590436918612e-08,
      "loss": 0.5375,
      "step": 15363
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1011896133422852,
      "learning_rate": 1.8557485221107897e-08,
      "loss": 0.4959,
      "step": 15364
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1148617267608643,
      "learning_rate": 1.8469275824244958e-08,
      "loss": 0.4878,
      "step": 15365
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1544920206069946,
      "learning_rate": 1.8381276182311004e-08,
      "loss": 0.4704,
      "step": 15366
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1003879308700562,
      "learning_rate": 1.8293486299011398e-08,
      "loss": 0.4997,
      "step": 15367
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.06473970413208,
      "learning_rate": 1.8205906178043186e-08,
      "loss": 0.5041,
      "step": 15368
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0241307020187378,
      "learning_rate": 1.811853582309453e-08,
      "loss": 0.5289,
      "step": 15369
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0392919778823853,
      "learning_rate": 1.803137523784304e-08,
      "loss": 0.4512,
      "step": 15370
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.028462290763855,
      "learning_rate": 1.7944424425959116e-08,
      "loss": 0.5017,
      "step": 15371
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9891314506530762,
      "learning_rate": 1.7857683391104273e-08,
      "loss": 0.4665,
      "step": 15372
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0323691368103027,
      "learning_rate": 1.7771152136931147e-08,
      "loss": 0.5099,
      "step": 15373
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1690553426742554,
      "learning_rate": 1.7684830667082377e-08,
      "loss": 0.5036,
      "step": 15374
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.048149824142456,
      "learning_rate": 1.759871898519394e-08,
      "loss": 0.4901,
      "step": 15375
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0809240341186523,
      "learning_rate": 1.7512817094890167e-08,
      "loss": 0.4707,
      "step": 15376
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0636286735534668,
      "learning_rate": 1.742712499978927e-08,
      "loss": 0.4795,
      "step": 15377
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0688414573669434,
      "learning_rate": 1.734164270349892e-08,
      "loss": 0.4704,
      "step": 15378
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0488619804382324,
      "learning_rate": 1.7256370209618458e-08,
      "loss": 0.5188,
      "step": 15379
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0834404230117798,
      "learning_rate": 1.71713075217389e-08,
      "loss": 0.5152,
      "step": 15380
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1331005096435547,
      "learning_rate": 1.7086454643441273e-08,
      "loss": 0.5231,
      "step": 15381
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0595972537994385,
      "learning_rate": 1.7001811578298832e-08,
      "loss": 0.4447,
      "step": 15382
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0525641441345215,
      "learning_rate": 1.6917378329875946e-08,
      "loss": 0.488,
      "step": 15383
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9919252991676331,
      "learning_rate": 1.6833154901726988e-08,
      "loss": 0.5188,
      "step": 15384
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0833450555801392,
      "learning_rate": 1.6749141297398574e-08,
      "loss": 0.4768,
      "step": 15385
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0467420816421509,
      "learning_rate": 1.6665337520428427e-08,
      "loss": 0.4613,
      "step": 15386
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0288331508636475,
      "learning_rate": 1.658174357434483e-08,
      "loss": 0.5197,
      "step": 15387
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1571677923202515,
      "learning_rate": 1.649835946266831e-08,
      "loss": 0.551,
      "step": 15388
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0284770727157593,
      "learning_rate": 1.6415185188909944e-08,
      "loss": 0.471,
      "step": 15389
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0496933460235596,
      "learning_rate": 1.6332220756570815e-08,
      "loss": 0.4814,
      "step": 15390
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.076085090637207,
      "learning_rate": 1.6249466169145354e-08,
      "loss": 0.5407,
      "step": 15391
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.032431960105896,
      "learning_rate": 1.6166921430118e-08,
      "loss": 0.4832,
      "step": 15392
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.089762806892395,
      "learning_rate": 1.608458654296319e-08,
      "loss": 0.4834,
      "step": 15393
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0818616151809692,
      "learning_rate": 1.600246151114926e-08,
      "loss": 0.4751,
      "step": 15394
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0302743911743164,
      "learning_rate": 1.5920546338133447e-08,
      "loss": 0.4924,
      "step": 15395
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0723204612731934,
      "learning_rate": 1.5838841027365215e-08,
      "loss": 0.5048,
      "step": 15396
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.052377462387085,
      "learning_rate": 1.5757345582285144e-08,
      "loss": 0.5257,
      "step": 15397
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.065084457397461,
      "learning_rate": 1.5676060006323267e-08,
      "loss": 0.5565,
      "step": 15398
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.114837884902954,
      "learning_rate": 1.559498430290407e-08,
      "loss": 0.4829,
      "step": 15399
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.132448434829712,
      "learning_rate": 1.5514118475440378e-08,
      "loss": 0.488,
      "step": 15400
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9648688435554504,
      "learning_rate": 1.5433462527337793e-08,
      "loss": 0.4395,
      "step": 15401
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0886834859848022,
      "learning_rate": 1.5353016461991387e-08,
      "loss": 0.5493,
      "step": 15402
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0746283531188965,
      "learning_rate": 1.5272780282789556e-08,
      "loss": 0.5297,
      "step": 15403
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2116729021072388,
      "learning_rate": 1.5192753993110155e-08,
      "loss": 0.4665,
      "step": 15404
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0766348838806152,
      "learning_rate": 1.5112937596323263e-08,
      "loss": 0.5302,
      "step": 15405
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0860892534255981,
      "learning_rate": 1.5033331095788973e-08,
      "loss": 0.4866,
      "step": 15406
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9706477522850037,
      "learning_rate": 1.4953934494860155e-08,
      "loss": 0.4174,
      "step": 15407
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.033424735069275,
      "learning_rate": 1.4874747796879142e-08,
      "loss": 0.4624,
      "step": 15408
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.146001935005188,
      "learning_rate": 1.4795771005181036e-08,
      "loss": 0.5519,
      "step": 15409
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0574760437011719,
      "learning_rate": 1.4717004123090406e-08,
      "loss": 0.4848,
      "step": 15410
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0617190599441528,
      "learning_rate": 1.463844715392404e-08,
      "loss": 0.5468,
      "step": 15411
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9709962010383606,
      "learning_rate": 1.4560100100989849e-08,
      "loss": 0.4465,
      "step": 15412
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0188851356506348,
      "learning_rate": 1.448196296758686e-08,
      "loss": 0.4785,
      "step": 15413
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0532549619674683,
      "learning_rate": 1.4404035757005219e-08,
      "loss": 0.4913,
      "step": 15414
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1245558261871338,
      "learning_rate": 1.4326318472525635e-08,
      "loss": 0.4962,
      "step": 15415
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0558838844299316,
