{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999245909056632,
  "eval_steps": 500,
  "global_step": 6630,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00015081818867355404,
      "grad_norm": 0.6922811636045936,
      "learning_rate": 4.999999747404737e-07,
      "loss": 1.6389,
      "step": 1
    },
    {
      "epoch": 0.0003016363773471081,
      "grad_norm": 0.7140881455188725,
      "learning_rate": 4.999998989619002e-07,
      "loss": 1.7317,
      "step": 2
    },
    {
      "epoch": 0.0004524545660206621,
      "grad_norm": 0.7901119831427035,
      "learning_rate": 4.999997726642967e-07,
      "loss": 1.6932,
      "step": 3
    },
    {
      "epoch": 0.0006032727546942162,
      "grad_norm": 0.779821728799067,
      "learning_rate": 4.999995958476916e-07,
      "loss": 1.6998,
      "step": 4
    },
    {
      "epoch": 0.0007540909433677702,
      "grad_norm": 0.7460901373887201,
      "learning_rate": 4.999993685121245e-07,
      "loss": 1.6461,
      "step": 5
    },
    {
      "epoch": 0.0009049091320413242,
      "grad_norm": 0.7488137173948498,
      "learning_rate": 4.999990906576464e-07,
      "loss": 1.7045,
      "step": 6
    },
    {
      "epoch": 0.0010557273207148783,
      "grad_norm": 0.8105539489064422,
      "learning_rate": 4.999987622843198e-07,
      "loss": 1.6732,
      "step": 7
    },
    {
      "epoch": 0.0012065455093884323,
      "grad_norm": 0.6994546178393476,
      "learning_rate": 4.999983833922184e-07,
      "loss": 1.6795,
      "step": 8
    },
    {
      "epoch": 0.0013573636980619864,
      "grad_norm": 0.7937671031757709,
      "learning_rate": 4.999979539814273e-07,
      "loss": 1.7008,
      "step": 9
    },
    {
      "epoch": 0.0015081818867355404,
      "grad_norm": 0.6602329299835072,
      "learning_rate": 4.999974740520428e-07,
      "loss": 1.6682,
      "step": 10
    },
    {
      "epoch": 0.0016590000754090944,
      "grad_norm": 0.6821099703015343,
      "learning_rate": 4.999969436041726e-07,
      "loss": 1.678,
      "step": 11
    },
    {
      "epoch": 0.0018098182640826485,
      "grad_norm": 0.7189347675969909,
      "learning_rate": 4.999963626379361e-07,
      "loss": 1.6921,
      "step": 12
    },
    {
      "epoch": 0.0019606364527562023,
      "grad_norm": 0.6454737446123784,
      "learning_rate": 4.999957311534635e-07,
      "loss": 1.6031,
      "step": 13
    },
    {
      "epoch": 0.0021114546414297566,
      "grad_norm": 0.6932215740279462,
      "learning_rate": 4.999950491508967e-07,
      "loss": 1.6368,
      "step": 14
    },
    {
      "epoch": 0.0022622728301033104,
      "grad_norm": 0.6697749479364712,
      "learning_rate": 4.999943166303888e-07,
      "loss": 1.7087,
      "step": 15
    },
    {
      "epoch": 0.0024130910187768646,
      "grad_norm": 0.6414871607410737,
      "learning_rate": 4.999935335921042e-07,
      "loss": 1.5736,
      "step": 16
    },
    {
      "epoch": 0.0025639092074504185,
      "grad_norm": 0.6524390383806831,
      "learning_rate": 4.999927000362188e-07,
      "loss": 1.6607,
      "step": 17
    },
    {
      "epoch": 0.0027147273961239727,
      "grad_norm": 0.5498201184163394,
      "learning_rate": 4.999918159629197e-07,
      "loss": 1.6549,
      "step": 18
    },
    {
      "epoch": 0.0028655455847975265,
      "grad_norm": 0.5906190966505179,
      "learning_rate": 4.999908813724054e-07,
      "loss": 1.6223,
      "step": 19
    },
    {
      "epoch": 0.003016363773471081,
      "grad_norm": 0.5409603294442903,
      "learning_rate": 4.999898962648858e-07,
      "loss": 1.646,
      "step": 20
    },
    {
      "epoch": 0.0031671819621446346,
      "grad_norm": 0.47639936391820453,
      "learning_rate": 4.999888606405821e-07,
      "loss": 1.6288,
      "step": 21
    },
    {
      "epoch": 0.003318000150818189,
      "grad_norm": 0.49001145145256797,
      "learning_rate": 4.999877744997268e-07,
      "loss": 1.6078,
      "step": 22
    },
    {
      "epoch": 0.0034688183394917427,
      "grad_norm": 0.4480835545958283,
      "learning_rate": 4.999866378425637e-07,
      "loss": 1.6913,
      "step": 23
    },
    {
      "epoch": 0.003619636528165297,
      "grad_norm": 0.47975660258284775,
      "learning_rate": 4.999854506693481e-07,
      "loss": 1.6562,
      "step": 24
    },
    {
      "epoch": 0.0037704547168388508,
      "grad_norm": 0.45739459732441357,
      "learning_rate": 4.999842129803465e-07,
      "loss": 1.5977,
      "step": 25
    },
    {
      "epoch": 0.003921272905512405,
      "grad_norm": 0.47345874678684624,
      "learning_rate": 4.999829247758368e-07,
      "loss": 1.6145,
      "step": 26
    },
    {
      "epoch": 0.004072091094185958,
      "grad_norm": 0.5277188876747779,
      "learning_rate": 4.999815860561083e-07,
      "loss": 1.6744,
      "step": 27
    },
    {
      "epoch": 0.004222909282859513,
      "grad_norm": 0.49781748452982383,
      "learning_rate": 4.999801968214616e-07,
      "loss": 1.6795,
      "step": 28
    },
    {
      "epoch": 0.004373727471533067,
      "grad_norm": 0.4058450594970341,
      "learning_rate": 4.999787570722085e-07,
      "loss": 1.5995,
      "step": 29
    },
    {
      "epoch": 0.004524545660206621,
      "grad_norm": 0.4688159734665827,
      "learning_rate": 4.999772668086723e-07,
      "loss": 1.6533,
      "step": 30
    },
    {
      "epoch": 0.004675363848880175,
      "grad_norm": 0.5382144698866896,
      "learning_rate": 4.999757260311877e-07,
      "loss": 1.6687,
      "step": 31
    },
    {
      "epoch": 0.004826182037553729,
      "grad_norm": 0.43364920622549574,
      "learning_rate": 4.999741347401006e-07,
      "loss": 1.6237,
      "step": 32
    },
    {
      "epoch": 0.004977000226227283,
      "grad_norm": 0.4417491883275204,
      "learning_rate": 4.999724929357683e-07,
      "loss": 1.6064,
      "step": 33
    },
    {
      "epoch": 0.005127818414900837,
      "grad_norm": 0.3798147688605585,
      "learning_rate": 4.999708006185593e-07,
      "loss": 1.5982,
      "step": 34
    },
    {
      "epoch": 0.005278636603574391,
      "grad_norm": 0.3722151540923329,
      "learning_rate": 4.999690577888537e-07,
      "loss": 1.5851,
      "step": 35
    },
    {
      "epoch": 0.005429454792247945,
      "grad_norm": 0.3670996345126827,
      "learning_rate": 4.999672644470428e-07,
      "loss": 1.6649,
      "step": 36
    },
    {
      "epoch": 0.005580272980921499,
      "grad_norm": 0.32863895232492335,
      "learning_rate": 4.999654205935294e-07,
      "loss": 1.6039,
      "step": 37
    },
    {
      "epoch": 0.005731091169595053,
      "grad_norm": 0.33479731199962554,
      "learning_rate": 4.999635262287272e-07,
      "loss": 1.6022,
      "step": 38
    },
    {
      "epoch": 0.005881909358268607,
      "grad_norm": 0.3340694893124953,
      "learning_rate": 4.999615813530618e-07,
      "loss": 1.7122,
      "step": 39
    },
    {
      "epoch": 0.006032727546942162,
      "grad_norm": 0.34129372778475764,
      "learning_rate": 4.999595859669698e-07,
      "loss": 1.7592,
      "step": 40
    },
    {
      "epoch": 0.006183545735615715,
      "grad_norm": 0.3737447888089227,
      "learning_rate": 4.999575400708991e-07,
      "loss": 1.675,
      "step": 41
    },
    {
      "epoch": 0.006334363924289269,
      "grad_norm": 0.31806492846143203,
      "learning_rate": 4.999554436653093e-07,
      "loss": 1.6369,
      "step": 42
    },
    {
      "epoch": 0.006485182112962823,
      "grad_norm": 4.465999509985267,
      "learning_rate": 4.999532967506708e-07,
      "loss": 1.768,
      "step": 43
    },
    {
      "epoch": 0.006636000301636378,
      "grad_norm": 0.38534249346532656,
      "learning_rate": 4.999510993274659e-07,
      "loss": 1.6387,
      "step": 44
    },
    {
      "epoch": 0.0067868184903099316,
      "grad_norm": 0.31913751532411,
      "learning_rate": 4.99948851396188e-07,
      "loss": 1.6166,
      "step": 45
    },
    {
      "epoch": 0.006937636678983485,
      "grad_norm": 2.426973413694192,
      "learning_rate": 4.999465529573416e-07,
      "loss": 1.6164,
      "step": 46
    },
    {
      "epoch": 0.007088454867657039,
      "grad_norm": 0.3265648082672616,
      "learning_rate": 4.999442040114429e-07,
      "loss": 1.6604,
      "step": 47
    },
    {
      "epoch": 0.007239273056330594,
      "grad_norm": 0.31660772435225665,
      "learning_rate": 4.999418045590194e-07,
      "loss": 1.5524,
      "step": 48
    },
    {
      "epoch": 0.007390091245004148,
      "grad_norm": 0.31800527693674274,
      "learning_rate": 4.999393546006097e-07,
      "loss": 1.6414,
      "step": 49
    },
    {
      "epoch": 0.0075409094336777015,
      "grad_norm": 0.3152640122154442,
      "learning_rate": 4.999368541367639e-07,
      "loss": 1.6063,
      "step": 50
    },
    {
      "epoch": 0.007691727622351255,
      "grad_norm": 0.32043620783105936,
      "learning_rate": 4.999343031680434e-07,
      "loss": 1.6677,
      "step": 51
    },
    {
      "epoch": 0.00784254581102481,
      "grad_norm": 0.3018244592626452,
      "learning_rate": 4.999317016950211e-07,
      "loss": 1.6265,
      "step": 52
    },
    {
      "epoch": 0.007993363999698364,
      "grad_norm": 0.3424930702846109,
      "learning_rate": 4.99929049718281e-07,
      "loss": 1.6404,
      "step": 53
    },
    {
      "epoch": 0.008144182188371917,
      "grad_norm": 0.30974610592242463,
      "learning_rate": 4.999263472384186e-07,
      "loss": 1.5421,
      "step": 54
    },
    {
      "epoch": 0.008295000377045472,
      "grad_norm": 0.30658278771053415,
      "learning_rate": 4.999235942560407e-07,
      "loss": 1.6901,
      "step": 55
    },
    {
      "epoch": 0.008445818565719026,
      "grad_norm": 0.3174958401672945,
      "learning_rate": 4.999207907717654e-07,
      "loss": 1.5751,
      "step": 56
    },
    {
      "epoch": 0.00859663675439258,
      "grad_norm": 0.38057071951040056,
      "learning_rate": 4.999179367862221e-07,
      "loss": 1.5924,
      "step": 57
    },
    {
      "epoch": 0.008747454943066134,
      "grad_norm": 0.3042205867066707,
      "learning_rate": 4.999150323000517e-07,
      "loss": 1.6212,
      "step": 58
    },
    {
      "epoch": 0.008898273131739689,
      "grad_norm": 0.29037474131564006,
      "learning_rate": 4.999120773139062e-07,
      "loss": 1.6054,
      "step": 59
    },
    {
      "epoch": 0.009049091320413242,
      "grad_norm": 0.33743493712479644,
      "learning_rate": 4.999090718284493e-07,
      "loss": 1.6234,
      "step": 60
    },
    {
      "epoch": 0.009199909509086796,
      "grad_norm": 0.3190295567941831,
      "learning_rate": 4.999060158443557e-07,
      "loss": 1.6061,
      "step": 61
    },
    {
      "epoch": 0.00935072769776035,
      "grad_norm": 0.3060496053543246,
      "learning_rate": 4.999029093623115e-07,
      "loss": 1.53,
      "step": 62
    },
    {
      "epoch": 0.009501545886433904,
      "grad_norm": 0.330007452265362,
      "learning_rate": 4.998997523830142e-07,
      "loss": 1.5874,
      "step": 63
    },
    {
      "epoch": 0.009652364075107459,
      "grad_norm": 0.2913872540073151,
      "learning_rate": 4.998965449071727e-07,
      "loss": 1.6215,
      "step": 64
    },
    {
      "epoch": 0.009803182263781011,
      "grad_norm": 0.3861447338808342,
      "learning_rate": 4.998932869355073e-07,
      "loss": 1.618,
      "step": 65
    },
    {
      "epoch": 0.009954000452454566,
      "grad_norm": 0.27958822059870586,
      "learning_rate": 4.998899784687492e-07,
      "loss": 1.6103,
      "step": 66
    },
    {
      "epoch": 0.01010481864112812,
      "grad_norm": 0.5931853797312046,
      "learning_rate": 4.998866195076416e-07,
      "loss": 1.5916,
      "step": 67
    },
    {
      "epoch": 0.010255636829801674,
      "grad_norm": 0.30425612593635215,
      "learning_rate": 4.998832100529383e-07,
      "loss": 1.6481,
      "step": 68
    },
    {
      "epoch": 0.010406455018475229,
      "grad_norm": 0.28173707892876365,
      "learning_rate": 4.998797501054052e-07,
      "loss": 1.6412,
      "step": 69
    },
    {
      "epoch": 0.010557273207148781,
      "grad_norm": 0.35484062928114324,
      "learning_rate": 4.998762396658189e-07,
      "loss": 1.6716,
      "step": 70
    },
    {
      "epoch": 0.010708091395822336,
      "grad_norm": 0.30696711292842577,
      "learning_rate": 4.998726787349676e-07,
      "loss": 1.6161,
      "step": 71
    },
    {
      "epoch": 0.01085890958449589,
      "grad_norm": 0.29830361910091235,
      "learning_rate": 4.99869067313651e-07,
      "loss": 1.5406,
      "step": 72
    },
    {
      "epoch": 0.011009727773169444,
      "grad_norm": 0.3257150413463108,
      "learning_rate": 4.998654054026798e-07,
      "loss": 1.6029,
      "step": 73
    },
    {
      "epoch": 0.011160545961842998,
      "grad_norm": 0.27608611779319664,
      "learning_rate": 4.998616930028763e-07,
      "loss": 1.5345,
      "step": 74
    },
    {
      "epoch": 0.011311364150516551,
      "grad_norm": 0.2801184141085234,
      "learning_rate": 4.998579301150741e-07,
      "loss": 1.6298,
      "step": 75
    },
    {
      "epoch": 0.011462182339190106,
      "grad_norm": 0.2877514992208577,
      "learning_rate": 4.998541167401179e-07,
      "loss": 1.5923,
      "step": 76
    },
    {
      "epoch": 0.01161300052786366,
      "grad_norm": 0.4974526858441575,
      "learning_rate": 4.998502528788641e-07,
      "loss": 1.6215,
      "step": 77
    },
    {
      "epoch": 0.011763818716537214,
      "grad_norm": 0.3049310065556733,
      "learning_rate": 4.998463385321801e-07,
      "loss": 1.6285,
      "step": 78
    },
    {
      "epoch": 0.011914636905210768,
      "grad_norm": 0.32712493893767625,
      "learning_rate": 4.99842373700945e-07,
      "loss": 1.6601,
      "step": 79
    },
    {
      "epoch": 0.012065455093884323,
      "grad_norm": 0.4232235008613389,
      "learning_rate": 4.998383583860487e-07,
      "loss": 1.6311,
      "step": 80
    },
    {
      "epoch": 0.012216273282557876,
      "grad_norm": 0.27530166675598666,
      "learning_rate": 4.99834292588393e-07,
      "loss": 1.5821,
      "step": 81
    },
    {
      "epoch": 0.01236709147123143,
      "grad_norm": 0.3199412588866916,
      "learning_rate": 4.998301763088906e-07,
      "loss": 1.7004,
      "step": 82
    },
    {
      "epoch": 0.012517909659904984,
      "grad_norm": 0.3765606480051871,
      "learning_rate": 4.99826009548466e-07,
      "loss": 1.572,
      "step": 83
    },
    {
      "epoch": 0.012668727848578538,
      "grad_norm": 0.27388681299904893,
      "learning_rate": 4.998217923080545e-07,
      "loss": 1.5805,
      "step": 84
    },
    {
      "epoch": 0.012819546037252093,
      "grad_norm": 0.28497127222933466,
      "learning_rate": 4.998175245886032e-07,
      "loss": 1.6096,
      "step": 85
    },
    {
      "epoch": 0.012970364225925646,
      "grad_norm": 0.288999756333011,
      "learning_rate": 4.998132063910701e-07,
      "loss": 1.629,
      "step": 86
    },
    {
      "epoch": 0.0131211824145992,
      "grad_norm": 0.2829852204496664,
      "learning_rate": 4.99808837716425e-07,
      "loss": 1.6398,
      "step": 87
    },
    {
      "epoch": 0.013272000603272755,
      "grad_norm": 0.3163616256907245,
      "learning_rate": 4.998044185656485e-07,
      "loss": 1.6663,
      "step": 88
    },
    {
      "epoch": 0.013422818791946308,
      "grad_norm": 0.29291670767485606,
      "learning_rate": 4.997999489397332e-07,
      "loss": 1.6779,
      "step": 89
    },
    {
      "epoch": 0.013573636980619863,
      "grad_norm": 0.3574975540951008,
      "learning_rate": 4.997954288396823e-07,
      "loss": 1.6485,
      "step": 90
    },
    {
      "epoch": 0.013724455169293416,
      "grad_norm": 0.2945864109199964,
      "learning_rate": 4.997908582665109e-07,
      "loss": 1.6502,
      "step": 91
    },
    {
      "epoch": 0.01387527335796697,
      "grad_norm": 0.2753915566219201,
      "learning_rate": 4.997862372212452e-07,
      "loss": 1.5492,
      "step": 92
    },
    {
      "epoch": 0.014026091546640525,
      "grad_norm": 0.26560879180698244,
      "learning_rate": 4.997815657049228e-07,
      "loss": 1.6694,
      "step": 93
    },
    {
      "epoch": 0.014176909735314078,
      "grad_norm": 0.28303864778500293,
      "learning_rate": 4.997768437185925e-07,
      "loss": 1.5992,
      "step": 94
    },
    {
      "epoch": 0.014327727923987633,
      "grad_norm": 0.27651869982324084,
      "learning_rate": 4.997720712633146e-07,
      "loss": 1.6188,
      "step": 95
    },
    {
      "epoch": 0.014478546112661188,
      "grad_norm": 0.26326254508243496,
      "learning_rate": 4.997672483401605e-07,
      "loss": 1.5778,
      "step": 96
    },
    {
      "epoch": 0.01462936430133474,
      "grad_norm": 0.2826952345075853,
      "learning_rate": 4.997623749502134e-07,
      "loss": 1.6432,
      "step": 97
    },
    {
      "epoch": 0.014780182490008295,
      "grad_norm": 0.27953137964722935,
      "learning_rate": 4.997574510945671e-07,
      "loss": 1.5695,
      "step": 98
    },
    {
      "epoch": 0.014931000678681848,
      "grad_norm": 0.2688364949225841,
      "learning_rate": 4.997524767743275e-07,
      "loss": 1.6274,
      "step": 99
    },
    {
      "epoch": 0.015081818867355403,
      "grad_norm": 0.3209356041339321,
      "learning_rate": 4.997474519906113e-07,
      "loss": 1.6223,
      "step": 100
    },
    {
      "epoch": 0.015232637056028958,
      "grad_norm": 1.0329372307400904,
      "learning_rate": 4.997423767445468e-07,
      "loss": 1.626,
      "step": 101
    },
    {
      "epoch": 0.01538345524470251,
      "grad_norm": 0.2925370222104452,
      "learning_rate": 4.997372510372735e-07,
      "loss": 1.5832,
      "step": 102
    },
    {
      "epoch": 0.015534273433376065,
      "grad_norm": 0.827547882991668,
      "learning_rate": 4.997320748699422e-07,
      "loss": 1.6391,
      "step": 103
    },
    {
      "epoch": 0.01568509162204962,
      "grad_norm": 0.28584586927783384,
      "learning_rate": 4.997268482437152e-07,
      "loss": 1.6098,
      "step": 104
    },
    {
      "epoch": 0.015835909810723175,
      "grad_norm": 0.2772653083687217,
      "learning_rate": 4.99721571159766e-07,
      "loss": 1.5489,
      "step": 105
    },
    {
      "epoch": 0.015986727999396728,
      "grad_norm": 0.3027002593820936,
      "learning_rate": 4.997162436192795e-07,
      "loss": 1.6175,
      "step": 106
    },
    {
      "epoch": 0.01613754618807028,
      "grad_norm": 0.26151021422707005,
      "learning_rate": 4.997108656234518e-07,
      "loss": 1.5712,
      "step": 107
    },
    {
      "epoch": 0.016288364376743834,
      "grad_norm": 0.2760524205676756,
      "learning_rate": 4.997054371734905e-07,
      "loss": 1.6033,
      "step": 108
    },
    {
      "epoch": 0.01643918256541739,
      "grad_norm": 0.2652710973293421,
      "learning_rate": 4.996999582706143e-07,
      "loss": 1.5926,
      "step": 109
    },
    {
      "epoch": 0.016590000754090943,
      "grad_norm": 0.27448663116920097,
      "learning_rate": 4.996944289160536e-07,
      "loss": 1.5478,
      "step": 110
    },
    {
      "epoch": 0.016740818942764496,
      "grad_norm": 0.27373441719549557,
      "learning_rate": 4.996888491110498e-07,
      "loss": 1.6436,
      "step": 111
    },
    {