      "learning_rate": 1.4248811117421046e-08,
      "loss": 0.5147,
      "step": 15416
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1222167015075684,
      "learning_rate": 1.4171513694954953e-08,
      "loss": 0.4893,
      "step": 15417
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1024917364120483,
      "learning_rate": 1.4094426208381972e-08,
      "loss": 0.5539,
      "step": 15418
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1723283529281616,
      "learning_rate": 1.4017548660947844e-08,
      "loss": 0.5152,
      "step": 15419
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.086130976676941,
      "learning_rate": 1.3940881055889976e-08,
      "loss": 0.5141,
      "step": 15420
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.167900562286377,
      "learning_rate": 1.3864423396436344e-08,
      "loss": 0.4828,
      "step": 15421
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0070527791976929,
      "learning_rate": 1.3788175685806594e-08,
      "loss": 0.5074,
      "step": 15422
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.103872299194336,
      "learning_rate": 1.3712137927210377e-08,
      "loss": 0.5252,
      "step": 15423
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1018625497817993,
      "learning_rate": 1.3636310123850694e-08,
      "loss": 0.5367,
      "step": 15424
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.981019139289856,
      "learning_rate": 1.3560692278919429e-08,
      "loss": 0.4795,
      "step": 15425
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0430066585540771,
      "learning_rate": 1.3485284395600707e-08,
      "loss": 0.4652,
      "step": 15426
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0918257236480713,
      "learning_rate": 1.3410086477069761e-08,
      "loss": 0.5407,
      "step": 15427
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0566918849945068,
      "learning_rate": 1.333509852649295e-08,
      "loss": 0.4817,
      "step": 15428
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0845723152160645,
      "learning_rate": 1.3260320547028305e-08,
      "loss": 0.5032,
      "step": 15429
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0401231050491333,
      "learning_rate": 1.3185752541823304e-08,
      "loss": 0.4924,
      "step": 15430
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.166698932647705,
      "learning_rate": 1.3111394514018772e-08,
      "loss": 0.4758,
      "step": 15431
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0580391883850098,
      "learning_rate": 1.3037246466745535e-08,
      "loss": 0.4882,
      "step": 15432
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.099125862121582,
      "learning_rate": 1.2963308403124985e-08,
      "loss": 0.4994,
      "step": 15433
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1514862775802612,
      "learning_rate": 1.2889580326271301e-08,
      "loss": 0.5011,
      "step": 15434
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0589500665664673,
      "learning_rate": 1.2816062239288107e-08,
      "loss": 0.4675,
      "step": 15435
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.115246057510376,
      "learning_rate": 1.2742754145271264e-08,
      "loss": 0.4854,
      "step": 15436
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0737168788909912,
      "learning_rate": 1.2669656047308299e-08,
      "loss": 0.4983,
      "step": 15437
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0704424381256104,
      "learning_rate": 1.2596767948476196e-08,
      "loss": 0.533,
      "step": 15438
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1217480897903442,
      "learning_rate": 1.2524089851844168e-08,
      "loss": 0.4673,
      "step": 15439
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.004804253578186,
      "learning_rate": 1.2451621760472544e-08,
      "loss": 0.4808,
      "step": 15440
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1652071475982666,
      "learning_rate": 1.237936367741277e-08,
      "loss": 0.4964,
      "step": 15441
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1194581985473633,
      "learning_rate": 1.2307315605707416e-08,
      "loss": 0.5033,
      "step": 15442
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.096777081489563,
      "learning_rate": 1.2235477548390162e-08,
      "loss": 0.5177,
      "step": 15443
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9652280807495117,
      "learning_rate": 1.2163849508485259e-08,
      "loss": 0.4436,
      "step": 15444
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0869945287704468,
      "learning_rate": 1.2092431489009738e-08,
      "loss": 0.5026,
      "step": 15445
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9712923169136047,
      "learning_rate": 1.202122349297008e-08,
      "loss": 0.4694,
      "step": 15446
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1346890926361084,
      "learning_rate": 1.1950225523365e-08,
      "loss": 0.5372,
      "step": 15447
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0068241357803345,
      "learning_rate": 1.1879437583183217e-08,
      "loss": 0.4844,
      "step": 15448
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0190941095352173,
      "learning_rate": 1.1808859675406236e-08,
      "loss": 0.5136,
      "step": 15449
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1415479183197021,
      "learning_rate": 1.173849180300557e-08,
      "loss": 0.4895,
      "step": 15450
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.080260992050171,
      "learning_rate": 1.16683339689444e-08,
      "loss": 0.5392,
      "step": 15451
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0502970218658447,
      "learning_rate": 1.1598386176175924e-08,
      "loss": 0.4963,
      "step": 15452
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0288969278335571,
      "learning_rate": 1.1528648427646671e-08,
      "loss": 0.4612,
      "step": 15453
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0865890979766846,
      "learning_rate": 1.1459120726292072e-08,
      "loss": 0.4803,
      "step": 15454
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.079679012298584,
      "learning_rate": 1.1389803075039785e-08,
      "loss": 0.4945,
      "step": 15455
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.050313949584961,
      "learning_rate": 1.1320695476809141e-08,
      "loss": 0.5243,
      "step": 15456
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.089991807937622,
      "learning_rate": 1.1251797934509478e-08,
      "loss": 0.5044,
      "step": 15457
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.107649326324463,
      "learning_rate": 1.1183110451042368e-08,
      "loss": 0.4493,
      "step": 15458
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0357794761657715,
      "learning_rate": 1.1114633029299382e-08,
      "loss": 0.4632,
      "step": 15459
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9970534443855286,
      "learning_rate": 1.1046365672163772e-08,
      "loss": 0.4763,
      "step": 15460
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.08147132396698,
      "learning_rate": 1.0978308382511016e-08,
      "loss": 0.5183,
      "step": 15461
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.112913966178894,
      "learning_rate": 1.0910461163206043e-08,
      "loss": 0.4865,
      "step": 15462
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2007641792297363,
      "learning_rate": 1.0842824017105458e-08,
      "loss": 0.5015,
      "step": 15463
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9819358587265015,
      "learning_rate": 1.0775396947057537e-08,
      "loss": 0.47,
      "step": 15464
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0319911241531372,
      "learning_rate": 1.0708179955901677e-08,
      "loss": 0.4666,
      "step": 15465
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0933454036712646,
      "learning_rate": 1.0641173046467833e-08,
      "loss": 0.5157,
      "step": 15466
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.100706696510315,
      "learning_rate": 1.0574376221577642e-08,
      "loss": 0.4744,
      "step": 15467
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0418654680252075,
      "learning_rate": 1.0507789484043295e-08,
      "loss": 0.4367,
      "step": 15468
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9827783107757568,
      "learning_rate": 1.0441412836668663e-08,
      "loss": 0.4627,
      "step": 15469
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.021515965461731,
      "learning_rate": 1.037524628224873e-08,
      "loss": 0.477,
      "step": 15470
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0083973407745361,
      "learning_rate": 1.0309289823569601e-08,
      "loss": 0.4883,
      "step": 15471
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1818205118179321,
      "learning_rate": 1.02435434634085e-08,
      "loss": 0.514,
      "step": 15472
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.253869652748108,
      "learning_rate": 1.0178007204533768e-08,
      "loss": 0.5216,
      "step": 15473
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1360281705856323,
      "learning_rate": 1.0112681049704865e-08,
      "loss": 0.5095,
      "step": 15474
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9967944622039795,
      "learning_rate": 1.004756500167181e-08,
      "loss": 0.4683,
      "step": 15475
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.064879059791565,
      "learning_rate": 9.982659063177413e-09,
      "loss": 0.4966,
      "step": 15476
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.049710988998413,
      "learning_rate": 9.917963236954487e-09,
      "loss": 0.4757,
      "step": 15477
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1126371622085571,
      "learning_rate": 9.853477525726962e-09,
      "loss": 0.4715,
      "step": 15478
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0898528099060059,
      "learning_rate": 9.789201932209335e-09,
      "loss": 0.5257,
      "step": 15479
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1489167213439941,
      "learning_rate": 9.725136459109441e-09,
      "loss": 0.5702,
      "step": 15480
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0590689182281494,
      "learning_rate": 9.66128110912401e-09,
      "loss": 0.4812,
      "step": 15481
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9705673456192017,
      "learning_rate": 9.597635884941447e-09,
      "loss": 0.4751,
      "step": 15482
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9974306225776672,
      "learning_rate": 9.534200789242388e-09,
      "loss": 0.5068,
      "step": 15483
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0707036256790161,
      "learning_rate": 9.470975824698025e-09,
      "loss": 0.5301,
      "step": 15484
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.145838737487793,
      "learning_rate": 9.407960993969567e-09,
      "loss": 0.5452,
      "step": 15485
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9681414365768433,
      "learning_rate": 9.345156299711e-09,
      "loss": 0.4709,
      "step": 15486
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0441086292266846,
      "learning_rate": 9.282561744566321e-09,
      "loss": 0.4832,
      "step": 15487
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0421568155288696,
      "learning_rate": 9.220177331172309e-09,
      "loss": 0.4879,
      "step": 15488
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0118240118026733,
      "learning_rate": 9.158003062154642e-09,
      "loss": 0.4499,
      "step": 15489
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0902498960494995,
      "learning_rate": 9.096038940131225e-09,
      "loss": 0.5196,
      "step": 15490
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9824011325836182,
      "learning_rate": 9.034284967711637e-09,
      "loss": 0.4471,
      "step": 15491
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0274479389190674,
      "learning_rate": 8.972741147496023e-09,
      "loss": 0.5007,
      "step": 15492
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0410804748535156,
      "learning_rate": 8.911407482076196e-09,
      "loss": 0.4576,
      "step": 15493
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0497487783432007,
      "learning_rate": 8.85028397403398e-09,
      "loss": 0.4976,
      "step": 15494
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.106119990348816,
      "learning_rate": 8.789370625943427e-09,
      "loss": 0.5142,
      "step": 15495
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0978714227676392,
      "learning_rate": 8.728667440369153e-09,
      "loss": 0.4892,
      "step": 15496
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0782625675201416,
      "learning_rate": 8.668174419867449e-09,
      "loss": 0.5343,
      "step": 15497
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.124019980430603,
      "learning_rate": 8.60789156698516e-09,
      "loss": 0.4786,
      "step": 15498
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0971437692642212,
      "learning_rate": 8.547818884260816e-09,
      "loss": 0.5118,
      "step": 15499
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0221912860870361,
      "learning_rate": 8.48795637422406e-09,
      "loss": 0.4627,
      "step": 15500
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0534876585006714,
      "learning_rate": 8.428304039395096e-09,
      "loss": 0.5088,
      "step": 15501
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0663495063781738,
      "learning_rate": 8.368861882285806e-09,
      "loss": 0.4988,
      "step": 15502
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0165753364562988,
      "learning_rate": 8.309629905399186e-09,
      "loss": 0.468,
      "step": 15503
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.038854956626892,
      "learning_rate": 8.250608111229352e-09,
      "loss": 0.5189,
      "step": 15504
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1807208061218262,
      "learning_rate": 8.191796502260985e-09,
      "loss": 0.5431,
      "step": 15505
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1552495956420898,
      "learning_rate": 8.13319508097099e-09,
      "loss": 0.5579,
      "step": 15506
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1485646963119507,
      "learning_rate": 8.074803849827395e-09,
      "loss": 0.5058,
      "step": 15507
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0058408975601196,
      "learning_rate": 8.016622811287123e-09,
      "loss": 0.4613,
      "step": 15508
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9777622818946838,
      "learning_rate": 7.958651967801545e-09,
      "loss": 0.4582,
      "step": 15509
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.062703251838684,
      "learning_rate": 7.900891321810932e-09,
      "loss": 0.4902,
      "step": 15510
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1520791053771973,
      "learning_rate": 7.843340875747785e-09,
      "loss": 0.5539,
      "step": 15511
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.029168725013733,
      "learning_rate": 7.786000632035163e-09,
      "loss": 0.4889,
      "step": 15512
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0008456707000732,
      "learning_rate": 7.728870593087246e-09,
      "loss": 0.5062,
      "step": 15513
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.185447096824646,
      "learning_rate": 7.671950761309333e-09,
      "loss": 0.4949,
      "step": 15514
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1570756435394287,
      "learning_rate": 7.61524113909895e-09,
      "loss": 0.4382,
      "step": 15515
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0614207983016968,
      "learning_rate": 7.558741728843633e-09,
      "loss": 0.4664,
      "step": 15516
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.094365119934082,
      "learning_rate": 7.502452532922033e-09,
      "loss": 0.4727,
      "step": 15517
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0895835161209106,
      "learning_rate": 7.446373553705033e-09,
      "loss": 0.525,
      "step": 15518
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.09317946434021,
      "learning_rate": 7.390504793552966e-09,
      "loss": 0.5406,
      "step": 15519
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0369621515274048,
      "learning_rate": 7.3348462548183955e-09,
      "loss": 0.5135,
      "step": 15520
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1538317203521729,
      "learning_rate": 7.279397939845556e-09,
      "loss": 0.5053,
      "step": 15521
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.189522624015808,
      "learning_rate": 7.2241598509686926e-09,