      "epoch": 0.016891637131438052,
      "grad_norm": 0.2823563689658009,
      "learning_rate": 4.996832188568556e-07,
      "loss": 1.6573,
      "step": 112
    },
    {
      "epoch": 0.017042455320111605,
      "grad_norm": 0.2905540170801979,
      "learning_rate": 4.996775381547355e-07,
      "loss": 1.6346,
      "step": 113
    },
    {
      "epoch": 0.01719327350878516,
      "grad_norm": 0.36710435208412284,
      "learning_rate": 4.996718070059646e-07,
      "loss": 1.6623,
      "step": 114
    },
    {
      "epoch": 0.017344091697458715,
      "grad_norm": 0.28150968628183515,
      "learning_rate": 4.996660254118298e-07,
      "loss": 1.621,
      "step": 115
    },
    {
      "epoch": 0.017494909886132268,
      "grad_norm": 0.2908537614777647,
      "learning_rate": 4.996601933736293e-07,
      "loss": 1.5806,
      "step": 116
    },
    {
      "epoch": 0.01764572807480582,
      "grad_norm": 4.782372308328432,
      "learning_rate": 4.996543108926726e-07,
      "loss": 1.6366,
      "step": 117
    },
    {
      "epoch": 0.017796546263479377,
      "grad_norm": 0.28382687643398863,
      "learning_rate": 4.996483779702805e-07,
      "loss": 1.6322,
      "step": 118
    },
    {
      "epoch": 0.01794736445215293,
      "grad_norm": 0.26019481261096256,
      "learning_rate": 4.99642394607785e-07,
      "loss": 1.6209,
      "step": 119
    },
    {
      "epoch": 0.018098182640826483,
      "grad_norm": 0.3154808626611058,
      "learning_rate": 4.996363608065297e-07,
      "loss": 1.6731,
      "step": 120
    },
    {
      "epoch": 0.01824900082950004,
      "grad_norm": 0.28724383174878004,
      "learning_rate": 4.996302765678691e-07,
      "loss": 1.6151,
      "step": 121
    },
    {
      "epoch": 0.018399819018173592,
      "grad_norm": 0.2673293043423508,
      "learning_rate": 4.996241418931695e-07,
      "loss": 1.6209,
      "step": 122
    },
    {
      "epoch": 0.018550637206847145,
      "grad_norm": 0.2714980441671276,
      "learning_rate": 4.996179567838084e-07,
      "loss": 1.6281,
      "step": 123
    },
    {
      "epoch": 0.0187014553955207,
      "grad_norm": 0.27001328981908584,
      "learning_rate": 4.996117212411742e-07,
      "loss": 1.5936,
      "step": 124
    },
    {
      "epoch": 0.018852273584194255,
      "grad_norm": 7.240553966508071,
      "learning_rate": 4.996054352666673e-07,
      "loss": 1.6544,
      "step": 125
    },
    {
      "epoch": 0.019003091772867808,
      "grad_norm": 0.2909128295098797,
      "learning_rate": 4.99599098861699e-07,
      "loss": 1.5634,
      "step": 126
    },
    {
      "epoch": 0.01915390996154136,
      "grad_norm": 0.2564976792990476,
      "learning_rate": 4.995927120276918e-07,
      "loss": 1.5892,
      "step": 127
    },
    {
      "epoch": 0.019304728150214917,
      "grad_norm": 0.28813897636995855,
      "learning_rate": 4.995862747660799e-07,
      "loss": 1.5968,
      "step": 128
    },
    {
      "epoch": 0.01945554633888847,
      "grad_norm": 0.26825660858447004,
      "learning_rate": 4.995797870783087e-07,
      "loss": 1.6441,
      "step": 129
    },
    {
      "epoch": 0.019606364527562023,
      "grad_norm": 0.8593441639393585,
      "learning_rate": 4.995732489658349e-07,
      "loss": 1.6078,
      "step": 130
    },
    {
      "epoch": 0.01975718271623558,
      "grad_norm": 0.2744427040666157,
      "learning_rate": 4.995666604301263e-07,
      "loss": 1.5881,
      "step": 131
    },
    {
      "epoch": 0.019908000904909132,
      "grad_norm": 0.2935832854888586,
      "learning_rate": 4.995600214726624e-07,
      "loss": 1.6288,
      "step": 132
    },
    {
      "epoch": 0.020058819093582685,
      "grad_norm": 0.29119246468057447,
      "learning_rate": 4.995533320949337e-07,
      "loss": 1.6326,
      "step": 133
    },
    {
      "epoch": 0.02020963728225624,
      "grad_norm": 0.3205215551752403,
      "learning_rate": 4.995465922984423e-07,
      "loss": 1.6077,
      "step": 134
    },
    {
      "epoch": 0.020360455470929795,
      "grad_norm": 0.2689930184821362,
      "learning_rate": 4.995398020847013e-07,
      "loss": 1.6249,
      "step": 135
    },
    {
      "epoch": 0.020511273659603348,
      "grad_norm": 0.28616284464875474,
      "learning_rate": 4.995329614552355e-07,
      "loss": 1.5789,
      "step": 136
    },
    {
      "epoch": 0.0206620918482769,
      "grad_norm": 0.28300795023532677,
      "learning_rate": 4.995260704115808e-07,
      "loss": 1.6373,
      "step": 137
    },
    {
      "epoch": 0.020812910036950457,
      "grad_norm": 0.2625446722721147,
      "learning_rate": 4.995191289552842e-07,
      "loss": 1.6015,
      "step": 138
    },
    {
      "epoch": 0.02096372822562401,
      "grad_norm": 0.2847634969602461,
      "learning_rate": 4.995121370879045e-07,
      "loss": 1.5301,
      "step": 139
    },
    {
      "epoch": 0.021114546414297563,
      "grad_norm": 0.5003742319182984,
      "learning_rate": 4.995050948110115e-07,
      "loss": 1.5855,
      "step": 140
    },
    {
      "epoch": 0.02126536460297112,
      "grad_norm": 0.3212774484618759,
      "learning_rate": 4.994980021261864e-07,
      "loss": 1.6243,
      "step": 141
    },
    {
      "epoch": 0.021416182791644672,
      "grad_norm": 0.26244417253211844,
      "learning_rate": 4.994908590350218e-07,
      "loss": 1.5763,
      "step": 142
    },
    {
      "epoch": 0.021567000980318225,
      "grad_norm": 0.2657090011462843,
      "learning_rate": 4.994836655391214e-07,
      "loss": 1.6437,
      "step": 143
    },
    {
      "epoch": 0.02171781916899178,
      "grad_norm": 0.27420630718454775,
      "learning_rate": 4.994764216401004e-07,
      "loss": 1.5617,
      "step": 144
    },
    {
      "epoch": 0.021868637357665335,
      "grad_norm": 0.2884668926439784,
      "learning_rate": 4.994691273395852e-07,
      "loss": 1.6123,
      "step": 145
    },
    {
      "epoch": 0.022019455546338888,
      "grad_norm": 0.3301484708204669,
      "learning_rate": 4.994617826392137e-07,
      "loss": 1.6564,
      "step": 146
    },
    {
      "epoch": 0.022170273735012444,
      "grad_norm": 0.2591901169274755,
      "learning_rate": 4.994543875406349e-07,
      "loss": 1.6437,
      "step": 147
    },
    {
      "epoch": 0.022321091923685997,
      "grad_norm": 0.2592192947358982,
      "learning_rate": 4.994469420455093e-07,
      "loss": 1.6852,
      "step": 148
    },
    {
      "epoch": 0.02247191011235955,
      "grad_norm": 0.25884995709127223,
      "learning_rate": 4.994394461555087e-07,
      "loss": 1.5904,
      "step": 149
    },
    {
      "epoch": 0.022622728301033103,
      "grad_norm": 0.2632847561494921,
      "learning_rate": 4.99431899872316e-07,
      "loss": 1.604,
      "step": 150
    },
    {
      "epoch": 0.02277354648970666,
      "grad_norm": 0.27459575640410483,
      "learning_rate": 4.994243031976255e-07,
      "loss": 1.5783,
      "step": 151
    },
    {
      "epoch": 0.022924364678380212,
      "grad_norm": 0.263011002646199,
      "learning_rate": 4.99416656133143e-07,
      "loss": 1.612,
      "step": 152
    },
    {
      "epoch": 0.023075182867053765,
      "grad_norm": 0.452258821903041,
      "learning_rate": 4.994089586805855e-07,
      "loss": 1.5412,
      "step": 153
    },
    {
      "epoch": 0.02322600105572732,
      "grad_norm": 0.2636086865413926,
      "learning_rate": 4.994012108416812e-07,
      "loss": 1.5354,
      "step": 154
    },
    {
      "epoch": 0.023376819244400875,
      "grad_norm": 0.2698908588403687,
      "learning_rate": 4.993934126181698e-07,
      "loss": 1.5574,
      "step": 155
    },
    {
      "epoch": 0.023527637433074428,
      "grad_norm": 0.2620736602088264,
      "learning_rate": 4.993855640118023e-07,
      "loss": 1.5399,
      "step": 156
    },
    {
      "epoch": 0.023678455621747984,
      "grad_norm": 0.2662131144061968,
      "learning_rate": 4.993776650243408e-07,
      "loss": 1.5513,
      "step": 157
    },
    {
      "epoch": 0.023829273810421537,
      "grad_norm": 0.33202641823726214,
      "learning_rate": 4.993697156575589e-07,
      "loss": 1.5834,
      "step": 158
    },
    {
      "epoch": 0.02398009199909509,
      "grad_norm": 0.383715850377704,
      "learning_rate": 4.993617159132415e-07,
      "loss": 1.6506,
      "step": 159
    },
    {
      "epoch": 0.024130910187768646,
      "grad_norm": 0.2642653431126336,
      "learning_rate": 4.993536657931846e-07,
      "loss": 1.5637,
      "step": 160
    },
    {
      "epoch": 0.0242817283764422,
      "grad_norm": 0.292189030769218,
      "learning_rate": 4.993455652991959e-07,
      "loss": 1.6019,
      "step": 161
    },
    {
      "epoch": 0.024432546565115752,
      "grad_norm": 0.27074399593196924,
      "learning_rate": 4.993374144330942e-07,
      "loss": 1.6746,
      "step": 162
    },
    {
      "epoch": 0.02458336475378931,
      "grad_norm": 0.2460380594866087,
      "learning_rate": 4.993292131967094e-07,
      "loss": 1.6073,
      "step": 163
    },
    {
      "epoch": 0.02473418294246286,
      "grad_norm": 0.9355172258960992,
      "learning_rate": 4.993209615918832e-07,
      "loss": 1.6106,
      "step": 164
    },
    {
      "epoch": 0.024885001131136415,
      "grad_norm": 0.2595553364190686,
      "learning_rate": 4.99312659620468e-07,
      "loss": 1.5158,
      "step": 165
    },
    {
      "epoch": 0.025035819319809968,
      "grad_norm": 0.31310147215799994,
      "learning_rate": 4.993043072843282e-07,
      "loss": 1.5527,
      "step": 166
    },
    {
      "epoch": 0.025186637508483524,
      "grad_norm": 0.30063636617290196,
      "learning_rate": 4.992959045853388e-07,
      "loss": 1.6818,
      "step": 167
    },
    {
      "epoch": 0.025337455697157077,
      "grad_norm": 0.26319554722690824,
      "learning_rate": 4.992874515253867e-07,
      "loss": 1.6324,
      "step": 168
    },
    {
      "epoch": 0.02548827388583063,
      "grad_norm": 0.2821630706130611,
      "learning_rate": 4.992789481063698e-07,
      "loss": 1.6648,
      "step": 169
    },
    {
      "epoch": 0.025639092074504186,
      "grad_norm": 0.2642207888569884,
      "learning_rate": 4.992703943301973e-07,
      "loss": 1.6741,
      "step": 170
    },
    {
      "epoch": 0.02578991026317774,
      "grad_norm": 0.26449734982639606,
      "learning_rate": 4.992617901987897e-07,
      "loss": 1.578,
      "step": 171
    },
    {
      "epoch": 0.025940728451851292,
      "grad_norm": 0.26710134208771796,
      "learning_rate": 4.99253135714079e-07,
      "loss": 1.5813,
      "step": 172
    },
    {
      "epoch": 0.02609154664052485,
      "grad_norm": 0.2561809201414807,
      "learning_rate": 4.992444308780084e-07,
      "loss": 1.566,
      "step": 173
    },
    {
      "epoch": 0.0262423648291984,
      "grad_norm": 0.29393431439094037,
      "learning_rate": 4.992356756925323e-07,
      "loss": 1.629,
      "step": 174
    },
    {
      "epoch": 0.026393183017871955,
      "grad_norm": 0.25906234137363926,
      "learning_rate": 4.992268701596166e-07,
      "loss": 1.5759,
      "step": 175
    },
    {
      "epoch": 0.02654400120654551,
      "grad_norm": 0.26821181727998744,
      "learning_rate": 4.992180142812383e-07,
      "loss": 1.6519,
      "step": 176
    },
    {
      "epoch": 0.026694819395219064,
      "grad_norm": 0.3624980940371162,
      "learning_rate": 4.992091080593858e-07,
      "loss": 1.605,
      "step": 177
    },
    {
      "epoch": 0.026845637583892617,
      "grad_norm": 0.2623854348627949,
      "learning_rate": 4.99200151496059e-07,
      "loss": 1.6117,
      "step": 178
    },
    {
      "epoch": 0.02699645577256617,
      "grad_norm": 0.2627813158352471,
      "learning_rate": 4.991911445932686e-07,
      "loss": 1.5814,
      "step": 179
    },
    {
      "epoch": 0.027147273961239726,
      "grad_norm": 0.2594695889060515,
      "learning_rate": 4.991820873530371e-07,
      "loss": 1.52,
      "step": 180
    },
    {
      "epoch": 0.02729809214991328,
      "grad_norm": 0.30046272167288224,
      "learning_rate": 4.991729797773981e-07,
      "loss": 1.6095,
      "step": 181
    },
    {
      "epoch": 0.027448910338586832,
      "grad_norm": 0.254940160039734,
      "learning_rate": 4.991638218683966e-07,
      "loss": 1.5254,
      "step": 182
    },
    {
      "epoch": 0.02759972852726039,
      "grad_norm": 1.303177270729485,
      "learning_rate": 4.991546136280887e-07,
      "loss": 1.5905,
      "step": 183
    },
    {
      "epoch": 0.02775054671593394,
      "grad_norm": 0.31877478376641744,
      "learning_rate": 4.991453550585418e-07,
      "loss": 1.588,
      "step": 184
    },
    {
      "epoch": 0.027901364904607494,
      "grad_norm": 0.24838455111117674,
      "learning_rate": 4.99136046161835e-07,
      "loss": 1.5996,
      "step": 185
    },
    {
      "epoch": 0.02805218309328105,
      "grad_norm": 0.2848153559432735,
      "learning_rate": 4.991266869400582e-07,
      "loss": 1.5963,
      "step": 186
    },
    {
      "epoch": 0.028203001281954604,
      "grad_norm": 0.25828269152635797,
      "learning_rate": 4.991172773953129e-07,
      "loss": 1.5755,
      "step": 187
    },
    {
      "epoch": 0.028353819470628157,
      "grad_norm": 0.2653958489725172,
      "learning_rate": 4.991078175297117e-07,
      "loss": 1.6307,
      "step": 188
    },
    {
      "epoch": 0.028504637659301713,
      "grad_norm": 0.2690080760631909,
      "learning_rate": 4.990983073453789e-07,
      "loss": 1.6036,
      "step": 189
    },
    {
      "epoch": 0.028655455847975266,
      "grad_norm": 0.2618006127959753,
      "learning_rate": 4.990887468444496e-07,
      "loss": 1.615,
      "step": 190
    },
    {
      "epoch": 0.02880627403664882,
      "grad_norm": 0.268044151489942,
      "learning_rate": 4.990791360290704e-07,
      "loss": 1.5995,
      "step": 191
    },
    {
      "epoch": 0.028957092225322376,
      "grad_norm": 0.26809538061657656,
      "learning_rate": 4.990694749013993e-07,
      "loss": 1.6634,
      "step": 192
    },
    {
      "epoch": 0.02910791041399593,
      "grad_norm": 0.26956024396526895,
      "learning_rate": 4.990597634636054e-07,
      "loss": 1.6116,
      "step": 193
    },
    {
      "epoch": 0.02925872860266948,
      "grad_norm": 0.25746207958100037,
      "learning_rate": 4.990500017178693e-07,
      "loss": 1.6472,
      "step": 194
    },
    {
      "epoch": 0.029409546791343034,
      "grad_norm": 0.26012498294026976,
      "learning_rate": 4.990401896663827e-07,
      "loss": 1.6604,
      "step": 195
    },
    {
      "epoch": 0.02956036498001659,
      "grad_norm": 0.3002010469788411,
      "learning_rate": 4.990303273113489e-07,
      "loss": 1.6303,
      "step": 196
    },
    {
      "epoch": 0.029711183168690144,
      "grad_norm": 0.27514115715685794,
      "learning_rate": 4.99020414654982e-07,
      "loss": 1.6351,
      "step": 197
    },
    {
      "epoch": 0.029862001357363697,
      "grad_norm": 0.2503927456470816,
      "learning_rate": 4.990104516995079e-07,
      "loss": 1.5543,
      "step": 198
    },
    {
      "epoch": 0.030012819546037253,
      "grad_norm": 0.27626109103815416,
      "learning_rate": 4.990004384471633e-07,
      "loss": 1.6182,
      "step": 199
    },
    {
      "epoch": 0.030163637734710806,
      "grad_norm": 0.2781518165419307,
      "learning_rate": 4.989903749001969e-07,
      "loss": 1.6649,
      "step": 200
    },
    {
      "epoch": 0.03031445592338436,
      "grad_norm": 0.25231222205522835,
      "learning_rate": 4.989802610608679e-07,
      "loss": 1.5731,
      "step": 201
    },
    {
      "epoch": 0.030465274112057916,
      "grad_norm": 0.26378173793047377,
      "learning_rate": 4.989700969314472e-07,
      "loss": 1.5373,
      "step": 202
    },
    {
      "epoch": 0.03061609230073147,
      "grad_norm": 0.3340714072382629,
      "learning_rate": 4.989598825142171e-07,
      "loss": 1.5694,
      "step": 203
    },
    {
      "epoch": 0.03076691048940502,
      "grad_norm": 0.26463313883572454,
      "learning_rate": 4.989496178114709e-07,
      "loss": 1.6428,
      "step": 204
    },
    {
      "epoch": 0.030917728678078578,
      "grad_norm": 0.2832551149133356,
      "learning_rate": 4.989393028255134e-07,
      "loss": 1.5773,
      "step": 205
    },
    {
      "epoch": 0.03106854686675213,
      "grad_norm": 0.27831436524066777,
      "learning_rate": 4.989289375586605e-07,
      "loss": 1.6353,
      "step": 206
    },
    {
      "epoch": 0.031219365055425684,
      "grad_norm": 0.26392410421703627,
      "learning_rate": 4.989185220132397e-07,
      "loss": 1.6607,
      "step": 207
    },
    {
      "epoch": 0.03137018324409924,
      "grad_norm": 0.2593580146156735,
      "learning_rate": 4.989080561915895e-07,
      "loss": 1.6365,
      "step": 208
    },
    {
      "epoch": 0.03152100143277279,
      "grad_norm": 0.29962645492914997,
      "learning_rate": 4.988975400960596e-07,
      "loss": 1.6178,
      "step": 209
    },
    {
      "epoch": 0.03167181962144635,
      "grad_norm": 0.2761177620930967,
      "learning_rate": 4.988869737290115e-07,
      "loss": 1.587,
      "step": 210
    },
    {
      "epoch": 0.0318226378101199,
      "grad_norm": 0.27633099258498567,
      "learning_rate": 4.988763570928174e-07,
      "loss": 1.6454,
      "step": 211
    },
    {
      "epoch": 0.031973455998793456,
      "grad_norm": 0.2463702497362132,
      "learning_rate": 4.988656901898612e-07,
      "loss": 1.5788,
      "step": 212
    },
    {
      "epoch": 0.03212427418746701,
      "grad_norm": 0.25496481204896715,
      "learning_rate": 4.988549730225378e-07,
      "loss": 1.5861,
      "step": 213
    },
    {
      "epoch": 0.03227509237614056,
      "grad_norm": 0.2580606120283535,
      "learning_rate": 4.988442055932536e-07,
      "loss": 1.5762,
      "step": 214
    },
    {
      "epoch": 0.032425910564814114,
      "grad_norm": 0.2771695020338603,
      "learning_rate": 4.988333879044262e-07,
      "loss": 1.581,
      "step": 215
    },
    {
      "epoch": 0.03257672875348767,
      "grad_norm": 0.2869703933581396,
      "learning_rate": 4.988225199584844e-07,
      "loss": 1.6003,
      "step": 216
    },
    {
      "epoch": 0.03272754694216123,
      "grad_norm": 0.2564015242431221,
      "learning_rate": 4.988116017578685e-07,
      "loss": 1.6345,
      "step": 217
    },
    {
      "epoch": 0.03287836513083478,
      "grad_norm": 0.5442682369824418,
      "learning_rate": 4.988006333050298e-07,
      "loss": 1.6365,
      "step": 218
    },
    {
      "epoch": 0.03302918331950833,
      "grad_norm": 0.25371420603547495,
      "learning_rate": 4.987896146024312e-07,
      "loss": 1.5389,
      "step": 219
    },
    {
      "epoch": 0.033180001508181886,
      "grad_norm": 0.2589325925383411,
      "learning_rate": 4.987785456525468e-07,
      "loss": 1.6327,
      "step": 220
    },
    {
      "epoch": 0.03333081969685544,
      "grad_norm": 0.4127227412991189,
      "learning_rate": 4.987674264578614e-07,
      "loss": 1.6354,
      "step": 221
    },
    {
      "epoch": 0.03348163788552899,
      "grad_norm": 0.2616917271457671,
      "learning_rate": 4.987562570208722e-07,
      "loss": 1.5691,
      "step": 222
    },
    {
      "epoch": 0.03363245607420255,
      "grad_norm": 0.262500735261184,
      "learning_rate": 4.987450373440868e-07,