      "loss": 0.515,
      "step": 15522
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0553511381149292,
      "learning_rate": 7.169131990514278e-09,
      "loss": 0.5348,
      "step": 15523
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.105460524559021,
      "learning_rate": 7.114314360798791e-09,
      "loss": 0.5216,
      "step": 15524
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0487605333328247,
      "learning_rate": 7.0597069641303865e-09,
      "loss": 0.5039,
      "step": 15525
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0938303470611572,
      "learning_rate": 7.00530980280889e-09,
      "loss": 0.4823,
      "step": 15526
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0104894638061523,
      "learning_rate": 6.951122879124139e-09,
      "loss": 0.4778,
      "step": 15527
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9984012246131897,
      "learning_rate": 6.89714619535764e-09,
      "loss": 0.4719,
      "step": 15528
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0359838008880615,
      "learning_rate": 6.84337975378313e-09,
      "loss": 0.4591,
      "step": 15529
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0148361921310425,
      "learning_rate": 6.789823556663799e-09,
      "loss": 0.4995,
      "step": 15530
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0572915077209473,
      "learning_rate": 6.736477606255065e-09,
      "loss": 0.4604,
      "step": 15531
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.04763662815094,
      "learning_rate": 6.683341904802909e-09,
      "loss": 0.491,
      "step": 15532
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1820831298828125,
      "learning_rate": 6.63041645454443e-09,
      "loss": 0.4814,
      "step": 15533
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1502306461334229,
      "learning_rate": 6.577701257708957e-09,
      "loss": 0.5676,
      "step": 15534
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0856949090957642,
      "learning_rate": 6.52519631651527e-09,
      "loss": 0.4669,
      "step": 15535
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1188011169433594,
      "learning_rate": 6.4729016331749325e-09,
      "loss": 0.5111,
      "step": 15536
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0615599155426025,
      "learning_rate": 6.420817209888963e-09,
      "loss": 0.4751,
      "step": 15537
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0602856874465942,
      "learning_rate": 6.368943048851162e-09,
      "loss": 0.4879,
      "step": 15538
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.146788477897644,
      "learning_rate": 6.317279152245892e-09,
      "loss": 0.4714,
      "step": 15539
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1277881860733032,
      "learning_rate": 6.265825522248082e-09,
      "loss": 0.5061,
      "step": 15540
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0490295886993408,
      "learning_rate": 6.2145821610243296e-09,
      "loss": 0.5056,
      "step": 15541
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1133843660354614,
      "learning_rate": 6.163549070732356e-09,
      "loss": 0.5406,
      "step": 15542
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1436570882797241,
      "learning_rate": 6.1127262535209955e-09,
      "loss": 0.5567,
      "step": 15543
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1176104545593262,
      "learning_rate": 6.062113711530204e-09,
      "loss": 0.5033,
      "step": 15544
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0910592079162598,
      "learning_rate": 6.01171144689161e-09,
      "loss": 0.4846,
      "step": 15545
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.08412766456604,
      "learning_rate": 5.96151946172685e-09,
      "loss": 0.5403,
      "step": 15546
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0198564529418945,
      "learning_rate": 5.911537758149233e-09,
      "loss": 0.4694,
      "step": 15547
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.014939785003662,
      "learning_rate": 5.861766338263741e-09,
      "loss": 0.4924,
      "step": 15548
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.121074914932251,
      "learning_rate": 5.812205204165922e-09,
      "loss": 0.4822,
      "step": 15549
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1740344762802124,
      "learning_rate": 5.762854357942993e-09,
      "loss": 0.538,
      "step": 15550
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0074889659881592,
      "learning_rate": 5.7137138016721825e-09,
      "loss": 0.4615,
      "step": 15551
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9784232974052429,
      "learning_rate": 5.6647835374229465e-09,
      "loss": 0.5068,
      "step": 15552
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0445855855941772,
      "learning_rate": 5.616063567255859e-09,
      "loss": 0.5426,
      "step": 15553
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1089900732040405,
      "learning_rate": 5.5675538932220555e-09,
      "loss": 0.5565,
      "step": 15554
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1083252429962158,
      "learning_rate": 5.519254517364347e-09,
      "loss": 0.462,
      "step": 15555
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0948292016983032,
      "learning_rate": 5.471165441716108e-09,
      "loss": 0.4796,
      "step": 15556
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0887798070907593,
      "learning_rate": 5.4232866683023856e-09,
      "loss": 0.5145,
      "step": 15557
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0607532262802124,
      "learning_rate": 5.375618199139343e-09,
      "loss": 0.4706,
      "step": 15558
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1024744510650635,
      "learning_rate": 5.328160036234264e-09,
      "loss": 0.5237,
      "step": 15559
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1816823482513428,
      "learning_rate": 5.280912181584441e-09,
      "loss": 0.5398,
      "step": 15560
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.034546971321106,
      "learning_rate": 5.233874637180503e-09,
      "loss": 0.5295,
      "step": 15561
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.04789137840271,
      "learning_rate": 5.1870474050025325e-09,
      "loss": 0.5054,
      "step": 15562
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0426819324493408,
      "learning_rate": 5.1404304870222856e-09,
      "loss": 0.4796,
      "step": 15563
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0558671951293945,
      "learning_rate": 5.094023885203192e-09,
      "loss": 0.4662,
      "step": 15564
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1009888648986816,
      "learning_rate": 5.0478276014981345e-09,
      "loss": 0.5328,
      "step": 15565
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2031971216201782,
      "learning_rate": 5.001841637852778e-09,
      "loss": 0.4895,
      "step": 15566
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0120888948440552,
      "learning_rate": 4.956065996203907e-09,
      "loss": 0.4846,
      "step": 15567
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0949715375900269,
      "learning_rate": 4.910500678478314e-09,
      "loss": 0.4925,
      "step": 15568
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.112021565437317,
      "learning_rate": 4.865145686595019e-09,
      "loss": 0.5211,
      "step": 15569
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9821246862411499,
      "learning_rate": 4.820001022463605e-09,
      "loss": 0.4812,
      "step": 15570
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1005215644836426,
      "learning_rate": 4.77506668798533e-09,
      "loss": 0.5139,
      "step": 15571
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.044965386390686,
      "learning_rate": 4.730342685051459e-09,
      "loss": 0.5013,
      "step": 15572
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0928858518600464,
      "learning_rate": 4.685829015545485e-09,
      "loss": 0.501,
      "step": 15573
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0987399816513062,
      "learning_rate": 4.641525681342019e-09,
      "loss": 0.5113,
      "step": 15574
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1467041969299316,
      "learning_rate": 4.597432684306236e-09,
      "loss": 0.5519,
      "step": 15575
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1859843730926514,
      "learning_rate": 4.553550026294984e-09,
      "loss": 0.5583,
      "step": 15576
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0004034042358398,
      "learning_rate": 4.5098777091556745e-09,
      "loss": 0.4807,
      "step": 15577
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0263206958770752,
      "learning_rate": 4.4664157347273916e-09,
      "loss": 0.5219,
      "step": 15578
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0322223901748657,
      "learning_rate": 4.423164104840339e-09,
      "loss": 0.5045,
      "step": 15579
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0323245525360107,
      "learning_rate": 4.38012282131528e-09,
      "loss": 0.4675,
      "step": 15580
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.053807258605957,
      "learning_rate": 4.3372918859652115e-09,
      "loss": 0.5007,
      "step": 15581
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0838310718536377,
      "learning_rate": 4.294671300592579e-09,
      "loss": 0.4959,
      "step": 15582
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0312323570251465,
      "learning_rate": 4.252261066993169e-09,
      "loss": 0.4954,
      "step": 15583
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.093133807182312,
      "learning_rate": 4.210061186951664e-09,
      "loss": 0.542,
      "step": 15584
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0975558757781982,
      "learning_rate": 4.168071662245532e-09,
      "loss": 0.5134,
      "step": 15585
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.079088807106018,
      "learning_rate": 4.1262924946422476e-09,
      "loss": 0.4907,
      "step": 15586
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.068571925163269,
      "learning_rate": 4.084723685901515e-09,
      "loss": 0.5307,
      "step": 15587
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.081530213356018,
      "learning_rate": 4.043365237774155e-09,
      "loss": 0.5256,
      "step": 15588
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0684465169906616,
      "learning_rate": 4.002217152000443e-09,
      "loss": 0.4959,
      "step": 15589
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9404188990592957,
      "learning_rate": 3.961279430313991e-09,
      "loss": 0.4006,
      "step": 15590
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0813419818878174,
      "learning_rate": 3.920552074437867e-09,
      "loss": 0.4954,
      "step": 15591
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1187856197357178,
      "learning_rate": 3.880035086086808e-09,
      "loss": 0.4615,
      "step": 15592
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0242195129394531,
      "learning_rate": 3.839728466967785e-09,
      "loss": 0.4851,
      "step": 15593
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1560113430023193,
      "learning_rate": 3.799632218777216e-09,
      "loss": 0.5075,
      "step": 15594
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1023149490356445,
      "learning_rate": 3.759746343203751e-09,
      "loss": 0.4827,
      "step": 15595
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1077873706817627,
      "learning_rate": 3.720070841926604e-09,
      "loss": 0.4832,
      "step": 15596
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0226209163665771,
      "learning_rate": 3.6806057166166585e-09,
      "loss": 0.4924,
      "step": 15597
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0712997913360596,
      "learning_rate": 3.6413509689353644e-09,
      "loss": 0.5191,
      "step": 15598
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0410377979278564,
      "learning_rate": 3.602306600535843e-09,
      "loss": 0.5057,
      "step": 15599
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.071582317352295,
      "learning_rate": 3.5634726130617802e-09,
      "loss": 0.5255,
      "step": 15600
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.977677583694458,
      "learning_rate": 3.5248490081485343e-09,
      "loss": 0.4713,
      "step": 15601
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9692546129226685,
      "learning_rate": 3.486435787422582e-09,
      "loss": 0.475,
      "step": 15602
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0329508781433105,
      "learning_rate": 3.4482329525009627e-09,
      "loss": 0.4914,
      "step": 15603
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0116748809814453,
      "learning_rate": 3.4102405049929455e-09,
      "loss": 0.5348,
      "step": 15604
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1092138290405273,
      "learning_rate": 3.372458446497251e-09,
      "loss": 0.5007,
      "step": 15605
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1361290216445923,
      "learning_rate": 3.3348867786059393e-09,
      "loss": 0.5157,
      "step": 15606
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0798801183700562,
      "learning_rate": 3.2975255028999675e-09,
      "loss": 0.4856,
      "step": 15607
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0976042747497559,
      "learning_rate": 3.2603746209530774e-09,
      "loss": 0.4749,
      "step": 15608
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.063442587852478,
      "learning_rate": 3.223434134329573e-09,
      "loss": 0.4877,
      "step": 15609
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1981234550476074,
      "learning_rate": 3.1867040445848764e-09,
      "loss": 0.5203,
      "step": 15610
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0773344039916992,
      "learning_rate": 3.1501843532649734e-09,
      "loss": 0.444,
      "step": 15611
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.059180498123169,
      "learning_rate": 3.113875061908078e-09,
      "loss": 0.4834,
      "step": 15612
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0928962230682373,
      "learning_rate": 3.077776172043523e-09,
      "loss": 0.4997,
      "step": 15613
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0543581247329712,
      "learning_rate": 3.0418876851900924e-09,
      "loss": 0.4387,
      "step": 15614
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.069734811782837,
      "learning_rate": 3.0062096028599108e-09,
      "loss": 0.4502,
      "step": 15615
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9861531257629395,
      "learning_rate": 2.9707419265551097e-09,
      "loss": 0.4883,
      "step": 15616
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0755091905593872,
      "learning_rate": 2.9354846577689387e-09,
      "loss": 0.531,
      "step": 15617
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0954896211624146,
      "learning_rate": 2.900437797986322e-09,
      "loss": 0.4631,
      "step": 15618
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0470771789550781,
      "learning_rate": 2.8656013486821897e-09,
      "loss": 0.5038,
      "step": 15619
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9958431124687195,
      "learning_rate": 2.8309753113237025e-09,
      "loss": 0.4625,
      "step": 15620
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9982820153236389,
      "learning_rate": 2.796559687369138e-09,
      "loss": 0.4949,
      "step": 15621
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9949272871017456,
      "learning_rate": 2.7623544782673372e-09,
      "loss": 0.4438,
      "step": 15622
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.064408779144287,
      "learning_rate": 2.7283596854588148e-09,
      "loss": 0.5071,
      "step": 15623
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0921839475631714,
      "learning_rate": 2.6945753103746475e-09,
      "loss": 0.4698,
      "step": 15624
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0886198282241821,
      "learning_rate": 2.661001354437587e-09,
      "loss": 0.4907,
      "step": 15625
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0785962343215942,
      "learning_rate": 2.6276378190615016e-09,
      "loss": 0.4858,
      "step": 15626
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1296824216842651,
      "learning_rate": 2.5944847056508237e-09,
      "loss": 0.463,
      "step": 15627
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1042518615722656,
      "learning_rate": 2.561542015601659e-09,