      "loss": 1.5843,
      "step": 223
    },
    {
      "epoch": 0.033783274262876105,
      "grad_norm": 2.907675552975996,
      "learning_rate": 4.987337674300242e-07,
      "loss": 1.6643,
      "step": 224
    },
    {
      "epoch": 0.03393409245154966,
      "grad_norm": 0.633109876707346,
      "learning_rate": 4.98722447281215e-07,
      "loss": 1.5513,
      "step": 225
    },
    {
      "epoch": 0.03408491064022321,
      "grad_norm": 0.31204975345051955,
      "learning_rate": 4.987110769002009e-07,
      "loss": 1.6166,
      "step": 226
    },
    {
      "epoch": 0.034235728828896764,
      "grad_norm": 0.2694667377758279,
      "learning_rate": 4.986996562895348e-07,
      "loss": 1.5886,
      "step": 227
    },
    {
      "epoch": 0.03438654701757032,
      "grad_norm": 0.3105182287125855,
      "learning_rate": 4.986881854517811e-07,
      "loss": 1.5918,
      "step": 228
    },
    {
      "epoch": 0.03453736520624387,
      "grad_norm": 0.25509511385760725,
      "learning_rate": 4.986766643895152e-07,
      "loss": 1.5409,
      "step": 229
    },
    {
      "epoch": 0.03468818339491743,
      "grad_norm": 6.222098257658109,
      "learning_rate": 4.98665093105324e-07,
      "loss": 1.6615,
      "step": 230
    },
    {
      "epoch": 0.03483900158359098,
      "grad_norm": 0.4035810998796959,
      "learning_rate": 4.986534716018054e-07,
      "loss": 1.591,
      "step": 231
    },
    {
      "epoch": 0.034989819772264535,
      "grad_norm": 0.25480384926234323,
      "learning_rate": 4.986417998815691e-07,
      "loss": 1.4973,
      "step": 232
    },
    {
      "epoch": 0.03514063796093809,
      "grad_norm": 0.26156856896321407,
      "learning_rate": 4.986300779472355e-07,
      "loss": 1.6332,
      "step": 233
    },
    {
      "epoch": 0.03529145614961164,
      "grad_norm": 0.2600770194351854,
      "learning_rate": 4.986183058014366e-07,
      "loss": 1.6398,
      "step": 234
    },
    {
      "epoch": 0.035442274338285194,
      "grad_norm": 0.25367766845188094,
      "learning_rate": 4.986064834468155e-07,
      "loss": 1.6245,
      "step": 235
    },
    {
      "epoch": 0.035593092526958754,
      "grad_norm": 0.26431562176833495,
      "learning_rate": 4.985946108860268e-07,
      "loss": 1.58,
      "step": 236
    },
    {
      "epoch": 0.03574391071563231,
      "grad_norm": 0.31760663960749624,
      "learning_rate": 4.985826881217361e-07,
      "loss": 1.572,
      "step": 237
    },
    {
      "epoch": 0.03589472890430586,
      "grad_norm": 0.24981232976175427,
      "learning_rate": 4.985707151566205e-07,
      "loss": 1.5037,
      "step": 238
    },
    {
      "epoch": 0.03604554709297941,
      "grad_norm": 0.2998208500980713,
      "learning_rate": 4.985586919933681e-07,
      "loss": 1.557,
      "step": 239
    },
    {
      "epoch": 0.036196365281652966,
      "grad_norm": 0.26446444256321233,
      "learning_rate": 4.985466186346787e-07,
      "loss": 1.5727,
      "step": 240
    },
    {
      "epoch": 0.03634718347032652,
      "grad_norm": 0.24675850916193817,
      "learning_rate": 4.985344950832631e-07,
      "loss": 1.5936,
      "step": 241
    },
    {
      "epoch": 0.03649800165900008,
      "grad_norm": 0.2766463983732949,
      "learning_rate": 4.985223213418433e-07,
      "loss": 1.6282,
      "step": 242
    },
    {
      "epoch": 0.03664881984767363,
      "grad_norm": 0.2557536990136019,
      "learning_rate": 4.985100974131526e-07,
      "loss": 1.6208,
      "step": 243
    },
    {
      "epoch": 0.036799638036347185,
      "grad_norm": 0.2599746525688643,
      "learning_rate": 4.984978232999357e-07,
      "loss": 1.6953,
      "step": 244
    },
    {
      "epoch": 0.03695045622502074,
      "grad_norm": 0.3175602608980726,
      "learning_rate": 4.984854990049486e-07,
      "loss": 1.5153,
      "step": 245
    },
    {
      "epoch": 0.03710127441369429,
      "grad_norm": 0.2677265443419408,
      "learning_rate": 4.984731245309581e-07,
      "loss": 1.5371,
      "step": 246
    },
    {
      "epoch": 0.037252092602367844,
      "grad_norm": 0.2641204549342982,
      "learning_rate": 4.984606998807431e-07,
      "loss": 1.6204,
      "step": 247
    },
    {
      "epoch": 0.0374029107910414,
      "grad_norm": 0.24941313609436178,
      "learning_rate": 4.984482250570931e-07,
      "loss": 1.6269,
      "step": 248
    },
    {
      "epoch": 0.037553728979714956,
      "grad_norm": 0.2638709649184524,
      "learning_rate": 4.98435700062809e-07,
      "loss": 1.6158,
      "step": 249
    },
    {
      "epoch": 0.03770454716838851,
      "grad_norm": 0.2656121521078653,
      "learning_rate": 4.984231249007031e-07,
      "loss": 1.611,
      "step": 250
    },
    {
      "epoch": 0.03785536535706206,
      "grad_norm": 0.28853552223521767,
      "learning_rate": 4.984104995735987e-07,
      "loss": 1.6577,
      "step": 251
    },
    {
      "epoch": 0.038006183545735615,
      "grad_norm": 0.2517118886233882,
      "learning_rate": 4.983978240843308e-07,
      "loss": 1.5289,
      "step": 252
    },
    {
      "epoch": 0.03815700173440917,
      "grad_norm": 3.1305018983774993,
      "learning_rate": 4.983850984357453e-07,
      "loss": 1.5685,
      "step": 253
    },
    {
      "epoch": 0.03830781992308272,
      "grad_norm": 0.2549610619095195,
      "learning_rate": 4.983723226306996e-07,
      "loss": 1.5981,
      "step": 254
    },
    {
      "epoch": 0.03845863811175628,
      "grad_norm": 0.26554576080157216,
      "learning_rate": 4.983594966720621e-07,
      "loss": 1.6547,
      "step": 255
    },
    {
      "epoch": 0.038609456300429834,
      "grad_norm": 0.3106860712826546,
      "learning_rate": 4.983466205627127e-07,
      "loss": 1.6248,
      "step": 256
    },
    {
      "epoch": 0.03876027448910339,
      "grad_norm": 0.26254029852902167,
      "learning_rate": 4.983336943055424e-07,
      "loss": 1.5561,
      "step": 257
    },
    {
      "epoch": 0.03891109267777694,
      "grad_norm": 0.27726026922899555,
      "learning_rate": 4.983207179034534e-07,
      "loss": 1.6218,
      "step": 258
    },
    {
      "epoch": 0.03906191086645049,
      "grad_norm": 0.29320880460173404,
      "learning_rate": 4.983076913593596e-07,
      "loss": 1.6269,
      "step": 259
    },
    {
      "epoch": 0.039212729055124046,
      "grad_norm": 0.29577469982501586,
      "learning_rate": 4.982946146761856e-07,
      "loss": 1.5869,
      "step": 260
    },
    {
      "epoch": 0.0393635472437976,
      "grad_norm": 0.2559213435574554,
      "learning_rate": 4.982814878568675e-07,
      "loss": 1.5463,
      "step": 261
    },
    {
      "epoch": 0.03951436543247116,
      "grad_norm": 0.25707483616437765,
      "learning_rate": 4.982683109043526e-07,
      "loss": 1.6543,
      "step": 262
    },
    {
      "epoch": 0.03966518362114471,
      "grad_norm": 0.41074618361188825,
      "learning_rate": 4.982550838215998e-07,
      "loss": 1.6011,
      "step": 263
    },
    {
      "epoch": 0.039816001809818265,
      "grad_norm": 0.25765331574805506,
      "learning_rate": 4.982418066115786e-07,
      "loss": 1.5752,
      "step": 264
    },
    {
      "epoch": 0.03996681999849182,
      "grad_norm": 0.256754926319508,
      "learning_rate": 4.982284792772704e-07,
      "loss": 1.5779,
      "step": 265
    },
    {
      "epoch": 0.04011763818716537,
      "grad_norm": 0.252978048670289,
      "learning_rate": 4.982151018216676e-07,
      "loss": 1.5641,
      "step": 266
    },
    {
      "epoch": 0.040268456375838924,
      "grad_norm": 0.2615264897773922,
      "learning_rate": 4.982016742477735e-07,
      "loss": 1.615,
      "step": 267
    },
    {
      "epoch": 0.04041927456451248,
      "grad_norm": 0.2470323509147562,
      "learning_rate": 4.981881965586032e-07,
      "loss": 1.6021,
      "step": 268
    },
    {
      "epoch": 0.040570092753186036,
      "grad_norm": 0.27623642277328214,
      "learning_rate": 4.981746687571828e-07,
      "loss": 1.5625,
      "step": 269
    },
    {
      "epoch": 0.04072091094185959,
      "grad_norm": 0.2681915393759785,
      "learning_rate": 4.981610908465498e-07,
      "loss": 1.6154,
      "step": 270
    },
    {
      "epoch": 0.04087172913053314,
      "grad_norm": 0.2545576182077647,
      "learning_rate": 4.981474628297527e-07,
      "loss": 1.5684,
      "step": 271
    },
    {
      "epoch": 0.041022547319206695,
      "grad_norm": 0.2751802445237094,
      "learning_rate": 4.981337847098515e-07,
      "loss": 1.598,
      "step": 272
    },
    {
      "epoch": 0.04117336550788025,
      "grad_norm": 0.3117722758748653,
      "learning_rate": 4.981200564899171e-07,
      "loss": 1.5139,
      "step": 273
    },
    {
      "epoch": 0.0413241836965538,
      "grad_norm": 0.6489078607208647,
      "learning_rate": 4.981062781730322e-07,
      "loss": 1.651,
      "step": 274
    },
    {
      "epoch": 0.04147500188522736,
      "grad_norm": 0.25981217410577917,
      "learning_rate": 4.980924497622901e-07,
      "loss": 1.5972,
      "step": 275
    },
    {
      "epoch": 0.041625820073900914,
      "grad_norm": 0.3774374807557159,
      "learning_rate": 4.98078571260796e-07,
      "loss": 1.6089,
      "step": 276
    },
    {
      "epoch": 0.04177663826257447,
      "grad_norm": 0.2563361962896586,
      "learning_rate": 4.980646426716658e-07,
      "loss": 1.6366,
      "step": 277
    },
    {
      "epoch": 0.04192745645124802,
      "grad_norm": 0.24794636056007546,
      "learning_rate": 4.98050663998027e-07,
      "loss": 1.541,
      "step": 278
    },
    {
      "epoch": 0.04207827463992157,
      "grad_norm": 0.2629941923704229,
      "learning_rate": 4.980366352430181e-07,
      "loss": 1.6291,
      "step": 279
    },
    {
      "epoch": 0.042229092828595126,
      "grad_norm": 0.2552870222410783,
      "learning_rate": 4.980225564097892e-07,
      "loss": 1.6124,
      "step": 280
    },
    {
      "epoch": 0.042379911017268686,
      "grad_norm": 0.3106822885855718,
      "learning_rate": 4.980084275015011e-07,
      "loss": 1.5888,
      "step": 281
    },
    {
      "epoch": 0.04253072920594224,
      "grad_norm": 0.25148761118484103,
      "learning_rate": 4.979942485213265e-07,
      "loss": 1.5989,
      "step": 282
    },
    {
      "epoch": 0.04268154739461579,
      "grad_norm": 0.25753173598937684,
      "learning_rate": 4.979800194724486e-07,
      "loss": 1.5735,
      "step": 283
    },
    {
      "epoch": 0.042832365583289345,
      "grad_norm": 2.4177527896744375,
      "learning_rate": 4.979657403580625e-07,
      "loss": 1.7179,
      "step": 284
    },
    {
      "epoch": 0.0429831837719629,
      "grad_norm": 0.2997357031955897,
      "learning_rate": 4.979514111813742e-07,
      "loss": 1.5998,
      "step": 285
    },
    {
      "epoch": 0.04313400196063645,
      "grad_norm": 0.25092531904846166,
      "learning_rate": 4.97937031945601e-07,
      "loss": 1.6163,
      "step": 286
    },
    {
      "epoch": 0.043284820149310003,
      "grad_norm": 0.2695635528659169,
      "learning_rate": 4.979226026539715e-07,
      "loss": 1.6604,
      "step": 287
    },
    {
      "epoch": 0.04343563833798356,
      "grad_norm": 0.2765857609818474,
      "learning_rate": 4.979081233097256e-07,
      "loss": 1.5919,
      "step": 288
    },
    {
      "epoch": 0.043586456526657116,
      "grad_norm": 0.32188937842884396,
      "learning_rate": 4.978935939161141e-07,
      "loss": 1.5846,
      "step": 289
    },
    {
      "epoch": 0.04373727471533067,
      "grad_norm": 0.2416812364013644,
      "learning_rate": 4.978790144763993e-07,
      "loss": 1.5054,
      "step": 290
    },
    {
      "epoch": 0.04388809290400422,
      "grad_norm": 0.26113644000145003,
      "learning_rate": 4.97864384993855e-07,
      "loss": 1.5911,
      "step": 291
    },
    {
      "epoch": 0.044038911092677775,
      "grad_norm": 0.2572965207604668,
      "learning_rate": 4.978497054717656e-07,
      "loss": 1.6129,
      "step": 292
    },
    {
      "epoch": 0.04418972928135133,
      "grad_norm": 2.8605673966482494,
      "learning_rate": 4.978349759134273e-07,
      "loss": 1.6077,
      "step": 293
    },
    {
      "epoch": 0.04434054747002489,
      "grad_norm": 0.27859193377435737,
      "learning_rate": 4.978201963221472e-07,
      "loss": 1.6179,
      "step": 294
    },
    {
      "epoch": 0.04449136565869844,
      "grad_norm": 0.26907059310024917,
      "learning_rate": 4.978053667012438e-07,
      "loss": 1.5576,
      "step": 295
    },
    {
      "epoch": 0.044642183847371994,
      "grad_norm": 0.29052532677945747,
      "learning_rate": 4.977904870540467e-07,
      "loss": 1.649,
      "step": 296
    },
    {
      "epoch": 0.04479300203604555,
      "grad_norm": 0.2472154773342736,
      "learning_rate": 4.977755573838969e-07,
      "loss": 1.5837,
      "step": 297
    },
    {
      "epoch": 0.0449438202247191,
      "grad_norm": 0.2559592266553768,
      "learning_rate": 4.977605776941466e-07,
      "loss": 1.5991,
      "step": 298
    },
    {
      "epoch": 0.04509463841339265,
      "grad_norm": 0.2580734058119581,
      "learning_rate": 4.977455479881591e-07,
      "loss": 1.6362,
      "step": 299
    },
    {
      "epoch": 0.045245456602066206,
      "grad_norm": 0.2509784344630137,
      "learning_rate": 4.977304682693089e-07,
      "loss": 1.5833,
      "step": 300
    },
    {
      "epoch": 0.045396274790739766,
      "grad_norm": 0.25643481159429315,
      "learning_rate": 4.97715338540982e-07,
      "loss": 1.602,
      "step": 301
    },
    {
      "epoch": 0.04554709297941332,
      "grad_norm": 0.25606930376257603,
      "learning_rate": 4.977001588065754e-07,
      "loss": 1.6303,
      "step": 302
    },
    {
      "epoch": 0.04569791116808687,
      "grad_norm": 0.25070608064964595,
      "learning_rate": 4.976849290694974e-07,
      "loss": 1.5829,
      "step": 303
    },
    {
      "epoch": 0.045848729356760425,
      "grad_norm": 0.25380401662714464,
      "learning_rate": 4.976696493331676e-07,
      "loss": 1.6079,
      "step": 304
    },
    {
      "epoch": 0.04599954754543398,
      "grad_norm": 0.3408665041576288,
      "learning_rate": 4.976543196010167e-07,
      "loss": 1.624,
      "step": 305
    },
    {
      "epoch": 0.04615036573410753,
      "grad_norm": 0.2659038642666762,
      "learning_rate": 4.976389398764865e-07,
      "loss": 1.5253,
      "step": 306
    },
    {
      "epoch": 0.04630118392278109,
      "grad_norm": 0.25436011693157684,
      "learning_rate": 4.976235101630303e-07,
      "loss": 1.5478,
      "step": 307
    },
    {
      "epoch": 0.04645200211145464,
      "grad_norm": 0.24834201891089555,
      "learning_rate": 4.976080304641127e-07,
      "loss": 1.6087,
      "step": 308
    },
    {
      "epoch": 0.046602820300128196,
      "grad_norm": 0.3203590066782244,
      "learning_rate": 4.975925007832091e-07,
      "loss": 1.5459,
      "step": 309
    },
    {
      "epoch": 0.04675363848880175,
      "grad_norm": 0.26593109134916937,
      "learning_rate": 4.975769211238065e-07,
      "loss": 1.6162,
      "step": 310
    },
    {
      "epoch": 0.0469044566774753,
      "grad_norm": 0.25633161565087326,
      "learning_rate": 4.97561291489403e-07,
      "loss": 1.5698,
      "step": 311
    },
    {
      "epoch": 0.047055274866148855,
      "grad_norm": 0.24959966756569993,
      "learning_rate": 4.975456118835078e-07,
      "loss": 1.6237,
      "step": 312
    },
    {
      "epoch": 0.047206093054822415,
      "grad_norm": 0.25286897448991674,
      "learning_rate": 4.975298823096416e-07,
      "loss": 1.6881,
      "step": 313
    },
    {
      "epoch": 0.04735691124349597,
      "grad_norm": 0.26036485809927373,
      "learning_rate": 4.975141027713359e-07,
      "loss": 1.6125,
      "step": 314
    },
    {
      "epoch": 0.04750772943216952,
      "grad_norm": 0.25618533844773894,
      "learning_rate": 4.974982732721338e-07,
      "loss": 1.55,
      "step": 315
    },
    {
      "epoch": 0.047658547620843074,
      "grad_norm": 0.33343355928017965,
      "learning_rate": 4.974823938155895e-07,
      "loss": 1.5757,
      "step": 316
    },
    {
      "epoch": 0.04780936580951663,
      "grad_norm": 0.27528927934531816,
      "learning_rate": 4.974664644052684e-07,
      "loss": 1.5858,
      "step": 317
    },
    {
      "epoch": 0.04796018399819018,
      "grad_norm": 0.25562091394247194,
      "learning_rate": 4.974504850447471e-07,
      "loss": 1.5687,
      "step": 318
    },
    {
      "epoch": 0.04811100218686373,
      "grad_norm": 0.24416275184111966,
      "learning_rate": 4.974344557376135e-07,
      "loss": 1.5786,
      "step": 319
    },
    {
      "epoch": 0.04826182037553729,
      "grad_norm": 0.24263644716720503,
      "learning_rate": 4.974183764874665e-07,
      "loss": 1.4961,
      "step": 320
    },
    {
      "epoch": 0.048412638564210846,
      "grad_norm": 0.3598724778136554,
      "learning_rate": 4.974022472979165e-07,
      "loss": 1.5871,
      "step": 321
    },
    {
      "epoch": 0.0485634567528844,
      "grad_norm": 0.25739161091804974,
      "learning_rate": 4.973860681725848e-07,
      "loss": 1.6068,
      "step": 322
    },
    {
      "epoch": 0.04871427494155795,
      "grad_norm": 0.258074132220957,
      "learning_rate": 4.973698391151043e-07,
      "loss": 1.5469,
      "step": 323
    },
    {
      "epoch": 0.048865093130231504,
      "grad_norm": 0.2486463490307011,
      "learning_rate": 4.973535601291188e-07,
      "loss": 1.5599,
      "step": 324
    },
    {
      "epoch": 0.04901591131890506,
      "grad_norm": 0.2614816055199508,
      "learning_rate": 4.973372312182834e-07,
      "loss": 1.5778,
      "step": 325
    },
    {
      "epoch": 0.04916672950757862,
      "grad_norm": 0.25989296790690797,
      "learning_rate": 4.973208523862643e-07,
      "loss": 1.6123,
      "step": 326
    },
    {
      "epoch": 0.04931754769625217,
      "grad_norm": 0.25399098043192375,
      "learning_rate": 4.973044236367393e-07,
      "loss": 1.5476,
      "step": 327
    },
    {
      "epoch": 0.04946836588492572,
      "grad_norm": 0.28911197979590025,
      "learning_rate": 4.972879449733968e-07,
      "loss": 1.5964,
      "step": 328
    },
    {
      "epoch": 0.049619184073599276,
      "grad_norm": 0.27658018770420756,
      "learning_rate": 4.97271416399937e-07,
      "loss": 1.5659,
      "step": 329
    },
    {
      "epoch": 0.04977000226227283,
      "grad_norm": 0.2557904118785333,
      "learning_rate": 4.97254837920071e-07,
      "loss": 1.5808,
      "step": 330
    },
    {
      "epoch": 0.04992082045094638,
      "grad_norm": 0.25726258372686667,
      "learning_rate": 4.972382095375211e-07,
      "loss": 1.5627,
      "step": 331
    },
    {
      "epoch": 0.050071638639619935,
      "grad_norm": 0.26655152839938784,
      "learning_rate": 4.972215312560208e-07,
      "loss": 1.5887,
      "step": 332
    },
    {
      "epoch": 0.050222456828293495,
      "grad_norm": 0.23932086734756317,
      "learning_rate": 4.97204803079315e-07,
      "loss": 1.5403,
      "step": 333
    },
    {
      "epoch": 0.05037327501696705,
      "grad_norm": 0.26431994526165736,
      "learning_rate": 4.971880250111596e-07,
      "loss": 1.5744,
      "step": 334
    },
    {
      "epoch": 0.0505240932056406,
      "grad_norm": 1.5201014117100498,