      "loss": 0.5477,
      "step": 15628
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1353806257247925,
      "learning_rate": 2.528809750301231e-09,
      "loss": 0.4806,
      "step": 15629
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1011172533035278,
      "learning_rate": 2.4962879111278813e-09,
      "loss": 0.4638,
      "step": 15630
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0281448364257812,
      "learning_rate": 2.4639764994505156e-09,
      "loss": 0.4835,
      "step": 15631
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0197134017944336,
      "learning_rate": 2.4318755166302668e-09,
      "loss": 0.4865,
      "step": 15632
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.008103609085083,
      "learning_rate": 2.399984964018276e-09,
      "loss": 0.4972,
      "step": 15633
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9819304943084717,
      "learning_rate": 2.3683048429573587e-09,
      "loss": 0.4532,
      "step": 15634
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0961347818374634,
      "learning_rate": 2.3368351547820023e-09,
      "loss": 0.5162,
      "step": 15635
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0458500385284424,
      "learning_rate": 2.3055759008167033e-09,
      "loss": 0.4518,
      "step": 15636
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0598288774490356,
      "learning_rate": 2.2745270823776312e-09,
      "loss": 0.4867,
      "step": 15637
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1259396076202393,
      "learning_rate": 2.243688700772628e-09,
      "loss": 0.538,
      "step": 15638
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0349446535110474,
      "learning_rate": 2.2130607573001006e-09,
      "loss": 0.5405,
      "step": 15639
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.058007001876831,
      "learning_rate": 2.1826432532495724e-09,
      "loss": 0.4848,
      "step": 15640
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0677138566970825,
      "learning_rate": 2.1524361899016853e-09,
      "loss": 0.5082,
      "step": 15641
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9990156888961792,
      "learning_rate": 2.1224395685282008e-09,
      "loss": 0.5306,
      "step": 15642
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9851405024528503,
      "learning_rate": 2.0926533903925516e-09,
      "loss": 0.4652,
      "step": 15643
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1061551570892334,
      "learning_rate": 2.0630776567492904e-09,
      "loss": 0.5053,
      "step": 15644
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0540603399276733,
      "learning_rate": 2.033712368842977e-09,
      "loss": 0.5004,
      "step": 15645
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0345008373260498,
      "learning_rate": 2.004557527909845e-09,
      "loss": 0.4679,
      "step": 15646
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0666303634643555,
      "learning_rate": 1.975613135178911e-09,
      "loss": 0.4971,
      "step": 15647
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1802502870559692,
      "learning_rate": 1.9468791918675345e-09,
      "loss": 0.4951,
      "step": 15648
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9935306906700134,
      "learning_rate": 1.918355699186414e-09,
      "loss": 0.5036,
      "step": 15649
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.019480586051941,
      "learning_rate": 1.8900426583357003e-09,
      "loss": 0.4298,
      "step": 15650
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9573292136192322,
      "learning_rate": 1.861940070508883e-09,
      "loss": 0.4635,
      "step": 15651
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.069796085357666,
      "learning_rate": 1.8340479368883502e-09,
      "loss": 0.5041,
      "step": 15652
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0155587196350098,
      "learning_rate": 1.8063662586481622e-09,
      "loss": 0.4627,
      "step": 15653
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9943329095840454,
      "learning_rate": 1.7788950369551638e-09,
      "loss": 0.4771,
      "step": 15654
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0415226221084595,
      "learning_rate": 1.751634272964542e-09,
      "loss": 0.471,
      "step": 15655
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1297926902770996,
      "learning_rate": 1.7245839678259323e-09,
      "loss": 0.4791,
      "step": 15656
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1199589967727661,
      "learning_rate": 1.6977441226767589e-09,
      "loss": 0.5399,
      "step": 15657
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9969205856323242,
      "learning_rate": 1.6711147386477833e-09,
      "loss": 0.4313,
      "step": 15658
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0321614742279053,
      "learning_rate": 1.644695816860331e-09,
      "loss": 0.4638,
      "step": 15659
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.067716360092163,
      "learning_rate": 1.6184873584268457e-09,
      "loss": 0.5376,
      "step": 15660
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0950990915298462,
      "learning_rate": 1.5924893644503336e-09,
      "loss": 0.5583,
      "step": 15661
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1522343158721924,
      "learning_rate": 1.566701836026585e-09,
      "loss": 0.5207,
      "step": 15662
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0168970823287964,
      "learning_rate": 1.541124774240288e-09,
      "loss": 0.5075,
      "step": 15663
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0221366882324219,
      "learning_rate": 1.5157581801689137e-09,
      "loss": 0.4748,
      "step": 15664
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0731167793273926,
      "learning_rate": 1.4906020548804968e-09,
      "loss": 0.503,
      "step": 15665
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0286649465560913,
      "learning_rate": 1.4656563994341898e-09,
      "loss": 0.47,
      "step": 15666
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1274816989898682,
      "learning_rate": 1.4409212148802643e-09,
      "loss": 0.5381,
      "step": 15667
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0476179122924805,
      "learning_rate": 1.416396502260664e-09,
      "loss": 0.4834,
      "step": 15668
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0860023498535156,
      "learning_rate": 1.3920822626078967e-09,
      "loss": 0.4954,
      "step": 15669
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0240345001220703,
      "learning_rate": 1.3679784969461429e-09,
      "loss": 0.528,
      "step": 15670
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.041835069656372,
      "learning_rate": 1.3440852062890364e-09,
      "loss": 0.5012,
      "step": 15671
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9991360902786255,
      "learning_rate": 1.3204023916435494e-09,
      "loss": 0.4802,
      "step": 15672
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0960899591445923,
      "learning_rate": 1.2969300540072171e-09,
      "loss": 0.4822,
      "step": 15673
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0605608224868774,
      "learning_rate": 1.2736681943675833e-09,
      "loss": 0.4904,
      "step": 15674
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0661795139312744,
      "learning_rate": 1.2506168137049747e-09,
      "loss": 0.5213,
      "step": 15675
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9501168131828308,
      "learning_rate": 1.2277759129886158e-09,
      "loss": 0.4456,
      "step": 15676
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0185068845748901,
      "learning_rate": 1.2051454931816254e-09,
      "loss": 0.4724,
      "step": 15677
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1506839990615845,
      "learning_rate": 1.1827255552365745e-09,
      "loss": 0.4651,
      "step": 15678
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0617283582687378,
      "learning_rate": 1.1605161000971532e-09,
      "loss": 0.5208,
      "step": 15679
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.061086893081665,
      "learning_rate": 1.1385171286992791e-09,
      "loss": 0.5032,
      "step": 15680
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1165785789489746,
      "learning_rate": 1.116728641967768e-09,
      "loss": 0.5692,
      "step": 15681
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0701615810394287,
      "learning_rate": 1.0951506408213298e-09,
      "loss": 0.507,
      "step": 15682
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0252995491027832,
      "learning_rate": 1.0737831261686815e-09,
      "loss": 0.5245,
      "step": 15683
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0762993097305298,
      "learning_rate": 1.052626098907994e-09,
      "loss": 0.4707,
      "step": 15684
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0970128774642944,
      "learning_rate": 1.0316795599318862e-09,
      "loss": 0.4847,
      "step": 15685
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0381357669830322,
      "learning_rate": 1.0109435101218757e-09,
      "loss": 0.4986,
      "step": 15686
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1097840070724487,
      "learning_rate": 9.90417950350042e-10,
      "loss": 0.4853,
      "step": 15687
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0169131755828857,
      "learning_rate": 9.701028814818047e-10,
      "loss": 0.5007,
      "step": 15688
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1126348972320557,
      "learning_rate": 9.499983043720351e-10,
      "loss": 0.5161,
      "step": 15689
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0900895595550537,
      "learning_rate": 9.301042198678334e-10,
      "loss": 0.5049,
      "step": 15690
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1760368347167969,
      "learning_rate": 9.104206288057527e-10,
      "loss": 0.4983,
      "step": 15691
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0283381938934326,
      "learning_rate": 8.909475320156846e-10,
      "loss": 0.4404,
      "step": 15692
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0624741315841675,
      "learning_rate": 8.71684930317529e-10,
      "loss": 0.4906,
      "step": 15693
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0771564245224,
      "learning_rate": 8.526328245217485e-10,
      "loss": 0.4918,
      "step": 15694
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0345393419265747,
      "learning_rate": 8.337912154304795e-10,
      "loss": 0.5149,
      "step": 15695
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0324221849441528,
      "learning_rate": 8.151601038375312e-10,
      "loss": 0.4918,
      "step": 15696
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0914480686187744,
      "learning_rate": 7.967394905278314e-10,
      "loss": 0.4865,
      "step": 15697
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9753479957580566,
      "learning_rate": 7.785293762757607e-10,
      "loss": 0.4864,
      "step": 15698
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0722107887268066,
      "learning_rate": 7.605297618495932e-10,
      "loss": 0.5488,
      "step": 15699
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.088541030883789,
      "learning_rate": 7.427406480059463e-10,
      "loss": 0.5284,
      "step": 15700
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9980146288871765,
      "learning_rate": 7.251620354942201e-10,
      "loss": 0.4872,
      "step": 15701
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0236068964004517,
      "learning_rate": 7.077939250549337e-10,
      "loss": 0.4758,
      "step": 15702
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0211738348007202,
      "learning_rate": 6.906363174191688e-10,
      "loss": 0.4903,
      "step": 15703
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9885705709457397,
      "learning_rate": 6.736892133091255e-10,
      "loss": 0.432,
      "step": 15704
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9520770311355591,
      "learning_rate": 6.569526134392324e-10,
      "loss": 0.4016,
      "step": 15705
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0254629850387573,
      "learning_rate": 6.404265185128155e-10,
      "loss": 0.5599,
      "step": 15706
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.090428352355957,
      "learning_rate": 6.241109292270953e-10,
      "loss": 0.4845,
      "step": 15707
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1226465702056885,
      "learning_rate": 6.080058462687444e-10,
      "loss": 0.5108,
      "step": 15708
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0290261507034302,
      "learning_rate": 5.92111270314999e-10,
      "loss": 0.4717,
      "step": 15709
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0893996953964233,
      "learning_rate": 5.764272020358785e-10,
      "loss": 0.5063,
      "step": 15710
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1903843879699707,
      "learning_rate": 5.609536420919659e-10,
      "loss": 0.5099,
      "step": 15711
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0996195077896118,
      "learning_rate": 5.456905911344068e-10,
      "loss": 0.4883,
      "step": 15712
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1661995649337769,
      "learning_rate": 5.306380498060204e-10,
      "loss": 0.5083,
      "step": 15713
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0439397096633911,
      "learning_rate": 5.15796018740744e-10,
      "loss": 0.5051,
      "step": 15714
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0537337064743042,
      "learning_rate": 5.011644985630781e-10,
      "loss": 0.4833,
      "step": 15715
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0489590167999268,
      "learning_rate": 4.867434898891965e-10,
      "loss": 0.5071,
      "step": 15716
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0084718465805054,
      "learning_rate": 4.725329933269463e-10,
      "loss": 0.5157,
      "step": 15717
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.024583101272583,
      "learning_rate": 4.5853300947418247e-10,
      "loss": 0.4831,
      "step": 15718
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0930402278900146,
      "learning_rate": 4.4474353892043356e-10,
      "loss": 0.5319,
      "step": 15719
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9511051774024963,
      "learning_rate": 4.311645822463462e-10,
      "loss": 0.4464,
      "step": 15720
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1021090745925903,
      "learning_rate": 4.177961400236852e-10,
      "loss": 0.5023,
      "step": 15721
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.030034065246582,
      "learning_rate": 4.046382128147786e-10,
      "loss": 0.4818,
      "step": 15722
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0933709144592285,
      "learning_rate": 3.916908011747378e-10,
      "loss": 0.5577,
      "step": 15723
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0729312896728516,
      "learning_rate": 3.7895390564868237e-10,
      "loss": 0.513,
      "step": 15724
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.008466362953186,
      "learning_rate": 3.664275267717399e-10,
      "loss": 0.4971,
      "step": 15725
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0403852462768555,
      "learning_rate": 3.541116650723764e-10,
      "loss": 0.4527,
      "step": 15726
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9811885952949524,
      "learning_rate": 3.4200632106906605e-10,
      "loss": 0.4783,
      "step": 15727
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9582056999206543,
      "learning_rate": 3.301114952708462e-10,
      "loss": 0.4753,
      "step": 15728
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1376382112503052,
      "learning_rate": 3.1842718817953755e-10,
      "loss": 0.537,
      "step": 15729
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1202480792999268,
      "learning_rate": 3.0695340028641383e-10,
      "loss": 0.568,
      "step": 15730
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0908098220825195,
      "learning_rate": 2.956901320744221e-10,
      "loss": 0.5215,
      "step": 15731
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0361547470092773,
      "learning_rate": 2.8463738401873776e-10,
      "loss": 0.5091,
      "step": 15732
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0398439168930054,
      "learning_rate": 2.7379515658398915e-10,
      "loss": 0.4834,
      "step": 15733
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0584646463394165,
      "learning_rate": 2.6316345022703307e-10,