      "learning_rate": 4.971711970553216e-07,
      "loss": 1.6221,
      "step": 335
    },
    {
      "epoch": 0.050674911394314154,
      "grad_norm": 0.26096064703562155,
      "learning_rate": 4.971543192155797e-07,
      "loss": 1.5479,
      "step": 336
    },
    {
      "epoch": 0.05082572958298771,
      "grad_norm": 0.25940489482173235,
      "learning_rate": 4.971373914957233e-07,
      "loss": 1.5641,
      "step": 337
    },
    {
      "epoch": 0.05097654777166126,
      "grad_norm": 0.30473473344604063,
      "learning_rate": 4.97120413899553e-07,
      "loss": 1.6051,
      "step": 338
    },
    {
      "epoch": 0.05112736596033482,
      "grad_norm": 0.24196960703987094,
      "learning_rate": 4.97103386430881e-07,
      "loss": 1.5729,
      "step": 339
    },
    {
      "epoch": 0.05127818414900837,
      "grad_norm": 0.24610108728619232,
      "learning_rate": 4.970863090935303e-07,
      "loss": 1.5358,
      "step": 340
    },
    {
      "epoch": 0.051429002337681926,
      "grad_norm": 0.2676476083686511,
      "learning_rate": 4.970691818913354e-07,
      "loss": 1.5793,
      "step": 341
    },
    {
      "epoch": 0.05157982052635548,
      "grad_norm": 0.27860897150629826,
      "learning_rate": 4.970520048281418e-07,
      "loss": 1.5466,
      "step": 342
    },
    {
      "epoch": 0.05173063871502903,
      "grad_norm": 0.2623220090940916,
      "learning_rate": 4.970347779078063e-07,
      "loss": 1.6198,
      "step": 343
    },
    {
      "epoch": 0.051881456903702584,
      "grad_norm": 0.25308668267137435,
      "learning_rate": 4.970175011341968e-07,
      "loss": 1.5745,
      "step": 344
    },
    {
      "epoch": 0.05203227509237614,
      "grad_norm": 0.24984219060622315,
      "learning_rate": 4.970001745111924e-07,
      "loss": 1.6151,
      "step": 345
    },
    {
      "epoch": 0.0521830932810497,
      "grad_norm": 0.24582469750896027,
      "learning_rate": 4.969827980426834e-07,
      "loss": 1.5943,
      "step": 346
    },
    {
      "epoch": 0.05233391146972325,
      "grad_norm": 0.7596166627025406,
      "learning_rate": 4.969653717325715e-07,
      "loss": 1.5662,
      "step": 347
    },
    {
      "epoch": 0.0524847296583968,
      "grad_norm": 0.2684913803684381,
      "learning_rate": 4.969478955847692e-07,
      "loss": 1.6064,
      "step": 348
    },
    {
      "epoch": 0.052635547847070356,
      "grad_norm": 0.26403543896139,
      "learning_rate": 4.969303696032005e-07,
      "loss": 1.5711,
      "step": 349
    },
    {
      "epoch": 0.05278636603574391,
      "grad_norm": 0.27319367968006386,
      "learning_rate": 4.969127937918006e-07,
      "loss": 1.6161,
      "step": 350
    },
    {
      "epoch": 0.05293718422441746,
      "grad_norm": 0.25929594252221,
      "learning_rate": 4.968951681545155e-07,
      "loss": 1.615,
      "step": 351
    },
    {
      "epoch": 0.05308800241309102,
      "grad_norm": 0.2566722943610633,
      "learning_rate": 4.96877492695303e-07,
      "loss": 1.5854,
      "step": 352
    },
    {
      "epoch": 0.053238820601764575,
      "grad_norm": 0.24811229538006918,
      "learning_rate": 4.968597674181316e-07,
      "loss": 1.6046,
      "step": 353
    },
    {
      "epoch": 0.05338963879043813,
      "grad_norm": 0.2856100330991902,
      "learning_rate": 4.96841992326981e-07,
      "loss": 1.6457,
      "step": 354
    },
    {
      "epoch": 0.05354045697911168,
      "grad_norm": 0.2748920596999085,
      "learning_rate": 4.968241674258426e-07,
      "loss": 1.5703,
      "step": 355
    },
    {
      "epoch": 0.053691275167785234,
      "grad_norm": 0.2622971771021598,
      "learning_rate": 4.968062927187183e-07,
      "loss": 1.5573,
      "step": 356
    },
    {
      "epoch": 0.05384209335645879,
      "grad_norm": 0.2560163824303438,
      "learning_rate": 4.967883682096214e-07,
      "loss": 1.5971,
      "step": 357
    },
    {
      "epoch": 0.05399291154513234,
      "grad_norm": 0.27367198030490836,
      "learning_rate": 4.967703939025768e-07,
      "loss": 1.5848,
      "step": 358
    },
    {
      "epoch": 0.0541437297338059,
      "grad_norm": 0.2664287206155345,
      "learning_rate": 4.967523698016201e-07,
      "loss": 1.5142,
      "step": 359
    },
    {
      "epoch": 0.05429454792247945,
      "grad_norm": 0.2812645453883054,
      "learning_rate": 4.967342959107982e-07,
      "loss": 1.6473,
      "step": 360
    },
    {
      "epoch": 0.054445366111153005,
      "grad_norm": 0.2639456931300287,
      "learning_rate": 4.967161722341692e-07,
      "loss": 1.6125,
      "step": 361
    },
    {
      "epoch": 0.05459618429982656,
      "grad_norm": 0.25849725873657714,
      "learning_rate": 4.966979987758025e-07,
      "loss": 1.5649,
      "step": 362
    },
    {
      "epoch": 0.05474700248850011,
      "grad_norm": 0.2391469770882484,
      "learning_rate": 4.966797755397785e-07,
      "loss": 1.5466,
      "step": 363
    },
    {
      "epoch": 0.054897820677173664,
      "grad_norm": 0.24853257602934095,
      "learning_rate": 4.96661502530189e-07,
      "loss": 1.5335,
      "step": 364
    },
    {
      "epoch": 0.055048638865847224,
      "grad_norm": 0.27168670853914295,
      "learning_rate": 4.966431797511366e-07,
      "loss": 1.5268,
      "step": 365
    },
    {
      "epoch": 0.05519945705452078,
      "grad_norm": 0.265670244111924,
      "learning_rate": 4.966248072067353e-07,
      "loss": 1.6445,
      "step": 366
    },
    {
      "epoch": 0.05535027524319433,
      "grad_norm": 0.2608115163988439,
      "learning_rate": 4.966063849011103e-07,
      "loss": 1.5765,
      "step": 367
    },
    {
      "epoch": 0.05550109343186788,
      "grad_norm": 0.2701215726834107,
      "learning_rate": 4.965879128383982e-07,
      "loss": 1.5365,
      "step": 368
    },
    {
      "epoch": 0.055651911620541436,
      "grad_norm": 0.25470002940222275,
      "learning_rate": 4.965693910227462e-07,
      "loss": 1.579,
      "step": 369
    },
    {
      "epoch": 0.05580272980921499,
      "grad_norm": 0.2548711844491008,
      "learning_rate": 4.965508194583132e-07,
      "loss": 1.6139,
      "step": 370
    },
    {
      "epoch": 0.05595354799788854,
      "grad_norm": 0.299974020575167,
      "learning_rate": 4.965321981492687e-07,
      "loss": 1.549,
      "step": 371
    },
    {
      "epoch": 0.0561043661865621,
      "grad_norm": 0.278607732945803,
      "learning_rate": 4.965135270997942e-07,
      "loss": 1.5023,
      "step": 372
    },
    {
      "epoch": 0.056255184375235655,
      "grad_norm": 0.25504464963999574,
      "learning_rate": 4.964948063140817e-07,
      "loss": 1.6278,
      "step": 373
    },
    {
      "epoch": 0.05640600256390921,
      "grad_norm": 0.26546345346554,
      "learning_rate": 4.964760357963345e-07,
      "loss": 1.5378,
      "step": 374
    },
    {
      "epoch": 0.05655682075258276,
      "grad_norm": 0.27984646045688566,
      "learning_rate": 4.964572155507671e-07,
      "loss": 1.6742,
      "step": 375
    },
    {
      "epoch": 0.056707638941256314,
      "grad_norm": 0.23650139431217956,
      "learning_rate": 4.964383455816053e-07,
      "loss": 1.6343,
      "step": 376
    },
    {
      "epoch": 0.05685845712992987,
      "grad_norm": 0.278604940494406,
      "learning_rate": 4.96419425893086e-07,
      "loss": 1.6187,
      "step": 377
    },
    {
      "epoch": 0.057009275318603427,
      "grad_norm": 0.256361842335032,
      "learning_rate": 4.96400456489457e-07,
      "loss": 1.5913,
      "step": 378
    },
    {
      "epoch": 0.05716009350727698,
      "grad_norm": 0.2545279810570703,
      "learning_rate": 4.963814373749777e-07,
      "loss": 1.6059,
      "step": 379
    },
    {
      "epoch": 0.05731091169595053,
      "grad_norm": 0.27221070074121395,
      "learning_rate": 4.963623685539184e-07,
      "loss": 1.59,
      "step": 380
    },
    {
      "epoch": 0.057461729884624085,
      "grad_norm": 0.26467844690156933,
      "learning_rate": 4.963432500305605e-07,
      "loss": 1.5966,
      "step": 381
    },
    {
      "epoch": 0.05761254807329764,
      "grad_norm": 0.24614201302924177,
      "learning_rate": 4.963240818091969e-07,
      "loss": 1.537,
      "step": 382
    },
    {
      "epoch": 0.05776336626197119,
      "grad_norm": 0.24763193545120962,
      "learning_rate": 4.96304863894131e-07,
      "loss": 1.5637,
      "step": 383
    },
    {
      "epoch": 0.05791418445064475,
      "grad_norm": 0.2666288900937848,
      "learning_rate": 4.962855962896782e-07,
      "loss": 1.6786,
      "step": 384
    },
    {
      "epoch": 0.058065002639318304,
      "grad_norm": 0.2705435333853324,
      "learning_rate": 4.962662790001645e-07,
      "loss": 1.5665,
      "step": 385
    },
    {
      "epoch": 0.05821582082799186,
      "grad_norm": 0.2607417613515552,
      "learning_rate": 4.962469120299272e-07,
      "loss": 1.6145,
      "step": 386
    },
    {
      "epoch": 0.05836663901666541,
      "grad_norm": 0.29842159732772466,
      "learning_rate": 4.962274953833147e-07,
      "loss": 1.5656,
      "step": 387
    },
    {
      "epoch": 0.05851745720533896,
      "grad_norm": 0.25621058334891145,
      "learning_rate": 4.962080290646865e-07,
      "loss": 1.5899,
      "step": 388
    },
    {
      "epoch": 0.058668275394012516,
      "grad_norm": 0.30606055120493514,
      "learning_rate": 4.961885130784137e-07,
      "loss": 1.5625,
      "step": 389
    },
    {
      "epoch": 0.05881909358268607,
      "grad_norm": 0.2524525980936105,
      "learning_rate": 4.961689474288779e-07,
      "loss": 1.6383,
      "step": 390
    },
    {
      "epoch": 0.05896991177135963,
      "grad_norm": 0.37840292284840893,
      "learning_rate": 4.961493321204722e-07,
      "loss": 1.5864,
      "step": 391
    },
    {
      "epoch": 0.05912072996003318,
      "grad_norm": 0.26299467187040465,
      "learning_rate": 4.961296671576009e-07,
      "loss": 1.6757,
      "step": 392
    },
    {
      "epoch": 0.059271548148706735,
      "grad_norm": 0.25575918441640594,
      "learning_rate": 4.961099525446793e-07,
      "loss": 1.6462,
      "step": 393
    },
    {
      "epoch": 0.05942236633738029,
      "grad_norm": 0.2536309275592947,
      "learning_rate": 4.96090188286134e-07,
      "loss": 1.4705,
      "step": 394
    },
    {
      "epoch": 0.05957318452605384,
      "grad_norm": 0.2467647941221563,
      "learning_rate": 4.960703743864025e-07,
      "loss": 1.5756,
      "step": 395
    },
    {
      "epoch": 0.059724002714727394,
      "grad_norm": 0.2542755135779441,
      "learning_rate": 4.960505108499337e-07,
      "loss": 1.5743,
      "step": 396
    },
    {
      "epoch": 0.059874820903400953,
      "grad_norm": 0.2454253182745587,
      "learning_rate": 4.960305976811874e-07,
      "loss": 1.6408,
      "step": 397
    },
    {
      "epoch": 0.060025639092074506,
      "grad_norm": 0.2878230831825292,
      "learning_rate": 4.960106348846349e-07,
      "loss": 1.5901,
      "step": 398
    },
    {
      "epoch": 0.06017645728074806,
      "grad_norm": 0.2396955029775844,
      "learning_rate": 4.959906224647584e-07,
      "loss": 1.6232,
      "step": 399
    },
    {
      "epoch": 0.06032727546942161,
      "grad_norm": 0.254504424632013,
      "learning_rate": 4.95970560426051e-07,
      "loss": 1.6626,
      "step": 400
    },
    {
      "epoch": 0.060478093658095165,
      "grad_norm": 0.24242887605105484,
      "learning_rate": 4.959504487730175e-07,
      "loss": 1.5979,
      "step": 401
    },
    {
      "epoch": 0.06062891184676872,
      "grad_norm": 0.2991997387118549,
      "learning_rate": 4.959302875101736e-07,
      "loss": 1.5772,
      "step": 402
    },
    {
      "epoch": 0.06077973003544227,
      "grad_norm": 0.25635844722182943,
      "learning_rate": 4.959100766420458e-07,
      "loss": 1.5976,
      "step": 403
    },
    {
      "epoch": 0.06093054822411583,
      "grad_norm": 0.25378689447575264,
      "learning_rate": 4.958898161731722e-07,
      "loss": 1.6138,
      "step": 404
    },
    {
      "epoch": 0.061081366412789384,
      "grad_norm": 0.2623760053921337,
      "learning_rate": 4.958695061081017e-07,
      "loss": 1.58,
      "step": 405
    },
    {
      "epoch": 0.06123218460146294,
      "grad_norm": 0.2454756265927881,
      "learning_rate": 4.958491464513949e-07,
      "loss": 1.5954,
      "step": 406
    },
    {
      "epoch": 0.06138300279013649,
      "grad_norm": 0.25948334826498115,
      "learning_rate": 4.958287372076228e-07,
      "loss": 1.663,
      "step": 407
    },
    {
      "epoch": 0.06153382097881004,
      "grad_norm": 0.25222923306553524,
      "learning_rate": 4.95808278381368e-07,
      "loss": 1.6045,
      "step": 408
    },
    {
      "epoch": 0.061684639167483596,
      "grad_norm": 0.25613180784161454,
      "learning_rate": 4.95787769977224e-07,
      "loss": 1.5402,
      "step": 409
    },
    {
      "epoch": 0.061835457356157156,
      "grad_norm": 0.2421104210544763,
      "learning_rate": 4.957672119997956e-07,
      "loss": 1.5712,
      "step": 410
    },
    {
      "epoch": 0.06198627554483071,
      "grad_norm": 0.29376753209022727,
      "learning_rate": 4.957466044536987e-07,
      "loss": 1.6257,
      "step": 411
    },
    {
      "epoch": 0.06213709373350426,
      "grad_norm": 0.24950584536301929,
      "learning_rate": 4.957259473435601e-07,
      "loss": 1.5806,
      "step": 412
    },
    {
      "epoch": 0.062287911922177815,
      "grad_norm": 0.29461741859092144,
      "learning_rate": 4.957052406740182e-07,
      "loss": 1.5995,
      "step": 413
    },
    {
      "epoch": 0.06243873011085137,
      "grad_norm": 0.2575517622877305,
      "learning_rate": 4.956844844497221e-07,
      "loss": 1.5807,
      "step": 414
    },
    {
      "epoch": 0.06258954829952493,
      "grad_norm": 0.2657087176303521,
      "learning_rate": 4.956636786753322e-07,
      "loss": 1.6092,
      "step": 415
    },
    {
      "epoch": 0.06274036648819847,
      "grad_norm": 0.25543517571389057,
      "learning_rate": 4.9564282335552e-07,
      "loss": 1.6333,
      "step": 416
    },
    {
      "epoch": 0.06289118467687203,
      "grad_norm": 0.2825468297311831,
      "learning_rate": 4.95621918494968e-07,
      "loss": 1.5348,
      "step": 417
    },
    {
      "epoch": 0.06304200286554558,
      "grad_norm": 0.44887284182032045,
      "learning_rate": 4.956009640983702e-07,
      "loss": 1.5886,
      "step": 418
    },
    {
      "epoch": 0.06319282105421914,
      "grad_norm": 0.2663367815663453,
      "learning_rate": 4.955799601704315e-07,
      "loss": 1.6115,
      "step": 419
    },
    {
      "epoch": 0.0633436392428927,
      "grad_norm": 0.5364400057464868,
      "learning_rate": 4.955589067158675e-07,
      "loss": 1.558,
      "step": 420
    },
    {
      "epoch": 0.06349445743156625,
      "grad_norm": 0.28752305934187716,
      "learning_rate": 4.955378037394056e-07,
      "loss": 1.5967,
      "step": 421
    },
    {
      "epoch": 0.0636452756202398,
      "grad_norm": 0.25938858933203635,
      "learning_rate": 4.955166512457841e-07,
      "loss": 1.5468,
      "step": 422
    },
    {
      "epoch": 0.06379609380891335,
      "grad_norm": 0.309542792113621,
      "learning_rate": 4.954954492397521e-07,
      "loss": 1.5073,
      "step": 423
    },
    {
      "epoch": 0.06394691199758691,
      "grad_norm": 0.2576084447428103,
      "learning_rate": 4.954741977260703e-07,
      "loss": 1.5529,
      "step": 424
    },
    {
      "epoch": 0.06409773018626046,
      "grad_norm": 0.2529029416333881,
      "learning_rate": 4.954528967095101e-07,
      "loss": 1.5312,
      "step": 425
    },
    {
      "epoch": 0.06424854837493402,
      "grad_norm": 0.25953587371630393,
      "learning_rate": 4.954315461948545e-07,
      "loss": 1.6635,
      "step": 426
    },
    {
      "epoch": 0.06439936656360758,
      "grad_norm": 0.26141101300702185,
      "learning_rate": 4.95410146186897e-07,
      "loss": 1.6324,
      "step": 427
    },
    {
      "epoch": 0.06455018475228112,
      "grad_norm": 0.2703321534764104,
      "learning_rate": 4.953886966904425e-07,
      "loss": 1.5481,
      "step": 428
    },
    {
      "epoch": 0.06470100294095468,
      "grad_norm": 0.2584943248632401,
      "learning_rate": 4.953671977103073e-07,
      "loss": 1.6144,
      "step": 429
    },
    {
      "epoch": 0.06485182112962823,
      "grad_norm": 1.222515281227258,
      "learning_rate": 4.953456492513184e-07,
      "loss": 1.601,
      "step": 430
    },
    {
      "epoch": 0.06500263931830179,
      "grad_norm": 0.2658086216081372,
      "learning_rate": 4.95324051318314e-07,
      "loss": 1.6449,
      "step": 431
    },
    {
      "epoch": 0.06515345750697533,
      "grad_norm": 0.38610334921098405,
      "learning_rate": 4.953024039161436e-07,
      "loss": 1.6104,
      "step": 432
    },
    {
      "epoch": 0.0653042756956489,
      "grad_norm": 0.2825054807195639,
      "learning_rate": 4.952807070496675e-07,
      "loss": 1.7244,
      "step": 433
    },
    {
      "epoch": 0.06545509388432245,
      "grad_norm": 0.30899364121719064,
      "learning_rate": 4.952589607237575e-07,
      "loss": 1.496,
      "step": 434
    },
    {
      "epoch": 0.065605912072996,
      "grad_norm": 0.2723169206900093,
      "learning_rate": 4.952371649432962e-07,
      "loss": 1.5453,
      "step": 435
    },
    {
      "epoch": 0.06575673026166956,
      "grad_norm": 7.523062534338598,
      "learning_rate": 4.952153197131773e-07,
      "loss": 1.5674,
      "step": 436
    },
    {
      "epoch": 0.0659075484503431,
      "grad_norm": 0.48768300544830984,
      "learning_rate": 4.951934250383057e-07,
      "loss": 1.5901,
      "step": 437
    },
    {
      "epoch": 0.06605836663901667,
      "grad_norm": 0.2599806063338143,
      "learning_rate": 4.951714809235974e-07,
      "loss": 1.5693,
      "step": 438
    },
    {
      "epoch": 0.06620918482769023,
      "grad_norm": 0.380773588575501,
      "learning_rate": 4.951494873739796e-07,
      "loss": 1.5482,
      "step": 439
    },
    {
      "epoch": 0.06636000301636377,
      "grad_norm": 0.24912068069172005,
      "learning_rate": 4.951274443943905e-07,
      "loss": 1.5877,
      "step": 440
    },
    {
      "epoch": 0.06651082120503733,
      "grad_norm": 2.064434986001262,
      "learning_rate": 4.951053519897792e-07,
      "loss": 1.6384,
      "step": 441
    },
    {
      "epoch": 0.06666163939371088,
      "grad_norm": 0.2536004094108816,
      "learning_rate": 4.950832101651062e-07,
      "loss": 1.5778,
      "step": 442
    },
    {
      "epoch": 0.06681245758238444,
      "grad_norm": 0.24866657130017011,
      "learning_rate": 4.950610189253431e-07,
      "loss": 1.5818,
      "step": 443
    },
    {
      "epoch": 0.06696327577105798,
      "grad_norm": 0.2621890595250137,
      "learning_rate": 4.950387782754722e-07,
      "loss": 1.6346,
      "step": 444
    },
    {
      "epoch": 0.06711409395973154,
      "grad_norm": 0.2510737223460099,
      "learning_rate": 4.950164882204875e-07,
      "loss": 1.5448,
      "step": 445
    },
    {
      "epoch": 0.0672649121484051,
      "grad_norm": 0.2476520678315182,
      "learning_rate": 4.949941487653936e-07,
      "loss": 1.6327,
      "step": 446
    },
    {
      "epoch": 0.06741573033707865,
      "grad_norm": 0.26233079006484494,
      "learning_rate": 4.949717599152063e-07,
      "loss": 1.6294,
      "step": 447
    },
    {
      "epoch": 0.06756654852575221,
      "grad_norm": 0.24716726351344304,
      "learning_rate": 4.949493216749525e-07,
      "loss": 1.5614,
      "step": 448
    },
    {
      "epoch": 0.06771736671442576,
      "grad_norm": 0.26748648356688093,
      "learning_rate": 4.949268340496706e-07,
      "loss": 1.5706,
      "step": 449
    },
    {
      "epoch": 0.06786818490309932,
      "grad_norm": 0.251706135191505,
      "learning_rate": 4.949042970444092e-07,
      "loss": 1.6084,
      "step": 450
    },
    {
      "epoch": 0.06801900309177286,
      "grad_norm": 0.25831712897398074,
      "learning_rate": 4.94881710664229e-07,
      "loss": 1.5103,
      "step": 451
    },
    {
      "epoch": 0.06816982128044642,
      "grad_norm": 0.2518111435993509,
      "learning_rate": 4.948590749142011e-07,
      "loss": 1.5769,
      "step": 452
    },
    {
      "epoch": 0.06832063946911998,
      "grad_norm": 0.25077512227541265,
      "learning_rate": 4.948363897994078e-07,
      "loss": 1.5992,
      "step": 453
    },
    {
      "epoch": 0.06847145765779353,
      "grad_norm": 0.25227964846590806,
      "learning_rate": 4.948136553249426e-07,
      "loss": 1.574,
      "step": 454
    },
    {
      "epoch": 0.06862227584646709,
      "grad_norm": 0.2457294422180872,
      "learning_rate": 4.947908714959102e-07,
      "loss": 1.5769,
      "step": 455
    },
    {
      "epoch": 0.06877309403514063,
      "grad_norm": 0.2496239078167245,
      "learning_rate": 4.94768038317426e-07,
      "loss": 1.5436,
      "step": 456
    },
    {
      "epoch": 0.0689239122238142,
      "grad_norm": 0.29062933294218735,
      "learning_rate": 4.94745155794617e-07,
      "loss": 1.5727,
      "step": 457
    },
    {
      "epoch": 0.06907473041248774,
      "grad_norm": 0.25627733322126045,
      "learning_rate": 4.947222239326207e-07,
      "loss": 1.61,
      "step": 458
    },
    {
      "epoch": 0.0692255486011613,
      "grad_norm": 0.2532772075106132,
      "learning_rate": 4.946992427365862e-07,
      "loss": 1.6688,
      "step": 459
    },
    {
      "epoch": 0.06937636678983486,
      "grad_norm": 0.2584382481261571,
      "learning_rate": 4.946762122116733e-07,
      "loss": 1.6578,
      "step": 460
    },
    {
      "epoch": 0.0695271849785084,
      "grad_norm": 0.26526267071855675,
      "learning_rate": 4.946531323630531e-07,
      "loss": 1.6717,
      "step": 461
    },
    {
      "epoch": 0.06967800316718196,
      "grad_norm": 0.25305928479834555,
      "learning_rate": 4.946300031959077e-07,
      "loss": 1.6204,
      "step": 462
    },
    {
      "epoch": 0.06982882135585551,
      "grad_norm": 0.28864451322495427,
      "learning_rate": 4.946068247154303e-07,
      "loss": 1.5951,
      "step": 463
    },
    {
      "epoch": 0.06997963954452907,
      "grad_norm": 0.24532974198684246,
      "learning_rate": 4.94583596926825e-07,
      "loss": 1.5852,
      "step": 464
    },
    {
      "epoch": 0.07013045773320263,
      "grad_norm": 3.851715573924243,
      "learning_rate": 4.945603198353073e-07,
      "loss": 1.5603,
      "step": 465
    },
    {
      "epoch": 0.07028127592187618,
      "grad_norm": 0.27613690978194105,
      "learning_rate": 4.945369934461034e-07,
      "loss": 1.6249,
      "step": 466
    },
    {
      "epoch": 0.07043209411054974,
      "grad_norm": 0.2514935014317457,
      "learning_rate": 4.945136177644509e-07,
      "loss": 1.5245,
      "step": 467
    },
    {
      "epoch": 0.07058291229922328,
      "grad_norm": 0.8068622232940601,
      "learning_rate": 4.944901927955982e-07,
      "loss": 1.6069,
      "step": 468
    },
    {
      "epoch": 0.07073373048789684,
      "grad_norm": 0.24230146231946875,
      "learning_rate": 4.94466718544805e-07,
      "loss": 1.565,
      "step": 469
    },
    {
      "epoch": 0.07088454867657039,
      "grad_norm": 0.25797039870888167,
      "learning_rate": 4.944431950173419e-07,
      "loss": 1.5914,
      "step": 470
    },
    {
      "epoch": 0.07103536686524395,
      "grad_norm": 0.28046800099114216,
      "learning_rate": 4.944196222184907e-07,
      "loss": 1.5747,
      "step": 471
    },
    {
      "epoch": 0.07118618505391751,
      "grad_norm": 0.26447198542846684,
      "learning_rate": 4.94396000153544e-07,
      "loss": 1.607,
      "step": 472
    },
    {
      "epoch": 0.07133700324259105,
      "grad_norm": 0.25729676049336647,
      "learning_rate": 4.943723288278059e-07,
      "loss": 1.5715,
      "step": 473
    },
    {
      "epoch": 0.07148782143126461,
      "grad_norm": 0.25074184707404273,
      "learning_rate": 4.94348608246591e-07,
      "loss": 1.6102,
      "step": 474
    },
    {
      "epoch": 0.07163863961993816,
      "grad_norm": 0.28631414643129693,
      "learning_rate": 4.943248384152255e-07,
      "loss": 1.5604,
      "step": 475
    },
    {
      "epoch": 0.07178945780861172,
      "grad_norm": 0.2491117852003348,
      "learning_rate": 4.943010193390463e-07,
      "loss": 1.5763,
      "step": 476
    },
    {
      "epoch": 0.07194027599728527,
      "grad_norm": 0.6450301432730091,
      "learning_rate": 4.942771510234015e-07,
      "loss": 1.6482,
      "step": 477
    },
    {
      "epoch": 0.07209109418595883,
      "grad_norm": 0.25779434871207685,
      "learning_rate": 4.942532334736502e-07,
      "loss": 1.6369,
      "step": 478
    },
    {
      "epoch": 0.07224191237463239,
      "grad_norm": 0.255749488451868,
      "learning_rate": 4.942292666951626e-07,
      "loss": 1.5625,
      "step": 479
    },
    {
      "epoch": 0.07239273056330593,
      "grad_norm": 0.24329694843771454,
      "learning_rate": 4.942052506933201e-07,
      "loss": 1.545,
      "step": 480
    },
    {
      "epoch": 0.07254354875197949,
      "grad_norm": 0.2639578426050398,
      "learning_rate": 4.941811854735148e-07,
      "loss": 1.5715,
      "step": 481
    },
    {
      "epoch": 0.07269436694065304,
      "grad_norm": 0.2555493641454379,
      "learning_rate": 4.9415707104115e-07,
      "loss": 1.5908,
      "step": 482
    },
    {
      "epoch": 0.0728451851293266,
      "grad_norm": 0.24771916790007276,
      "learning_rate": 4.941329074016404e-07,
      "loss": 1.6032,
      "step": 483
    },
    {
      "epoch": 0.07299600331800016,
      "grad_norm": 0.25813027877620537,
      "learning_rate": 4.94108694560411e-07,
      "loss": 1.6065,
      "step": 484
    },
    {
      "epoch": 0.0731468215066737,
      "grad_norm": 0.2617232017934573,
      "learning_rate": 4.940844325228986e-07,
      "loss": 1.5383,
      "step": 485
    },
    {
      "epoch": 0.07329763969534726,
      "grad_norm": 0.9969708511757501,
      "learning_rate": 4.940601212945508e-07,
      "loss": 1.5394,
      "step": 486
    },
    {
      "epoch": 0.07344845788402081,
      "grad_norm": 0.26599817022694655,
      "learning_rate": 4.940357608808258e-07,
      "loss": 1.56,
      "step": 487
    },
    {
      "epoch": 0.07359927607269437,
      "grad_norm": 0.2882472701105029,
      "learning_rate": 4.940113512871937e-07,
      "loss": 1.5607,
      "step": 488
    },
    {
      "epoch": 0.07375009426136792,
      "grad_norm": 0.26344165366825933,
      "learning_rate": 4.939868925191347e-07,
      "loss": 1.5867,
      "step": 489
    },
    {
      "epoch": 0.07390091245004148,
      "grad_norm": 0.2559759462159106,
      "learning_rate": 4.939623845821408e-07,
      "loss": 1.661,
      "step": 490
    },
    {
      "epoch": 0.07405173063871504,
      "grad_norm": 0.2494797400109043,
      "learning_rate": 4.939378274817147e-07,
      "loss": 1.5513,
      "step": 491
    },
    {
      "epoch": 0.07420254882738858,
      "grad_norm": 0.2564563520992107,
      "learning_rate": 4.939132212233701e-07,
      "loss": 1.6087,
      "step": 492
    },
    {
      "epoch": 0.07435336701606214,
      "grad_norm": 0.2501704535942094,
      "learning_rate": 4.938885658126319e-07,
      "loss": 1.5295,
      "step": 493
    },
    {
      "epoch": 0.07450418520473569,
      "grad_norm": 0.2707003912399451,
      "learning_rate": 4.93863861255036e-07,
      "loss": 1.6212,
      "step": 494
    },
    {
      "epoch": 0.07465500339340925,
      "grad_norm": 0.2610328885695742,
      "learning_rate": 4.938391075561292e-07,
      "loss": 1.6279,
      "step": 495
    },
    {
      "epoch": 0.0748058215820828,
      "grad_norm": 0.24482384823431727,
      "learning_rate": 4.938143047214695e-07,
      "loss": 1.6057,
      "step": 496
    },
    {
      "epoch": 0.07495663977075635,
      "grad_norm": 0.2662320490545685,
      "learning_rate": 4.937894527566257e-07,
      "loss": 1.5947,
      "step": 497
    },
    {
      "epoch": 0.07510745795942991,
      "grad_norm": 0.24651792237577014,
      "learning_rate": 4.93764551667178e-07,
      "loss": 1.6602,
      "step": 498
    },
    {
      "epoch": 0.07525827614810346,
      "grad_norm": 0.24301487876995637,
      "learning_rate": 4.937396014587173e-07,
      "loss": 1.5758,
      "step": 499
    },
    {
      "epoch": 0.07540909433677702,
      "grad_norm": 0.24840742131244414,
      "learning_rate": 4.937146021368456e-07,
      "loss": 1.562,
      "step": 500
    },
    {
      "epoch": 0.07555991252545056,
      "grad_norm": 0.3866236043766646,
      "learning_rate": 4.936895537071761e-07,
      "loss": 1.6029,
      "step": 501
    },
    {
      "epoch": 0.07571073071412412,
      "grad_norm": 0.26485826630020837,
      "learning_rate": 4.936644561753329e-07,
      "loss": 1.5556,
      "step": 502
    },
    {
      "epoch": 0.07586154890279767,
      "grad_norm": 0.2489058113043106,
      "learning_rate": 4.93639309546951e-07,
      "loss": 1.5792,
      "step": 503
    },
    {
      "epoch": 0.07601236709147123,
      "grad_norm": 0.26597721973837385,
      "learning_rate": 4.936141138276766e-07,
      "loss": 1.5991,
      "step": 504
    },
    {
      "epoch": 0.07616318528014479,
      "grad_norm": 0.2607421928543872,
      "learning_rate": 4.93588869023167e-07,
      "loss": 1.5498,
      "step": 505
    },
    {
      "epoch": 0.07631400346881834,
      "grad_norm": 0.2895465689620655,
      "learning_rate": 4.935635751390903e-07,
      "loss": 1.5785,
      "step": 506
    },
    {
      "epoch": 0.0764648216574919,
      "grad_norm": 0.26139600362308274,
      "learning_rate": 4.935382321811255e-07,
      "loss": 1.5805,
      "step": 507
    },
    {
      "epoch": 0.07661563984616544,
      "grad_norm": 0.2638969582987421,
      "learning_rate": 4.935128401549633e-07,
      "loss": 1.625,
      "step": 508
    },
    {
      "epoch": 0.076766458034839,
      "grad_norm": 0.4096579467303753,
      "learning_rate": 4.934873990663046e-07,
      "loss": 1.6052,
      "step": 509
    },
    {
      "epoch": 0.07691727622351256,
      "grad_norm": 0.2651690016150467,
      "learning_rate": 4.934619089208617e-07,
      "loss": 1.5982,
      "step": 510
    },
    {
      "epoch": 0.07706809441218611,
      "grad_norm": 0.2608025596360598,
      "learning_rate": 4.93436369724358e-07,
      "loss": 1.5265,
      "step": 511
    },
    {
      "epoch": 0.07721891260085967,
      "grad_norm": 0.24775914712884162,
      "learning_rate": 4.934107814825277e-07,
      "loss": 1.6348,
      "step": 512
    },
    {
      "epoch": 0.07736973078953321,
      "grad_norm": 0.23982029055690735,
      "learning_rate": 4.93385144201116e-07,
      "loss": 1.5263,
      "step": 513
    },
    {
      "epoch": 0.07752054897820677,
      "grad_norm": 0.24440445690472806,
      "learning_rate": 4.933594578858796e-07,
      "loss": 1.6087,
      "step": 514
    },
    {
      "epoch": 0.07767136716688032,
      "grad_norm": 0.29016356740639526,
      "learning_rate": 4.933337225425854e-07,
      "loss": 1.582,
      "step": 515
    },
    {
      "epoch": 0.07782218535555388,
      "grad_norm": 4.683371732453072,
      "learning_rate": 4.93307938177012e-07,
      "loss": 1.6012,
      "step": 516
    },
    {
      "epoch": 0.07797300354422744,
      "grad_norm": 0.32311562025321017,
      "learning_rate": 4.932821047949486e-07,
      "loss": 1.6044,
      "step": 517
    },
    {
      "epoch": 0.07812382173290099,
      "grad_norm": 0.2557945762461551,
      "learning_rate": 4.932562224021955e-07,
      "loss": 1.5517,
      "step": 518
    },
    {
      "epoch": 0.07827463992157455,
      "grad_norm": 0.3261593598067262,
      "learning_rate": 4.932302910045642e-07,
      "loss": 1.5681,
      "step": 519
    },
    {
      "epoch": 0.07842545811024809,
      "grad_norm": 0.3704288545771335,
      "learning_rate": 4.932043106078772e-07,
      "loss": 1.5855,
      "step": 520
    },
    {
      "epoch": 0.07857627629892165,
      "grad_norm": 0.27059697971029173,
      "learning_rate": 4.931782812179674e-07,
      "loss": 1.5812,
      "step": 521
    },
    {
      "epoch": 0.0787270944875952,
      "grad_norm": 0.256274062882238,
      "learning_rate": 4.931522028406795e-07,
      "loss": 1.5326,
      "step": 522
    },
    {
      "epoch": 0.07887791267626876,
      "grad_norm": 0.3795916976539053,
      "learning_rate": 4.931260754818688e-07,
      "loss": 1.5828,
      "step": 523
    },
    {
      "epoch": 0.07902873086494232,
      "grad_norm": 0.2872991368841093,
      "learning_rate": 4.930998991474017e-07,
      "loss": 1.5398,
      "step": 524
    },
    {
      "epoch": 0.07917954905361586,
      "grad_norm": 0.24318254003701104,
      "learning_rate": 4.930736738431553e-07,
      "loss": 1.5622,
      "step": 525
    },
    {
      "epoch": 0.07933036724228942,
      "grad_norm": 0.25638241361288794,
      "learning_rate": 4.930473995750181e-07,
      "loss": 1.5693,
      "step": 526
    },
    {
      "epoch": 0.07948118543096297,
      "grad_norm": 0.25050071504745147,
      "learning_rate": 4.930210763488896e-07,
      "loss": 1.5284,
      "step": 527
    },
    {
      "epoch": 0.07963200361963653,
      "grad_norm": 0.24898287643190337,
      "learning_rate": 4.929947041706799e-07,
      "loss": 1.5946,
      "step": 528
    },
    {
      "epoch": 0.07978282180831008,
      "grad_norm": 0.2506867625821907,
      "learning_rate": 4.929682830463105e-07,
      "loss": 1.5109,
      "step": 529
    },
    {
      "epoch": 0.07993363999698364,
      "grad_norm": 0.24183988226515585,
      "learning_rate": 4.929418129817135e-07,
      "loss": 1.5645,
      "step": 530
    },
    {
      "epoch": 0.0800844581856572,
      "grad_norm": 0.2543083634039026,
      "learning_rate": 4.929152939828324e-07,
      "loss": 1.5981,
      "step": 531
    },
    {
      "epoch": 0.08023527637433074,
      "grad_norm": 0.2590778218402505,
      "learning_rate": 4.928887260556214e-07,
      "loss": 1.6484,
      "step": 532
    },
    {
      "epoch": 0.0803860945630043,
      "grad_norm": 0.254819692911704,
      "learning_rate": 4.928621092060457e-07,
      "loss": 1.56,
      "step": 533
    },
    {
      "epoch": 0.08053691275167785,
      "grad_norm": 0.2452102835738855,
      "learning_rate": 4.928354434400817e-07,
      "loss": 1.5742,
      "step": 534
    },
    {
      "epoch": 0.08068773094035141,
      "grad_norm": 0.2525657819458198,
      "learning_rate": 4.928087287637166e-07,
      "loss": 1.5722,
      "step": 535
    },
    {
      "epoch": 0.08083854912902497,
      "grad_norm": 0.2509074090624925,
      "learning_rate": 4.927819651829486e-07,
      "loss": 1.5607,
      "step": 536
    },
    {
      "epoch": 0.08098936731769851,
      "grad_norm": 0.25115918674482207,
      "learning_rate": 4.927551527037868e-07,
      "loss": 1.5402,
      "step": 537
    },
    {
      "epoch": 0.08114018550637207,
      "grad_norm": 0.24342202294545484,
      "learning_rate": 4.927282913322516e-07,
      "loss": 1.6124,
      "step": 538
    },
    {
      "epoch": 0.08129100369504562,
      "grad_norm": 0.2832028861128905,
      "learning_rate": 4.927013810743739e-07,
      "loss": 1.6304,
      "step": 539
    },
    {
      "epoch": 0.08144182188371918,
      "grad_norm": 0.2508552678273836,
      "learning_rate": 4.926744219361962e-07,
      "loss": 1.5912,
      "step": 540
    },
    {
      "epoch": 0.08159264007239272,
      "grad_norm": 0.2437905171553333,
      "learning_rate": 4.926474139237713e-07,
      "loss": 1.4971,
      "step": 541
    },
    {
      "epoch": 0.08174345826106628,
      "grad_norm": 0.24457479507873567,
      "learning_rate": 4.926203570431633e-07,
      "loss": 1.6038,
      "step": 542
    },
    {
      "epoch": 0.08189427644973984,
      "grad_norm": 0.2513609050923867,
      "learning_rate": 4.925932513004475e-07,
      "loss": 1.5137,
      "step": 543
    },
    {
      "epoch": 0.08204509463841339,
      "grad_norm": 0.2539532399251499,
      "learning_rate": 4.925660967017097e-07,
      "loss": 1.5587,
      "step": 544
    },
    {
      "epoch": 0.08219591282708695,
      "grad_norm": 0.2637711952751605,
      "learning_rate": 4.925388932530469e-07,
      "loss": 1.5986,
      "step": 545
    },
    {
      "epoch": 0.0823467310157605,
      "grad_norm": 0.25790794585338084,
      "learning_rate": 4.925116409605672e-07,
      "loss": 1.5677,
      "step": 546
    },
    {
      "epoch": 0.08249754920443406,
      "grad_norm": 0.3729386590116179,
      "learning_rate": 4.924843398303895e-07,
      "loss": 1.5004,
      "step": 547
    },
    {
      "epoch": 0.0826483673931076,
      "grad_norm": 0.2503856209292245,
      "learning_rate": 4.924569898686435e-07,
      "loss": 1.5944,
      "step": 548
    },
    {
      "epoch": 0.08279918558178116,
      "grad_norm": 0.24346750355850505,
      "learning_rate": 4.924295910814703e-07,
      "loss": 1.5523,
      "step": 549
    },
    {
      "epoch": 0.08295000377045472,
      "grad_norm": 0.24624694608717898,
      "learning_rate": 4.924021434750218e-07,
      "loss": 1.6148,
      "step": 550
    },
    {
      "epoch": 0.08310082195912827,
      "grad_norm": 0.2535930007858961,
      "learning_rate": 4.923746470554604e-07,
      "loss": 1.5205,
      "step": 551
    },
    {
      "epoch": 0.08325164014780183,
      "grad_norm": 0.24827404197581623,
      "learning_rate": 4.923471018289602e-07,
      "loss": 1.5817,
      "step": 552
    },
    {
      "epoch": 0.08340245833647537,
      "grad_norm": 0.2503802600818671,
      "learning_rate": 4.923195078017058e-07,
      "loss": 1.5844,
      "step": 553
    },
    {
      "epoch": 0.08355327652514893,
      "grad_norm": 0.27055106614329255,
      "learning_rate": 4.922918649798929e-07,
      "loss": 1.5673,
      "step": 554
    },
    {
      "epoch": 0.0837040947138225,
      "grad_norm": 0.25205217660052004,
      "learning_rate": 4.922641733697281e-07,
      "loss": 1.5818,
      "step": 555
    },
    {
      "epoch": 0.08385491290249604,
      "grad_norm": 0.25579080228321643,
      "learning_rate": 4.922364329774287e-07,
      "loss": 1.5709,
      "step": 556
    },
    {
      "epoch": 0.0840057310911696,
      "grad_norm": 0.249415396867393,
      "learning_rate": 4.922086438092238e-07,
      "loss": 1.5273,
      "step": 557
    },
    {
      "epoch": 0.08415654927984315,
      "grad_norm": 0.28056615620061776,
      "learning_rate": 4.921808058713524e-07,
      "loss": 1.5693,
      "step": 558
    },
    {
      "epoch": 0.0843073674685167,
      "grad_norm": 0.2844995932486672,
      "learning_rate": 4.92152919170065e-07,