      "loss": 0.484,
      "step": 15734
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.104377269744873,
      "learning_rate": 2.5274226539584445e-10,
      "loss": 0.4811,
      "step": 15735
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.004241704940796,
      "learning_rate": 2.4253160252840636e-10,
      "loss": 0.4199,
      "step": 15736
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0890564918518066,
      "learning_rate": 2.3253146205493014e-10,
      "loss": 0.4789,
      "step": 15737
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0498672723770142,
      "learning_rate": 2.227418443967455e-10,
      "loss": 0.5008,
      "step": 15738
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0728520154953003,
      "learning_rate": 2.1316274996630026e-10,
      "loss": 0.5534,
      "step": 15739
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9895668029785156,
      "learning_rate": 2.0379417916605027e-10,
      "loss": 0.4431,
      "step": 15740
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1057844161987305,
      "learning_rate": 1.946361323912349e-10,
      "loss": 0.4936,
      "step": 15741
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0985236167907715,
      "learning_rate": 1.856886100276567e-10,
      "loss": 0.5129,
      "step": 15742
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0444600582122803,
      "learning_rate": 1.7695161245112613e-10,
      "loss": 0.4916,
      "step": 15743
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1546952724456787,
      "learning_rate": 1.6842514003023724e-10,
      "loss": 0.5168,
      "step": 15744
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0188254117965698,
      "learning_rate": 1.6010919312359208e-10,
      "loss": 0.467,
      "step": 15745
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0582890510559082,
      "learning_rate": 1.520037720820211e-10,
      "loss": 0.4842,
      "step": 15746
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0392037630081177,
      "learning_rate": 1.4410887724580768e-10,
      "loss": 0.4333,
      "step": 15747
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0552963018417358,
      "learning_rate": 1.3642450894801872e-10,
      "loss": 0.4416,
      "step": 15748
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0796518325805664,
      "learning_rate": 1.2895066751283935e-10,
      "loss": 0.4768,
      "step": 15749
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0383802652359009,
      "learning_rate": 1.2168735325335246e-10,
      "loss": 0.511,
      "step": 15750
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0444761514663696,
      "learning_rate": 1.1463456647708982e-10,
      "loss": 0.5094,
      "step": 15751
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.060904622077942,
      "learning_rate": 1.0779230747992587e-10,
      "loss": 0.5082,
      "step": 15752
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1346057653427124,
      "learning_rate": 1.0116057654996348e-10,
      "loss": 0.5678,
      "step": 15753
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.065916657447815,
      "learning_rate": 9.473937396697885e-11,
      "loss": 0.4356,
      "step": 15754
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0521526336669922,
      "learning_rate": 8.85287000013113e-11,
      "loss": 0.496,
      "step": 15755
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0416265726089478,
      "learning_rate": 8.252855491386325e-11,
      "loss": 0.4501,
      "step": 15756
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0565155744552612,
      "learning_rate": 7.673893895776551e-11,
      "loss": 0.4994,
      "step": 15757
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0739134550094604,
      "learning_rate": 7.115985237726719e-11,
      "loss": 0.5306,
      "step": 15758
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0463873147964478,
      "learning_rate": 6.57912954060702e-11,
      "loss": 0.5017,
      "step": 15759
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0840357542037964,
      "learning_rate": 6.063326827121518e-11,
      "loss": 0.4765,
      "step": 15760
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.083588719367981,
      "learning_rate": 5.5685771189750714e-11,
      "loss": 0.4727,
      "step": 15761
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0757077932357788,
      "learning_rate": 5.0948804369843616e-11,
      "loss": 0.4973,
      "step": 15762
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.004027247428894,
      "learning_rate": 4.642236801022382e-11,
      "loss": 0.5167,
      "step": 15763
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1284191608428955,
      "learning_rate": 4.210646230295989e-11,
      "loss": 0.4831,
      "step": 15764
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0468326807022095,
      "learning_rate": 3.800108742846309e-11,
      "loss": 0.508,
      "step": 15765
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1716840267181396,
      "learning_rate": 3.410624356048331e-11,
      "loss": 0.5461,
      "step": 15766
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0520511865615845,
      "learning_rate": 3.0421930862778446e-11,
      "loss": 0.5136,
      "step": 15767
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0537054538726807,
      "learning_rate": 2.6948149490224618e-11,
      "loss": 0.5008,
      "step": 15768
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0120649337768555,
      "learning_rate": 2.3684899589371256e-11,
      "loss": 0.4472,
      "step": 15769
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0789854526519775,
      "learning_rate": 2.063218129733091e-11,
      "loss": 0.5109,
      "step": 15770
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1495561599731445,
      "learning_rate": 1.7789994742889448e-11,
      "loss": 0.5457,
      "step": 15771
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0533004999160767,
      "learning_rate": 1.5158340045395847e-11,
      "loss": 0.5097,
      "step": 15772
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.058885931968689,
      "learning_rate": 1.2737217315872407e-11,
      "loss": 0.4954,
      "step": 15773
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.064213514328003,
      "learning_rate": 1.0526626656459648e-11,
      "loss": 0.4911,
      "step": 15774
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0072990655899048,
      "learning_rate": 8.526568160416304e-12,
      "loss": 0.521,
      "step": 15775
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1676479578018188,
      "learning_rate": 6.737041911564213e-12,
      "loss": 0.5327,
      "step": 15776
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1192338466644287,
      "learning_rate": 5.158047984843428e-12,
      "loss": 0.4802,
      "step": 15777
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1080821752548218,
      "learning_rate": 3.789586447422444e-12,
      "loss": 0.472,
      "step": 15778
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0763338804244995,
      "learning_rate": 2.631657356477746e-12,
      "loss": 0.5215,
      "step": 15779
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0144696235656738,
      "learning_rate": 1.6842607614142582e-12,
      "loss": 0.4771,
      "step": 15780
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0714669227600098,
      "learning_rate": 9.473967016448982e-13,
      "loss": 0.5592,
      "step": 15781
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1242960691452026,
      "learning_rate": 4.210652082559108e-13,
      "loss": 0.4959,
      "step": 15782
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0352383852005005,
      "learning_rate": 1.0526630289664496e-13,
      "loss": 0.4915,
      "step": 15783
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.006693720817566,
      "learning_rate": 0.0,
      "loss": 0.4885,
      "step": 15784
    },
    {
      "epoch": 1.0,
      "step": 15784,
      "total_flos": 3.799196684334504e+19,
      "train_loss": 0.5349509626951727,
      "train_runtime": 163223.245,
      "train_samples_per_second": 24.756,
      "train_steps_per_second": 0.097
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 15784,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 8000,
  "total_flos": 3.799196684334504e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}