      "loss": 1.5175,
      "step": 559
    },
    {
      "epoch": 0.08445818565719025,
      "grad_norm": 0.2608428394441073,
      "learning_rate": 4.921249837116231e-07,
      "loss": 1.6353,
      "step": 560
    },
    {
      "epoch": 0.08460900384586381,
      "grad_norm": 0.24713048021592293,
      "learning_rate": 4.920969995022992e-07,
      "loss": 1.5715,
      "step": 561
    },
    {
      "epoch": 0.08475982203453737,
      "grad_norm": 0.46568785557221054,
      "learning_rate": 4.92068966548376e-07,
      "loss": 1.612,
      "step": 562
    },
    {
      "epoch": 0.08491064022321092,
      "grad_norm": 0.2434062672346236,
      "learning_rate": 4.920408848561483e-07,
      "loss": 1.5673,
      "step": 563
    },
    {
      "epoch": 0.08506145841188448,
      "grad_norm": 0.28399027202323046,
      "learning_rate": 4.920127544319208e-07,
      "loss": 1.5214,
      "step": 564
    },
    {
      "epoch": 0.08521227660055802,
      "grad_norm": 0.2504036265338775,
      "learning_rate": 4.9198457528201e-07,
      "loss": 1.6365,
      "step": 565
    },
    {
      "epoch": 0.08536309478923158,
      "grad_norm": 0.2559157777595285,
      "learning_rate": 4.919563474127428e-07,
      "loss": 1.5891,
      "step": 566
    },
    {
      "epoch": 0.08551391297790513,
      "grad_norm": 0.29339247140505426,
      "learning_rate": 4.91928070830457e-07,
      "loss": 1.6133,
      "step": 567
    },
    {
      "epoch": 0.08566473116657869,
      "grad_norm": 0.2373443895115084,
      "learning_rate": 4.918997455415017e-07,
      "loss": 1.6091,
      "step": 568
    },
    {
      "epoch": 0.08581554935525225,
      "grad_norm": 0.2551840980360237,
      "learning_rate": 4.918713715522367e-07,
      "loss": 1.5759,
      "step": 569
    },
    {
      "epoch": 0.0859663675439258,
      "grad_norm": 0.2874474579323459,
      "learning_rate": 4.918429488690329e-07,
      "loss": 1.5687,
      "step": 570
    },
    {
      "epoch": 0.08611718573259936,
      "grad_norm": 0.24438925052557178,
      "learning_rate": 4.918144774982718e-07,
      "loss": 1.6112,
      "step": 571
    },
    {
      "epoch": 0.0862680039212729,
      "grad_norm": 0.24410876540004492,
      "learning_rate": 4.917859574463461e-07,
      "loss": 1.5914,
      "step": 572
    },
    {
      "epoch": 0.08641882210994646,
      "grad_norm": 0.27929429191497995,
      "learning_rate": 4.917573887196596e-07,
      "loss": 1.6207,
      "step": 573
    },
    {
      "epoch": 0.08656964029862001,
      "grad_norm": 0.2521870396625732,
      "learning_rate": 4.917287713246266e-07,
      "loss": 1.5708,
      "step": 574
    },
    {
      "epoch": 0.08672045848729357,
      "grad_norm": 0.2689103195119159,
      "learning_rate": 4.917001052676726e-07,
      "loss": 1.6346,
      "step": 575
    },
    {
      "epoch": 0.08687127667596713,
      "grad_norm": 0.24086423482406905,
      "learning_rate": 4.916713905552341e-07,
      "loss": 1.6522,
      "step": 576
    },
    {
      "epoch": 0.08702209486464067,
      "grad_norm": 0.2627511498959766,
      "learning_rate": 4.91642627193758e-07,
      "loss": 1.5834,
      "step": 577
    },
    {
      "epoch": 0.08717291305331423,
      "grad_norm": 0.256738420406437,
      "learning_rate": 4.916138151897029e-07,
      "loss": 1.5748,
      "step": 578
    },
    {
      "epoch": 0.08732373124198778,
      "grad_norm": 0.24545574836575051,
      "learning_rate": 4.915849545495378e-07,
      "loss": 1.5448,
      "step": 579
    },
    {
      "epoch": 0.08747454943066134,
      "grad_norm": 0.31967227244550045,
      "learning_rate": 4.915560452797427e-07,
      "loss": 1.532,
      "step": 580
    },
    {
      "epoch": 0.0876253676193349,
      "grad_norm": 0.2968673763523125,
      "learning_rate": 4.915270873868087e-07,
      "loss": 1.6339,
      "step": 581
    },
    {
      "epoch": 0.08777618580800844,
      "grad_norm": 0.28728768833492335,
      "learning_rate": 4.914980808772375e-07,
      "loss": 1.5853,
      "step": 582
    },
    {
      "epoch": 0.087927003996682,
      "grad_norm": 0.2462113832568665,
      "learning_rate": 4.914690257575422e-07,
      "loss": 1.5725,
      "step": 583
    },
    {
      "epoch": 0.08807782218535555,
      "grad_norm": 0.23829183228187617,
      "learning_rate": 4.914399220342461e-07,
      "loss": 1.5606,
      "step": 584
    },
    {
      "epoch": 0.08822864037402911,
      "grad_norm": 0.3152762690498193,
      "learning_rate": 4.914107697138843e-07,
      "loss": 1.5535,
      "step": 585
    },
    {
      "epoch": 0.08837945856270266,
      "grad_norm": 0.24961819664817447,
      "learning_rate": 4.91381568803002e-07,
      "loss": 1.5568,
      "step": 586
    },
    {
      "epoch": 0.08853027675137622,
      "grad_norm": 0.2676329327281572,
      "learning_rate": 4.913523193081558e-07,
      "loss": 1.6027,
      "step": 587
    },
    {
      "epoch": 0.08868109494004978,
      "grad_norm": 0.2441440934420083,
      "learning_rate": 4.913230212359131e-07,
      "loss": 1.5824,
      "step": 588
    },
    {
      "epoch": 0.08883191312872332,
      "grad_norm": 0.2538836446076873,
      "learning_rate": 4.91293674592852e-07,
      "loss": 1.5529,
      "step": 589
    },
    {
      "epoch": 0.08898273131739688,
      "grad_norm": 0.41769465370401243,
      "learning_rate": 4.912642793855619e-07,
      "loss": 1.5893,
      "step": 590
    },
    {
      "epoch": 0.08913354950607043,
      "grad_norm": 0.278803470964098,
      "learning_rate": 4.912348356206426e-07,
      "loss": 1.5982,
      "step": 591
    },
    {
      "epoch": 0.08928436769474399,
      "grad_norm": 0.5837272165753525,
      "learning_rate": 4.912053433047054e-07,
      "loss": 1.5819,
      "step": 592
    },
    {
      "epoch": 0.08943518588341753,
      "grad_norm": 0.25424687096423615,
      "learning_rate": 4.911758024443721e-07,
      "loss": 1.6297,
      "step": 593
    },
    {
      "epoch": 0.0895860040720911,
      "grad_norm": 0.2601339721261789,
      "learning_rate": 4.911462130462752e-07,
      "loss": 1.5876,
      "step": 594
    },
    {
      "epoch": 0.08973682226076465,
      "grad_norm": 0.3393900866903272,
      "learning_rate": 4.911165751170587e-07,
      "loss": 1.5898,
      "step": 595
    },
    {
      "epoch": 0.0898876404494382,
      "grad_norm": 0.7180948377283978,
      "learning_rate": 4.910868886633771e-07,
      "loss": 1.6494,
      "step": 596
    },
    {
      "epoch": 0.09003845863811176,
      "grad_norm": 0.2653601165598602,
      "learning_rate": 4.910571536918958e-07,
      "loss": 1.6252,
      "step": 597
    },
    {
      "epoch": 0.0901892768267853,
      "grad_norm": 0.24954419790830248,
      "learning_rate": 4.910273702092913e-07,
      "loss": 1.5972,
      "step": 598
    },
    {
      "epoch": 0.09034009501545887,
      "grad_norm": 0.2677338162505918,
      "learning_rate": 4.909975382222508e-07,
      "loss": 1.5955,
      "step": 599
    },
    {
      "epoch": 0.09049091320413241,
      "grad_norm": 0.25788426217463334,
      "learning_rate": 4.909676577374722e-07,
      "loss": 1.6125,
      "step": 600
    },
    {
      "epoch": 0.09064173139280597,
      "grad_norm": 0.25188480941816443,
      "learning_rate": 4.909377287616649e-07,
      "loss": 1.5452,
      "step": 601
    },
    {
      "epoch": 0.09079254958147953,
      "grad_norm": 0.27570661850707645,
      "learning_rate": 4.909077513015488e-07,
      "loss": 1.6192,
      "step": 602
    },
    {
      "epoch": 0.09094336777015308,
      "grad_norm": 0.24156826274527765,
      "learning_rate": 4.908777253638545e-07,
      "loss": 1.554,
      "step": 603
    },
    {
      "epoch": 0.09109418595882664,
      "grad_norm": 0.25224612585820494,
      "learning_rate": 4.908476509553239e-07,
      "loss": 1.5563,
      "step": 604
    },
    {
      "epoch": 0.09124500414750018,
      "grad_norm": 0.306557401649218,
      "learning_rate": 4.908175280827094e-07,
      "loss": 1.6025,
      "step": 605
    },
    {
      "epoch": 0.09139582233617374,
      "grad_norm": 0.23952494340105374,
      "learning_rate": 4.907873567527744e-07,
      "loss": 1.5947,
      "step": 606
    },
    {
      "epoch": 0.0915466405248473,
      "grad_norm": 0.24856125896105166,
      "learning_rate": 4.907571369722936e-07,
      "loss": 1.6013,
      "step": 607
    },
    {
      "epoch": 0.09169745871352085,
      "grad_norm": 0.2508258193786405,
      "learning_rate": 4.907268687480518e-07,
      "loss": 1.5779,
      "step": 608
    },
    {
      "epoch": 0.09184827690219441,
      "grad_norm": 0.25781742553896553,
      "learning_rate": 4.906965520868456e-07,
      "loss": 1.6299,
      "step": 609
    },
    {
      "epoch": 0.09199909509086796,
      "grad_norm": 0.2550011985174638,
      "learning_rate": 4.906661869954815e-07,
      "loss": 1.6228,
      "step": 610
    },
    {
      "epoch": 0.09214991327954151,
      "grad_norm": 0.2470914926519195,
      "learning_rate": 4.906357734807775e-07,
      "loss": 1.6089,
      "step": 611
    },
    {
      "epoch": 0.09230073146821506,
      "grad_norm": 0.2610545421881108,
      "learning_rate": 4.906053115495624e-07,
      "loss": 1.6606,
      "step": 612
    },
    {
      "epoch": 0.09245154965688862,
      "grad_norm": 0.26931972604669374,
      "learning_rate": 4.905748012086756e-07,
      "loss": 1.6334,
      "step": 613
    },
    {
      "epoch": 0.09260236784556218,
      "grad_norm": 0.36010264957923965,
      "learning_rate": 4.905442424649679e-07,
      "loss": 1.5844,
      "step": 614
    },
    {
      "epoch": 0.09275318603423573,
      "grad_norm": 0.2562554805903904,
      "learning_rate": 4.905136353253003e-07,
      "loss": 1.5654,
      "step": 615
    },
    {
      "epoch": 0.09290400422290929,
      "grad_norm": 0.2648496245111277,
      "learning_rate": 4.904829797965452e-07,
      "loss": 1.6259,
      "step": 616
    },
    {
      "epoch": 0.09305482241158283,
      "grad_norm": 0.28004205716450364,
      "learning_rate": 4.904522758855855e-07,
      "loss": 1.6303,
      "step": 617
    },
    {
      "epoch": 0.09320564060025639,
      "grad_norm": 0.2543298564080976,
      "learning_rate": 4.904215235993152e-07,
      "loss": 1.5825,
      "step": 618
    },
    {
      "epoch": 0.09335645878892994,
      "grad_norm": 0.2708365964823325,
      "learning_rate": 4.903907229446392e-07,
      "loss": 1.6123,
      "step": 619
    },
    {
      "epoch": 0.0935072769776035,
      "grad_norm": 0.2434521472967361,
      "learning_rate": 4.90359873928473e-07,
      "loss": 1.6018,
      "step": 620
    },
    {
      "epoch": 0.09365809516627706,
      "grad_norm": 0.33511641774018264,
      "learning_rate": 4.90328976557743e-07,
      "loss": 1.5607,
      "step": 621
    },
    {
      "epoch": 0.0938089133549506,
      "grad_norm": 0.2958147063552225,
      "learning_rate": 4.902980308393868e-07,
      "loss": 1.556,
      "step": 622
    },
    {
      "epoch": 0.09395973154362416,
      "grad_norm": 0.28854360372385784,
      "learning_rate": 4.902670367803526e-07,
      "loss": 1.6016,
      "step": 623
    },
    {
      "epoch": 0.09411054973229771,
      "grad_norm": 0.2508804168130778,
      "learning_rate": 4.902359943875992e-07,
      "loss": 1.6528,
      "step": 624
    },
    {
      "epoch": 0.09426136792097127,
      "grad_norm": 0.24637422112807927,
      "learning_rate": 4.902049036680967e-07,
      "loss": 1.5707,
      "step": 625
    },
    {
      "epoch": 0.09441218610964483,
      "grad_norm": 0.27532911514444836,
      "learning_rate": 4.901737646288259e-07,
      "loss": 1.5714,
      "step": 626
    },
    {
      "epoch": 0.09456300429831838,
      "grad_norm": 0.2510146750886741,
      "learning_rate": 4.901425772767784e-07,
      "loss": 1.5771,
      "step": 627
    },
    {
      "epoch": 0.09471382248699194,
      "grad_norm": 0.3652566414579146,
      "learning_rate": 4.901113416189567e-07,
      "loss": 1.5934,
      "step": 628
    },
    {
      "epoch": 0.09486464067566548,
      "grad_norm": 0.264710597437996,
      "learning_rate": 4.90080057662374e-07,
      "loss": 1.5961,
      "step": 629
    },
    {
      "epoch": 0.09501545886433904,
      "grad_norm": 0.2722196620450524,
      "learning_rate": 4.900487254140546e-07,
      "loss": 1.5784,
      "step": 630
    },
    {
      "epoch": 0.09516627705301259,
      "grad_norm": 0.2494468175683074,
      "learning_rate": 4.900173448810334e-07,
      "loss": 1.5588,
      "step": 631
    },
    {
      "epoch": 0.09531709524168615,
      "grad_norm": 0.25564128850962825,
      "learning_rate": 4.899859160703563e-07,
      "loss": 1.5296,
      "step": 632
    },
    {
      "epoch": 0.09546791343035971,
      "grad_norm": 0.24464482213775124,
      "learning_rate": 4.899544389890798e-07,
      "loss": 1.5731,
      "step": 633
    },
    {
      "epoch": 0.09561873161903325,
      "grad_norm": 0.25286317801092784,
      "learning_rate": 4.899229136442717e-07,
      "loss": 1.6368,
      "step": 634
    },
    {
      "epoch": 0.09576954980770681,
      "grad_norm": 0.2616059791844806,
      "learning_rate": 4.898913400430103e-07,
      "loss": 1.5624,
      "step": 635
    },
    {
      "epoch": 0.09592036799638036,
      "grad_norm": 0.2543080959885784,
      "learning_rate": 4.898597181923847e-07,
      "loss": 1.612,
      "step": 636
    },
    {
      "epoch": 0.09607118618505392,
      "grad_norm": 0.26165761435102747,
      "learning_rate": 4.89828048099495e-07,
      "loss": 1.5321,
      "step": 637
    },
    {
      "epoch": 0.09622200437372747,
      "grad_norm": 0.25832811520315874,
      "learning_rate": 4.89796329771452e-07,
      "loss": 1.5823,
      "step": 638
    },
    {
      "epoch": 0.09637282256240103,
      "grad_norm": 0.24902143318525818,
      "learning_rate": 4.897645632153773e-07,
      "loss": 1.5841,
      "step": 639
    },
    {
      "epoch": 0.09652364075107459,
      "grad_norm": 2.155448583198823,
      "learning_rate": 4.897327484384037e-07,
      "loss": 1.5538,
      "step": 640
    },
    {
      "epoch": 0.09667445893974813,
      "grad_norm": 0.29087209938764197,
      "learning_rate": 4.897008854476741e-07,
      "loss": 1.5588,
      "step": 641
    },
    {
      "epoch": 0.09682527712842169,
      "grad_norm": 0.2653741871413294,
      "learning_rate": 4.896689742503432e-07,
      "loss": 1.5673,
      "step": 642
    },
    {
      "epoch": 0.09697609531709524,
      "grad_norm": 0.25968820863722764,
      "learning_rate": 4.896370148535756e-07,
      "loss": 1.5111,
      "step": 643
    },
    {
      "epoch": 0.0971269135057688,
      "grad_norm": 0.25714475293882755,
      "learning_rate": 4.896050072645471e-07,
      "loss": 1.5501,
      "step": 644
    },
    {
      "epoch": 0.09727773169444234,
      "grad_norm": 0.2567542152360163,
      "learning_rate": 4.895729514904447e-07,
      "loss": 1.582,
      "step": 645
    },
    {
      "epoch": 0.0974285498831159,
      "grad_norm": 0.2437153380303258,
      "learning_rate": 4.895408475384656e-07,
      "loss": 1.5133,
      "step": 646
    },
    {
      "epoch": 0.09757936807178946,
      "grad_norm": 0.298717931310982,
      "learning_rate": 4.895086954158181e-07,
      "loss": 1.5918,
      "step": 647
    },
    {
      "epoch": 0.09773018626046301,
      "grad_norm": 0.2545301551858995,
      "learning_rate": 4.894764951297213e-07,
      "loss": 1.5648,
      "step": 648
    },
    {
      "epoch": 0.09788100444913657,
      "grad_norm": 0.26439815919936716,
      "learning_rate": 4.894442466874052e-07,
      "loss": 1.6693,
      "step": 649
    },
    {
      "epoch": 0.09803182263781011,
      "grad_norm": 0.2567876593677127,
      "learning_rate": 4.894119500961103e-07,
      "loss": 1.6272,
      "step": 650
    },
    {
      "epoch": 0.09818264082648367,
      "grad_norm": 0.2703561761668927,
      "learning_rate": 4.893796053630881e-07,
      "loss": 1.6037,
      "step": 651
    },
    {
      "epoch": 0.09833345901515723,
      "grad_norm": 0.24462862373595118,
      "learning_rate": 4.893472124956013e-07,
      "loss": 1.6007,
      "step": 652
    },
    {
      "epoch": 0.09848427720383078,
      "grad_norm": 0.27868827457946876,
      "learning_rate": 4.893147715009226e-07,
      "loss": 1.5594,
      "step": 653
    },
    {
      "epoch": 0.09863509539250434,
      "grad_norm": 0.2694330648878804,
      "learning_rate": 4.892822823863363e-07,
      "loss": 1.5226,
      "step": 654
    },
    {
      "epoch": 0.09878591358117789,
      "grad_norm": 0.2508658391074199,
      "learning_rate": 4.89249745159137e-07,
      "loss": 1.5539,
      "step": 655
    },
    {
      "epoch": 0.09893673176985145,
      "grad_norm": 0.25138448088883353,
      "learning_rate": 4.892171598266304e-07,
      "loss": 1.6533,
      "step": 656
    },
    {
      "epoch": 0.09908754995852499,
      "grad_norm": 0.25195751337900707,
      "learning_rate": 4.891845263961325e-07,
      "loss": 1.5837,
      "step": 657
    },
    {
      "epoch": 0.09923836814719855,
      "grad_norm": 0.2665540846353779,
      "learning_rate": 4.891518448749708e-07,
      "loss": 1.5776,
      "step": 658
    },
    {
      "epoch": 0.09938918633587211,
      "grad_norm": 0.2593627553827404,
      "learning_rate": 4.891191152704831e-07,
      "loss": 1.5844,
      "step": 659
    },
    {
      "epoch": 0.09954000452454566,
      "grad_norm": 0.2649200832665812,
      "learning_rate": 4.890863375900182e-07,
      "loss": 1.5452,
      "step": 660
    },
    {
      "epoch": 0.09969082271321922,
      "grad_norm": 0.26580937811421534,
      "learning_rate": 4.890535118409355e-07,
      "loss": 1.6071,
      "step": 661
    },
    {
      "epoch": 0.09984164090189276,
      "grad_norm": 0.2951878007593459,
      "learning_rate": 4.890206380306056e-07,
      "loss": 1.5651,
      "step": 662
    },
    {
      "epoch": 0.09999245909056632,
      "grad_norm": 0.25869622571347445,
      "learning_rate": 4.889877161664096e-07,
      "loss": 1.5353,
      "step": 663
    },
    {
      "epoch": 0.10014327727923987,
      "grad_norm": 0.24044540900199096,
      "learning_rate": 4.889547462557391e-07,
      "loss": 1.5694,
      "step": 664
    },
    {
      "epoch": 0.10029409546791343,
      "grad_norm": 0.2565865665249769,
      "learning_rate": 4.889217283059971e-07,
      "loss": 1.6183,
      "step": 665
    },
    {
      "epoch": 0.10044491365658699,
      "grad_norm": 0.27025535522679817,
      "learning_rate": 4.888886623245969e-07,
      "loss": 1.6222,
      "step": 666
    },
    {
      "epoch": 0.10059573184526054,
      "grad_norm": 0.24652564398079863,
      "learning_rate": 4.888555483189632e-07,
      "loss": 1.5629,
      "step": 667
    },
    {
      "epoch": 0.1007465500339341,
      "grad_norm": 0.4814577649778317,
      "learning_rate": 4.888223862965304e-07,
      "loss": 1.589,
      "step": 668
    },
    {
      "epoch": 0.10089736822260764,
      "grad_norm": 0.23937179549886292,
      "learning_rate": 4.887891762647449e-07,
      "loss": 1.5531,
      "step": 669
    },
    {
      "epoch": 0.1010481864112812,
      "grad_norm": 0.2466135765774502,
      "learning_rate": 4.887559182310629e-07,
      "loss": 1.6277,
      "step": 670
    },
    {
      "epoch": 0.10119900459995475,
      "grad_norm": 0.24673261333382188,
      "learning_rate": 4.887226122029522e-07,
      "loss": 1.5716,
      "step": 671
    },
    {
      "epoch": 0.10134982278862831,
      "grad_norm": 0.35516279706374765,
      "learning_rate": 4.886892581878906e-07,
      "loss": 1.5149,
      "step": 672
    },
    {
      "epoch": 0.10150064097730187,
      "grad_norm": 0.24643752858072748,
      "learning_rate": 4.886558561933673e-07,
      "loss": 1.5287,
      "step": 673
    },
    {
      "epoch": 0.10165145916597541,
      "grad_norm": 0.25727848949346177,
      "learning_rate": 4.886224062268819e-07,
      "loss": 1.544,
      "step": 674
    },
    {
      "epoch": 0.10180227735464897,
      "grad_norm": 0.2672375663532377,
      "learning_rate": 4.88588908295945e-07,
      "loss": 1.6145,
      "step": 675
    },
    {
      "epoch": 0.10195309554332252,
      "grad_norm": 0.23783832868779106,
      "learning_rate": 4.885553624080777e-07,
      "loss": 1.5434,
      "step": 676
    },
    {
      "epoch": 0.10210391373199608,
      "grad_norm": 0.2628061693510539,
      "learning_rate": 4.885217685708123e-07,
      "loss": 1.6046,
      "step": 677
    },
    {
      "epoch": 0.10225473192066964,
      "grad_norm": 0.2498006629530914,
      "learning_rate": 4.884881267916913e-07,
      "loss": 1.5738,
      "step": 678
    },
    {
      "epoch": 0.10240555010934319,
      "grad_norm": 0.2670401903771944,
      "learning_rate": 4.884544370782682e-07,
      "loss": 1.6719,
      "step": 679
    },
    {
      "epoch": 0.10255636829801675,
      "grad_norm": 0.26786252982477077,
      "learning_rate": 4.884206994381078e-07,
      "loss": 1.5068,
      "step": 680
    },
    {
      "epoch": 0.10270718648669029,
      "grad_norm": 0.2424177526813027,
      "learning_rate": 4.883869138787846e-07,
      "loss": 1.6749,
      "step": 681
    },
    {
      "epoch": 0.10285800467536385,
      "grad_norm": 0.2513530982644478,
      "learning_rate": 4.883530804078849e-07,
      "loss": 1.5723,
      "step": 682
    },
    {
      "epoch": 0.1030088228640374,
      "grad_norm": 0.25744150006151734,
      "learning_rate": 4.883191990330051e-07,
      "loss": 1.5466,
      "step": 683
    },
    {
      "epoch": 0.10315964105271096,
      "grad_norm": 0.27196436155706344,
      "learning_rate": 4.882852697617525e-07,
      "loss": 1.594,
      "step": 684
    },
    {
      "epoch": 0.10331045924138452,
      "grad_norm": 0.2533812358300188,
      "learning_rate": 4.882512926017453e-07,
      "loss": 1.5704,
      "step": 685
    },
    {
      "epoch": 0.10346127743005806,
      "grad_norm": 0.2651528199593723,
      "learning_rate": 4.882172675606125e-07,
      "loss": 1.5445,
      "step": 686
    },
    {
      "epoch": 0.10361209561873162,
      "grad_norm": 0.26571572796536885,
      "learning_rate": 4.881831946459933e-07,
      "loss": 1.54,
      "step": 687
    },
    {
      "epoch": 0.10376291380740517,
      "grad_norm": 0.24786037357426793,
      "learning_rate": 4.881490738655386e-07,
      "loss": 1.5229,
      "step": 688
    },
    {
      "epoch": 0.10391373199607873,
      "grad_norm": 0.25004542288508147,
      "learning_rate": 4.88114905226909e-07,
      "loss": 1.5828,
      "step": 689
    },
    {
      "epoch": 0.10406455018475227,
      "grad_norm": 0.2473729216074345,
      "learning_rate": 4.880806887377768e-07,
      "loss": 1.5187,
      "step": 690
    },
    {
      "epoch": 0.10421536837342583,
      "grad_norm": 0.251002717251045,
      "learning_rate": 4.880464244058242e-07,
      "loss": 1.591,
      "step": 691
    },
    {
      "epoch": 0.1043661865620994,
      "grad_norm": 0.2625610469104862,
      "learning_rate": 4.880121122387447e-07,
      "loss": 1.6434,
      "step": 692
    },
    {
      "epoch": 0.10451700475077294,
      "grad_norm": 0.3151656817305251,
      "learning_rate": 4.879777522442425e-07,
      "loss": 1.5391,
      "step": 693
    },
    {
      "epoch": 0.1046678229394465,
      "grad_norm": 0.24445992489148152,
      "learning_rate": 4.879433444300323e-07,
      "loss": 1.5501,
      "step": 694
    },
    {
      "epoch": 0.10481864112812005,
      "grad_norm": 4.8427358903679405,
      "learning_rate": 4.879088888038397e-07,
      "loss": 1.6371,
      "step": 695
    },
    {
      "epoch": 0.1049694593167936,
      "grad_norm": 0.26416228036696915,
      "learning_rate": 4.87874385373401e-07,
      "loss": 1.6385,
      "step": 696
    },
    {
      "epoch": 0.10512027750546717,
      "grad_norm": 0.2496641311371161,
      "learning_rate": 4.878398341464631e-07,
      "loss": 1.5365,
      "step": 697
    },
    {
      "epoch": 0.10527109569414071,
      "grad_norm": 0.2545167000001841,
      "learning_rate": 4.87805235130784e-07,
      "loss": 1.5949,
      "step": 698
    },
    {
      "epoch": 0.10542191388281427,
      "grad_norm": 0.27524392309011514,
      "learning_rate": 4.877705883341319e-07,
      "loss": 1.6025,
      "step": 699
    },
    {
      "epoch": 0.10557273207148782,
      "grad_norm": 0.2531395600898579,
      "learning_rate": 4.877358937642863e-07,
      "loss": 1.5657,
      "step": 700
    },
    {
      "epoch": 0.10572355026016138,
      "grad_norm": 0.24654464345054394,
      "learning_rate": 4.877011514290369e-07,
      "loss": 1.5843,
      "step": 701
    },
    {
      "epoch": 0.10587436844883492,
      "grad_norm": 0.3149996418781548,
      "learning_rate": 4.876663613361844e-07,
      "loss": 1.5075,
      "step": 702
    },
    {
      "epoch": 0.10602518663750848,
      "grad_norm": 0.2589830127512137,
      "learning_rate": 4.876315234935403e-07,
      "loss": 1.6373,
      "step": 703
    },
    {
      "epoch": 0.10617600482618204,
      "grad_norm": 0.31577873026015885,
      "learning_rate": 4.875966379089267e-07,
      "loss": 1.5944,
      "step": 704
    },
    {
      "epoch": 0.10632682301485559,
      "grad_norm": 12.03237047152764,
      "learning_rate": 4.875617045901763e-07,
      "loss": 1.5977,
      "step": 705
    },
    {
      "epoch": 0.10647764120352915,
      "grad_norm": 0.41114235828480006,
      "learning_rate": 4.875267235451328e-07,
      "loss": 1.5744,
      "step": 706
    },
    {
      "epoch": 0.1066284593922027,
      "grad_norm": 0.25165928626641787,
      "learning_rate": 4.874916947816504e-07,
      "loss": 1.5568,
      "step": 707
    },
    {
      "epoch": 0.10677927758087626,
      "grad_norm": 0.2548859391150305,
      "learning_rate": 4.874566183075942e-07,
      "loss": 1.503,
      "step": 708
    },
    {
      "epoch": 0.1069300957695498,
      "grad_norm": 0.3172895827908948,
      "learning_rate": 4.874214941308395e-07,
      "loss": 1.6655,
      "step": 709
    },
    {
      "epoch": 0.10708091395822336,
      "grad_norm": 0.2738482410825109,
      "learning_rate": 4.873863222592732e-07,
      "loss": 1.5105,
      "step": 710
    },
    {
      "epoch": 0.10723173214689692,
      "grad_norm": 0.2551922177918959,
      "learning_rate": 4.873511027007921e-07,
      "loss": 1.5545,
      "step": 711
    },
    {
      "epoch": 0.10738255033557047,
      "grad_norm": 0.26786371214739546,
      "learning_rate": 4.873158354633041e-07,
      "loss": 1.5731,
      "step": 712
    },
    {
      "epoch": 0.10753336852424403,
      "grad_norm": 0.251925518723259,
      "learning_rate": 4.872805205547276e-07,
      "loss": 1.5932,
      "step": 713
    },
    {
      "epoch": 0.10768418671291757,
      "grad_norm": 0.2430974358880398,
      "learning_rate": 4.872451579829922e-07,
      "loss": 1.5401,
      "step": 714
    },
    {
      "epoch": 0.10783500490159113,
      "grad_norm": 0.25083348588807275,
      "learning_rate": 4.872097477560374e-07,
      "loss": 1.5348,
      "step": 715
    },
    {
      "epoch": 0.10798582309026468,
      "grad_norm": 0.24851262093970097,
      "learning_rate": 4.87174289881814e-07,
      "loss": 1.5336,
      "step": 716
    },
    {
      "epoch": 0.10813664127893824,
      "grad_norm": 0.2580987044490573,
      "learning_rate": 4.871387843682834e-07,
      "loss": 1.5337,
      "step": 717
    },
    {
      "epoch": 0.1082874594676118,
      "grad_norm": 0.24269265338452864,
      "learning_rate": 4.871032312234175e-07,
      "loss": 1.6074,
      "step": 718
    },
    {
      "epoch": 0.10843827765628535,
      "grad_norm": 0.262540479014235,
      "learning_rate": 4.870676304551991e-07,
      "loss": 1.5434,
      "step": 719
    },
    {
      "epoch": 0.1085890958449589,
      "grad_norm": 0.25733851943997205,
      "learning_rate": 4.870319820716215e-07,
      "loss": 1.5998,
      "step": 720
    },
    {
      "epoch": 0.10873991403363245,
      "grad_norm": 0.30099795570434335,
      "learning_rate": 4.869962860806889e-07,
      "loss": 1.5917,
      "step": 721
    },
    {
      "epoch": 0.10889073222230601,
      "grad_norm": 0.24529089302262863,
      "learning_rate": 4.869605424904161e-07,
      "loss": 1.6178,
      "step": 722
    },
    {
      "epoch": 0.10904155041097957,
      "grad_norm": 0.3216523435582841,
      "learning_rate": 4.869247513088285e-07,
      "loss": 1.6293,
      "step": 723
    },
    {
      "epoch": 0.10919236859965312,
      "grad_norm": 0.2537191894534603,
      "learning_rate": 4.868889125439622e-07,
      "loss": 1.5027,
      "step": 724
    },
    {
      "epoch": 0.10934318678832668,
      "grad_norm": 0.23406712357409545,
      "learning_rate": 4.868530262038642e-07,
      "loss": 1.5506,
      "step": 725
    },
    {
      "epoch": 0.10949400497700022,
      "grad_norm": 0.24424934141970756,
      "learning_rate": 4.868170922965921e-07,
      "loss": 1.62,
      "step": 726
    },
    {
      "epoch": 0.10964482316567378,
      "grad_norm": 0.2529106313948061,
      "learning_rate": 4.867811108302139e-07,
      "loss": 1.6838,
      "step": 727
    },
    {
      "epoch": 0.10979564135434733,
      "grad_norm": 0.24268045391162135,
      "learning_rate": 4.867450818128086e-07,
      "loss": 1.5843,
      "step": 728
    },
    {
      "epoch": 0.10994645954302089,
      "grad_norm": 0.2782845964267312,
      "learning_rate": 4.867090052524657e-07,
      "loss": 1.5276,
      "step": 729
    },
    {
      "epoch": 0.11009727773169445,
      "grad_norm": 0.2607417299475057,
      "learning_rate": 4.866728811572855e-07,
      "loss": 1.6077,
      "step": 730
    },
    {
      "epoch": 0.110248095920368,
      "grad_norm": 0.2995072103389239,
      "learning_rate": 4.866367095353788e-07,
      "loss": 1.531,
      "step": 731
    },
    {
      "epoch": 0.11039891410904155,
      "grad_norm": 0.26231666932903636,
      "learning_rate": 4.866004903948674e-07,
      "loss": 1.6113,
      "step": 732
    },
    {
      "epoch": 0.1105497322977151,
      "grad_norm": 0.255610027357904,
      "learning_rate": 4.865642237438832e-07,
      "loss": 1.5377,
      "step": 733
    },
    {
      "epoch": 0.11070055048638866,
      "grad_norm": 0.24761027967399435,
      "learning_rate": 4.865279095905696e-07,
      "loss": 1.5794,
      "step": 734
    },
    {
      "epoch": 0.1108513686750622,
      "grad_norm": 0.3683492891040185,
      "learning_rate": 4.864915479430799e-07,
      "loss": 1.6031,
      "step": 735
    },
    {
      "epoch": 0.11100218686373577,
      "grad_norm": 0.2984691028171984,
      "learning_rate": 4.864551388095782e-07,
      "loss": 1.5659,
      "step": 736
    },
    {
      "epoch": 0.11115300505240933,
      "grad_norm": 0.24189257268944578,
      "learning_rate": 4.864186821982397e-07,
      "loss": 1.5652,
      "step": 737
    },
    {
      "epoch": 0.11130382324108287,
      "grad_norm": 0.2370044032129546,
      "learning_rate": 4.863821781172499e-07,
      "loss": 1.571,
      "step": 738
    },
    {
      "epoch": 0.11145464142975643,
      "grad_norm": 0.2555003933014576,
      "learning_rate": 4.863456265748048e-07,
      "loss": 1.5888,
      "step": 739
    },
    {
      "epoch": 0.11160545961842998,
      "grad_norm": 0.543423385722154,
      "learning_rate": 4.863090275791116e-07,
      "loss": 1.5519,
      "step": 740
    },
    {
      "epoch": 0.11175627780710354,
      "grad_norm": 0.27599159260109296,
      "learning_rate": 4.862723811383877e-07,
      "loss": 1.602,
      "step": 741
    },
    {
      "epoch": 0.11190709599577708,
      "grad_norm": 0.2465362472728453,
      "learning_rate": 4.862356872608612e-07,
      "loss": 1.5474,
      "step": 742
    },
    {
      "epoch": 0.11205791418445064,
      "grad_norm": 0.2800510702511786,
      "learning_rate": 4.861989459547712e-07,
      "loss": 1.5227,
      "step": 743
    },
    {
      "epoch": 0.1122087323731242,
      "grad_norm": 0.32072850809900044,
      "learning_rate": 4.861621572283669e-07,
      "loss": 1.6118,
      "step": 744
    },
    {
      "epoch": 0.11235955056179775,
      "grad_norm": 0.2510399301195033,
      "learning_rate": 4.861253210899088e-07,
      "loss": 1.6143,
      "step": 745
    },
    {
      "epoch": 0.11251036875047131,
      "grad_norm": 0.2410159118441585,
      "learning_rate": 4.860884375476672e-07,
      "loss": 1.5808,
      "step": 746
    },
    {
      "epoch": 0.11266118693914486,
      "grad_norm": 0.2564249523301703,
      "learning_rate": 4.86051506609924e-07,
      "loss": 1.5642,
      "step": 747
    },
    {
      "epoch": 0.11281200512781842,
      "grad_norm": 0.2697763677983635,
      "learning_rate": 4.86014528284971e-07,
      "loss": 1.6284,
      "step": 748
    },
    {
      "epoch": 0.11296282331649198,
      "grad_norm": 0.24748116614837487,
      "learning_rate": 4.85977502581111e-07,
      "loss": 1.5848,
      "step": 749
    },
    {
      "epoch": 0.11311364150516552,
      "grad_norm": 0.24419802618009506,
      "learning_rate": 4.859404295066573e-07,
      "loss": 1.5726,
      "step": 750
    },
    {
      "epoch": 0.11326445969383908,
      "grad_norm": 0.2658317522730182,
      "learning_rate": 4.859033090699338e-07,
      "loss": 1.5658,
      "step": 751
    },
    {
      "epoch": 0.11341527788251263,
      "grad_norm": 0.2585887332026061,
      "learning_rate": 4.858661412792754e-07,
      "loss": 1.6313,
      "step": 752
    },
    {
      "epoch": 0.11356609607118619,
      "grad_norm": 0.265337834018552,
      "learning_rate": 4.858289261430271e-07,
      "loss": 1.5918,
      "step": 753
    },
    {
      "epoch": 0.11371691425985973,
      "grad_norm": 0.28811771044154877,
      "learning_rate": 4.857916636695448e-07,
      "loss": 1.7301,
      "step": 754
    },
    {
      "epoch": 0.1138677324485333,
      "grad_norm": 0.2775887071260114,
      "learning_rate": 4.857543538671952e-07,
      "loss": 1.6004,
      "step": 755
    },
    {
      "epoch": 0.11401855063720685,
      "grad_norm": 1.307067680039299,
      "learning_rate": 4.857169967443553e-07,
      "loss": 1.5489,
      "step": 756
    },
    {
      "epoch": 0.1141693688258804,
      "grad_norm": 0.25603868221014153,
      "learning_rate": 4.856795923094128e-07,
      "loss": 1.5895,
      "step": 757
    },
    {
      "epoch": 0.11432018701455396,
      "grad_norm": 0.26609703406081053,
      "learning_rate": 4.856421405707662e-07,
      "loss": 1.5814,
      "step": 758
    },
    {
      "epoch": 0.1144710052032275,
      "grad_norm": 0.24417096229180008,
      "learning_rate": 4.856046415368244e-07,
      "loss": 1.5387,
      "step": 759
    },
    {
      "epoch": 0.11462182339190106,
      "grad_norm": 0.25096964734102967,
      "learning_rate": 4.855670952160073e-07,
      "loss": 1.5615,
      "step": 760
    },
    {
      "epoch": 0.11477264158057461,
      "grad_norm": 0.28760577193425085,
      "learning_rate": 4.855295016167447e-07,
      "loss": 1.5391,
      "step": 761
    },
    {
      "epoch": 0.11492345976924817,
      "grad_norm": 0.2699115214561961,
      "learning_rate": 4.854918607474777e-07,
      "loss": 1.5244,
      "step": 762
    },
    {
      "epoch": 0.11507427795792173,
      "grad_norm": 0.25905586243470197,
      "learning_rate": 4.854541726166579e-07,
      "loss": 1.4722,
      "step": 763
    },
    {
      "epoch": 0.11522509614659528,
      "grad_norm": 0.275105267887951,
      "learning_rate": 4.854164372327471e-07,
      "loss": 1.5588,
      "step": 764
    },
    {
      "epoch": 0.11537591433526884,
      "grad_norm": 0.23764545222034308,
      "learning_rate": 4.853786546042184e-07,
      "loss": 1.5995,
      "step": 765
    },
    {
      "epoch": 0.11552673252394238,
      "grad_norm": 0.2509047647028819,
      "learning_rate": 4.853408247395545e-07,
      "loss": 1.5454,
      "step": 766
    },
    {
      "epoch": 0.11567755071261594,
      "grad_norm": 0.24100699069586992,
      "learning_rate": 4.853029476472499e-07,
      "loss": 1.6153,
      "step": 767
    },
    {
      "epoch": 0.1158283689012895,
      "grad_norm": 0.24582457784503464,
      "learning_rate": 4.852650233358088e-07,
      "loss": 1.5976,
      "step": 768
    },
    {
      "epoch": 0.11597918708996305,
      "grad_norm": 0.2515327610744451,
      "learning_rate": 4.852270518137463e-07,
      "loss": 1.5088,
      "step": 769
    },
    {
      "epoch": 0.11613000527863661,
      "grad_norm": 0.28108513747998987,
      "learning_rate": 4.851890330895882e-07,
      "loss": 1.5592,
      "step": 770
    },
    {
      "epoch": 0.11628082346731015,
      "grad_norm": 0.25724095740981906,
      "learning_rate": 4.851509671718708e-07,
      "loss": 1.5793,
      "step": 771
    },
    {
      "epoch": 0.11643164165598371,
      "grad_norm": 0.3379293339765616,
      "learning_rate": 4.851128540691409e-07,
      "loss": 1.5388,
      "step": 772
    },
    {
      "epoch": 0.11658245984465726,
      "grad_norm": 0.266271201005545,
      "learning_rate": 4.850746937899562e-07,
      "loss": 1.6516,
      "step": 773
    },
    {
      "epoch": 0.11673327803333082,
      "grad_norm": 0.27413359706065926,
      "learning_rate": 4.850364863428847e-07,
      "loss": 1.5685,
      "step": 774
    },
    {
      "epoch": 0.11688409622200438,
      "grad_norm": 0.2532201394692201,
      "learning_rate": 4.84998231736505e-07,
      "loss": 1.6207,
      "step": 775
    },
    {
      "epoch": 0.11703491441067793,
      "grad_norm": 0.24793843250269984,
      "learning_rate": 4.849599299794065e-07,
      "loss": 1.6016,
      "step": 776
    },
    {
      "epoch": 0.11718573259935149,
      "grad_norm": 2.5487673069833954,
      "learning_rate": 4.849215810801891e-07,
      "loss": 1.5861,
      "step": 777
    },
    {
      "epoch": 0.11733655078802503,
      "grad_norm": 0.2585038483321872,
      "learning_rate": 4.84883185047463e-07,
      "loss": 1.547,
      "step": 778
    },
    {
      "epoch": 0.11748736897669859,
      "grad_norm": 0.26306792481033825,
      "learning_rate": 4.848447418898495e-07,
      "loss": 1.5976,
      "step": 779
    },
    {
      "epoch": 0.11763818716537214,
      "grad_norm": 0.25036190245725615,
      "learning_rate": 4.8480625161598e-07,
      "loss": 1.6552,
      "step": 780
    },
    {
      "epoch": 0.1177890053540457,
      "grad_norm": 0.2788788641094789,
      "learning_rate": 4.847677142344968e-07,
      "loss": 1.5277,
      "step": 781
    },
    {
      "epoch": 0.11793982354271926,
      "grad_norm": 0.26398224032511813,
      "learning_rate": 4.847291297540528e-07,
      "loss": 1.6165,
      "step": 782
    },
    {
      "epoch": 0.1180906417313928,
      "grad_norm": 0.2923197817267704,
      "learning_rate": 4.84690498183311e-07,
      "loss": 1.6169,
      "step": 783
    },
    {
      "epoch": 0.11824145992006636,
      "grad_norm": 0.2663331591259635,
      "learning_rate": 4.846518195309456e-07,
      "loss": 1.5577,
      "step": 784
    },
    {
      "epoch": 0.11839227810873991,
      "grad_norm": 0.2535735629174223,
      "learning_rate": 4.846130938056408e-07,
      "loss": 1.6244,
      "step": 785
    },
    {
      "epoch": 0.11854309629741347,
      "grad_norm": 0.4299089038786934,
      "learning_rate": 4.84574321016092e-07,
      "loss": 1.5036,
      "step": 786
    },
    {
      "epoch": 0.11869391448608702,
      "grad_norm": 0.2611073794072947,
      "learning_rate": 4.845355011710046e-07,
      "loss": 1.6248,
      "step": 787
    },
    {
      "epoch": 0.11884473267476058,
      "grad_norm": 0.34084186540244277,
      "learning_rate": 4.844966342790949e-07,
      "loss": 1.5502,
      "step": 788
    },
    {
      "epoch": 0.11899555086343414,
      "grad_norm": 0.6743003737430772,
      "learning_rate": 4.844577203490895e-07,
      "loss": 1.4726,
      "step": 789
    },
    {
      "epoch": 0.11914636905210768,
      "grad_norm": 0.25381971085918803,
      "learning_rate": 4.844187593897258e-07,
      "loss": 1.5452,
      "step": 790
    },
    {
      "epoch": 0.11929718724078124,
      "grad_norm": 0.27775315318526334,
      "learning_rate": 4.843797514097518e-07,
      "loss": 1.5751,
      "step": 791
    },
    {
      "epoch": 0.11944800542945479,
      "grad_norm": 0.27009087648015045,
      "learning_rate": 4.843406964179256e-07,
      "loss": 1.4924,
      "step": 792
    },
    {
      "epoch": 0.11959882361812835,
      "grad_norm": 0.2625048270437366,
      "learning_rate": 4.843015944230165e-07,
      "loss": 1.6697,
      "step": 793
    },
    {
      "epoch": 0.11974964180680191,
      "grad_norm": 0.25396359169098826,
      "learning_rate": 4.84262445433804e-07,
      "loss": 1.5235,
      "step": 794
    },
    {
      "epoch": 0.11990045999547545,
      "grad_norm": 0.24250088924809574,
      "learning_rate": 4.842232494590779e-07,
      "loss": 1.6287,
      "step": 795
    },
    {
      "epoch": 0.12005127818414901,
      "grad_norm": 0.26372563090965606,
      "learning_rate": 4.841840065076392e-07,
      "loss": 1.6462,
      "step": 796
    },
    {
      "epoch": 0.12020209637282256,
      "grad_norm": 0.2527762179440438,
      "learning_rate": 4.841447165882988e-07,
      "loss": 1.5983,
      "step": 797
    },
    {
      "epoch": 0.12035291456149612,
      "grad_norm": 0.2408908261902159,
      "learning_rate": 4.841053797098787e-07,
      "loss": 1.6323,
      "step": 798
    },
    {
      "epoch": 0.12050373275016966,
      "grad_norm": 0.33754799043680367,
      "learning_rate": 4.84065995881211e-07,
      "loss": 1.5269,
      "step": 799
    },
    {
      "epoch": 0.12065455093884322,
      "grad_norm": 0.23290067250893015,
      "learning_rate": 4.840265651111384e-07,
      "loss": 1.5957,
      "step": 800
    },
    {
      "epoch": 0.12080536912751678,
      "grad_norm": 0.2521050639135633,
      "learning_rate": 4.839870874085147e-07,
      "loss": 1.5405,
      "step": 801
    },
    {
      "epoch": 0.12095618731619033,
      "grad_norm": 0.2658430314654051,
      "learning_rate": 4.839475627822033e-07,
      "loss": 1.5346,
      "step": 802
    },
    {
      "epoch": 0.12110700550486389,
      "grad_norm": 0.2605551980486632,
      "learning_rate": 4.839079912410788e-07,
      "loss": 1.5528,
      "step": 803
    },
    {
      "epoch": 0.12125782369353744,
      "grad_norm": 0.25637582876320203,
      "learning_rate": 4.838683727940262e-07,
      "loss": 1.514,
      "step": 804
    },
    {
      "epoch": 0.121408641882211,
      "grad_norm": 0.2964856263950721,
      "learning_rate": 4.83828707449941e-07,
      "loss": 1.5536,
      "step": 805
    },
    {
      "epoch": 0.12155946007088454,
      "grad_norm": 0.24779737219807155,
      "learning_rate": 4.837889952177293e-07,
      "loss": 1.5783,
      "step": 806
    },
    {
      "epoch": 0.1217102782595581,
      "grad_norm": 0.24262288865254855,
      "learning_rate": 4.837492361063075e-07,
      "loss": 1.5754,
      "step": 807
    },
    {
      "epoch": 0.12186109644823166,
      "grad_norm": 0.25703278119585443,
      "learning_rate": 4.837094301246027e-07,
      "loss": 1.5389,
      "step": 808
    },
    {
      "epoch": 0.12201191463690521,
      "grad_norm": 0.24606774271785903,
      "learning_rate": 4.836695772815527e-07,
      "loss": 1.5562,
      "step": 809
    },
    {
      "epoch": 0.12216273282557877,
      "grad_norm": 0.2638024822506071,
      "learning_rate": 4.836296775861053e-07,
      "loss": 1.5749,
      "step": 810
    },
    {
      "epoch": 0.12231355101425231,
      "grad_norm": 0.2443613491949051,
      "learning_rate": 4.835897310472194e-07,
      "loss": 1.5665,
      "step": 811
    },
    {
      "epoch": 0.12246436920292587,
      "grad_norm": 0.26880244150319627,
      "learning_rate": 4.835497376738641e-07,
      "loss": 1.5263,
      "step": 812
    },
    {
      "epoch": 0.12261518739159942,
      "grad_norm": 0.24512045156197035,
      "learning_rate": 4.83509697475019e-07,
      "loss": 1.5692,
      "step": 813
    },
    {
      "epoch": 0.12276600558027298,
      "grad_norm": 0.24820930430496924,
      "learning_rate": 4.834696104596743e-07,
      "loss": 1.5603,
      "step": 814
    },
    {
      "epoch": 0.12291682376894654,
      "grad_norm": 0.42535360243645115,
      "learning_rate": 4.834294766368308e-07,
      "loss": 1.6325,
      "step": 815
    },
    {
      "epoch": 0.12306764195762009,
      "grad_norm": 0.25780958887909644,
      "learning_rate": 4.833892960154997e-07,
      "loss": 1.6142,
      "step": 816
    },
    {
      "epoch": 0.12321846014629365,
      "grad_norm": 0.27275400494629926,
      "learning_rate": 4.833490686047026e-07,
      "loss": 1.5863,
      "step": 817
    },
    {
      "epoch": 0.12336927833496719,
      "grad_norm": 0.2476330246188185,
      "learning_rate": 4.833087944134719e-07,
      "loss": 1.6803,
      "step": 818
    },
    {
      "epoch": 0.12352009652364075,
      "grad_norm": 0.24146678674912134,
      "learning_rate": 4.832684734508501e-07,
      "loss": 1.528,
      "step": 819
    },
    {
      "epoch": 0.12367091471231431,
      "grad_norm": 0.250043437502425,
      "learning_rate": 4.832281057258907e-07,
      "loss": 1.5728,
      "step": 820
    },
    {
      "epoch": 0.12382173290098786,
      "grad_norm": 0.24954721250421497,
      "learning_rate": 4.831876912476572e-07,
      "loss": 1.6055,
      "step": 821
    },
    {
      "epoch": 0.12397255108966142,
      "grad_norm": 0.40705819595065323,
      "learning_rate": 4.83147230025224e-07,
      "loss": 1.6186,
      "step": 822
    },
    {
      "epoch": 0.12412336927833496,
      "grad_norm": 0.32773363272612815,
      "learning_rate": 4.831067220676756e-07,
      "loss": 1.5751,
      "step": 823
    },
    {
      "epoch": 0.12427418746700852,
      "grad_norm": 0.26807163443396315,
      "learning_rate": 4.830661673841075e-07,
      "loss": 1.5544,
      "step": 824
    },
    {
      "epoch": 0.12442500565568207,
      "grad_norm": 0.24261889068804798,
      "learning_rate": 4.830255659836252e-07,
      "loss": 1.6377,
      "step": 825
    },
    {
      "epoch": 0.12457582384435563,
      "grad_norm": 0.34382157172301875,
      "learning_rate": 4.829849178753449e-07,
      "loss": 1.5715,
      "step": 826
    },
    {
      "epoch": 0.12472664203302919,
      "grad_norm": 0.316169860210875,
      "learning_rate": 4.829442230683935e-07,
      "loss": 1.581,
      "step": 827
    },
    {
      "epoch": 0.12487746022170274,
      "grad_norm": 0.2865289199919012,
      "learning_rate": 4.82903481571908e-07,
      "loss": 1.5738,
      "step": 828
    },
    {
      "epoch": 0.12502827841037628,
      "grad_norm": 0.2648798194129721,
      "learning_rate": 4.828626933950359e-07,
      "loss": 1.624,
      "step": 829
    },
    {
      "epoch": 0.12517909659904985,
      "grad_norm": 0.29238962734359203,
      "learning_rate": 4.828218585469357e-07,
      "loss": 1.509,
      "step": 830
    },
    {
      "epoch": 0.1253299147877234,
      "grad_norm": 0.5511438047008799,
      "learning_rate": 4.827809770367757e-07,
      "loss": 1.5524,
      "step": 831
    },
    {
      "epoch": 0.12548073297639695,
      "grad_norm": 0.26740428045118964,
      "learning_rate": 4.827400488737351e-07,
      "loss": 1.5907,
      "step": 832
    },
    {
      "epoch": 0.12563155116507052,
      "grad_norm": 0.26308461790104193,
      "learning_rate": 4.826990740670035e-07,
      "loss": 1.64,
      "step": 833
    },
    {
      "epoch": 0.12578236935374407,
      "grad_norm": 0.24263120576507818,
      "learning_rate": 4.826580526257809e-07,
      "loss": 1.5427,
      "step": 834
    },
    {
      "epoch": 0.1259331875424176,
      "grad_norm": 0.24279432366476036,
      "learning_rate": 4.826169845592778e-07,
      "loss": 1.5294,
      "step": 835
    },
    {
      "epoch": 0.12608400573109116,
      "grad_norm": 0.2590934265442616,
      "learning_rate": 4.825758698767152e-07,
      "loss": 1.607,
      "step": 836
    },
    {
      "epoch": 0.12623482391976473,
      "grad_norm": 1.123102233200613,
      "learning_rate": 4.825347085873245e-07,
      "loss": 1.4723,
      "step": 837
    },
    {
      "epoch": 0.12638564210843828,
      "grad_norm": 0.27876322471312537,
      "learning_rate": 4.824935007003478e-07,
      "loss": 1.5882,
      "step": 838
    },
    {
      "epoch": 0.12653646029711182,
      "grad_norm": 0.26911085288321157,
      "learning_rate": 4.824522462250371e-07,
      "loss": 1.6014,
      "step": 839
    },
    {
      "epoch": 0.1266872784857854,
      "grad_norm": 0.2425461341792373,
      "learning_rate": 4.824109451706556e-07,
      "loss": 1.5061,
      "step": 840
    },
    {
      "epoch": 0.12683809667445894,
      "grad_norm": 0.24613790002709662,
      "learning_rate": 4.823695975464764e-07,
      "loss": 1.5627,
      "step": 841
    },
    {
      "epoch": 0.1269889148631325,
      "grad_norm": 0.25215547445900216,
      "learning_rate": 4.823282033617832e-07,
      "loss": 1.5776,
      "step": 842
    },
    {
      "epoch": 0.12713973305180604,
      "grad_norm": 0.24529552413588745,
      "learning_rate": 4.822867626258704e-07,
      "loss": 1.6023,
      "step": 843
    },
    {
      "epoch": 0.1272905512404796,
      "grad_norm": 0.2550713983118439,
      "learning_rate": 4.822452753480425e-07,
      "loss": 1.6527,
      "step": 844
    },
    {
      "epoch": 0.12744136942915316,
      "grad_norm": 0.2676107121499945,
      "learning_rate": 4.822037415376146e-07,
      "loss": 1.6495,
      "step": 845
    },
    {
      "epoch": 0.1275921876178267,
      "grad_norm": 0.2819066663763421,
      "learning_rate": 4.821621612039123e-07,
      "loss": 1.5682,
      "step": 846
    },
    {
      "epoch": 0.12774300580650028,
      "grad_norm": 0.326742076073443,
      "learning_rate": 4.821205343562714e-07,
      "loss": 1.561,
      "step": 847
    },
    {
      "epoch": 0.12789382399517382,
      "grad_norm": 0.253293756714276,
      "learning_rate": 4.820788610040388e-07,
      "loss": 1.5217,
      "step": 848
    },
    {
      "epoch": 0.12804464218384737,
      "grad_norm": 0.24698120971930787,
      "learning_rate": 4.82037141156571e-07,
      "loss": 1.5475,
      "step": 849
    },
    {
      "epoch": 0.12819546037252091,
      "grad_norm": 0.2552342548833313,
      "learning_rate": 4.819953748232354e-07,
      "loss": 1.6362,
      "step": 850
    },
    {
      "epoch": 0.1283462785611945,
      "grad_norm": 0.2489531724751376,
      "learning_rate": 4.819535620134098e-07,
      "loss": 1.4661,
      "step": 851
    },
    {
      "epoch": 0.12849709674986803,
      "grad_norm": 0.2561641259473689,
      "learning_rate": 4.819117027364824e-07,
      "loss": 1.6147,
      "step": 852
    },
    {
      "epoch": 0.12864791493854158,
      "grad_norm": 0.248751198019234,
      "learning_rate": 4.818697970018517e-07,
      "loss": 1.5935,
      "step": 853
    },
    {
      "epoch": 0.12879873312721515,
      "grad_norm": 0.259225532039798,
      "learning_rate": 4.81827844818927e-07,
      "loss": 1.5667,
      "step": 854
    },
    {
      "epoch": 0.1289495513158887,
      "grad_norm": 0.36433050928979993,
      "learning_rate": 4.817858461971275e-07,
      "loss": 1.5721,
      "step": 855
    },
    {
      "epoch": 0.12910036950456225,
      "grad_norm": 0.2495349007213541,
      "learning_rate": 4.817438011458835e-07,
      "loss": 1.5774,
      "step": 856
    },
    {
      "epoch": 0.1292511876932358,
      "grad_norm": 0.26017903422953986,
      "learning_rate": 4.817017096746349e-07,
      "loss": 1.6409,
      "step": 857
    },
    {
      "epoch": 0.12940200588190937,
      "grad_norm": 0.2577039923092752,
      "learning_rate": 4.816595717928327e-07,
      "loss": 1.5082,
      "step": 858
    },
    {
      "epoch": 0.1295528240705829,
      "grad_norm": 0.25190571322660016,
      "learning_rate": 4.81617387509938e-07,
      "loss": 1.6199,
      "step": 859
    },
    {
      "epoch": 0.12970364225925646,
      "grad_norm": 0.25108026522987853,
      "learning_rate": 4.815751568354225e-07,
      "loss": 1.608,
      "step": 860
    },
    {
      "epoch": 0.12985446044793003,
      "grad_norm": 0.2812535177205302,
      "learning_rate": 4.815328797787681e-07,
      "loss": 1.5383,
      "step": 861
    },
    {
      "epoch": 0.13000527863660358,
      "grad_norm": 0.2912235666420491,
      "learning_rate": 4.814905563494674e-07,
      "loss": 1.6123,
      "step": 862
    },
    {
      "epoch": 0.13015609682527712,
      "grad_norm": 0.23793333653321677,
      "learning_rate": 4.81448186557023e-07,
      "loss": 1.5155,
      "step": 863
    },
    {
      "epoch": 0.13030691501395067,
      "grad_norm": 0.24211422876016309,
      "learning_rate": 4.814057704109483e-07,
      "loss": 1.5627,
      "step": 864
    },
    {
      "epoch": 0.13045773320262424,
      "grad_norm": 0.3005613543933885,
      "learning_rate": 4.81363307920767e-07,
      "loss": 1.5363,
      "step": 865
    },
    {
      "epoch": 0.1306085513912978,
      "grad_norm": 0.26748788516683486,
      "learning_rate": 4.813207990960131e-07,
      "loss": 1.5645,
      "step": 866
    },
    {
      "epoch": 0.13075936957997134,
      "grad_norm": 0.2467019690508945,
      "learning_rate": 4.812782439462311e-07,
      "loss": 1.6062,
      "step": 867
    },
    {
      "epoch": 0.1309101877686449,
      "grad_norm": 0.24050701469204028,
      "learning_rate": 4.812356424809756e-07,
      "loss": 1.5592,
      "step": 868
    },
    {
      "epoch": 0.13106100595731845,
      "grad_norm": 0.24566700637291505,
      "learning_rate": 4.811929947098125e-07,
      "loss": 1.4948,
      "step": 869
    },
    {
      "epoch": 0.131211824145992,
      "grad_norm": 0.24578461985233002,
      "learning_rate": 4.811503006423168e-07,
      "loss": 1.5725,
      "step": 870
    },
    {
      "epoch": 0.13136264233466555,
      "grad_norm": 0.2766474882116995,
      "learning_rate": 4.81107560288075e-07,
      "loss": 1.5085,
      "step": 871
    },
    {
      "epoch": 0.13151346052333912,
      "grad_norm": 0.3032106098903552,
      "learning_rate": 4.810647736566833e-07,
      "loss": 1.5573,
      "step": 872
    },
    {
      "epoch": 0.13166427871201267,
      "grad_norm": 0.26227182055075776,
      "learning_rate": 4.810219407577488e-07,
      "loss": 1.5504,
      "step": 873
    },
    {
      "epoch": 0.1318150969006862,
      "grad_norm": 0.30231661285945577,
      "learning_rate": 4.809790616008884e-07,
      "loss": 1.6157,
      "step": 874
    },
    {
      "epoch": 0.1319659150893598,
      "grad_norm": 0.24138594952190437,
      "learning_rate": 4.8093613619573e-07,
      "loss": 1.5301,
      "step": 875
    },
    {
      "epoch": 0.13211673327803333,
      "grad_norm": 0.28015148103115745,
      "learning_rate": 4.808931645519115e-07,
      "loss": 1.6091,
      "step": 876
    },
    {
      "epoch": 0.13226755146670688,
      "grad_norm": 0.25583472974312804,
      "learning_rate": 4.808501466790813e-07,
      "loss": 1.6112,
      "step": 877
    },
    {
      "epoch": 0.13241836965538045,
      "grad_norm": 0.2550840996404767,
      "learning_rate": 4.808070825868981e-07,
      "loss": 1.5721,
      "step": 878
    },
    {
      "epoch": 0.132569187844054,
      "grad_norm": 0.2386014317840139,
      "learning_rate": 4.80763972285031e-07,
      "loss": 1.561,
      "step": 879
    },
    {
      "epoch": 0.13272000603272754,
      "grad_norm": 0.26463646113746936,
      "learning_rate": 4.807208157831598e-07,
      "loss": 1.5963,
      "step": 880
    },
    {
      "epoch": 0.1328708242214011,
      "grad_norm": 0.24789221098898107,
      "learning_rate": 4.806776130909741e-07,
      "loss": 1.5585,
      "step": 881
    },
    {
      "epoch": 0.13302164241007466,
      "grad_norm": 0.27954549479073476,
      "learning_rate": 4.806343642181743e-07,
      "loss": 1.5257,
      "step": 882
    },
    {
      "epoch": 0.1331724605987482,
      "grad_norm": 0.2583555304842065,
      "learning_rate": 4.80591069174471e-07,
      "loss": 1.6395,
      "step": 883
    },
    {
      "epoch": 0.13332327878742176,
      "grad_norm": 0.2747169063226922,
      "learning_rate": 4.805477279695852e-07,
      "loss": 1.6829,
      "step": 884
    },
    {
      "epoch": 0.13347409697609533,
      "grad_norm": 0.24600349536308114,
      "learning_rate": 4.805043406132482e-07,
      "loss": 1.5371,
      "step": 885
    },
    {
      "epoch": 0.13362491516476888,
      "grad_norm": 0.25980027742127176,
      "learning_rate": 4.804609071152018e-07,
      "loss": 1.6184,
      "step": 886
    },
    {
      "epoch": 0.13377573335344242,
      "grad_norm": 0.24508614968710926,
      "learning_rate": 4.804174274851981e-07,
      "loss": 1.6147,
      "step": 887
    },
    {
      "epoch": 0.13392655154211597,
      "grad_norm": 0.24804260240667214,
      "learning_rate": 4.803739017329996e-07,
      "loss": 1.6067,
      "step": 888
    },
    {
      "epoch": 0.13407736973078954,
      "grad_norm": 0.25405274236319236,
      "learning_rate": 4.803303298683789e-07,
      "loss": 1.5351,
      "step": 889
    },
    {
      "epoch": 0.1342281879194631,
      "grad_norm": 0.27238479371505225,
      "learning_rate": 4.802867119011193e-07,
      "loss": 1.5735,
      "step": 890
    },
    {
      "epoch": 0.13437900610813663,
      "grad_norm": 0.23928993849704544,
      "learning_rate": 4.802430478410142e-07,
      "loss": 1.542,
      "step": 891
    },
    {
      "epoch": 0.1345298242968102,
      "grad_norm": 0.27162527646949547,
      "learning_rate": 4.801993376978676e-07,
      "loss": 1.5939,
      "step": 892
    },
    {
      "epoch": 0.13468064248548375,
      "grad_norm": 0.24118953502840654,
      "learning_rate": 4.801555814814936e-07,
      "loss": 1.5485,
      "step": 893
    },
    {
      "epoch": 0.1348314606741573,
      "grad_norm": 0.24615309745922606,
      "learning_rate": 4.801117792017168e-07,
      "loss": 1.6679,
      "step": 894
    },
    {
      "epoch": 0.13498227886283085,
      "grad_norm": 0.26089481234827916,
      "learning_rate": 4.80067930868372e-07,
      "loss": 1.5373,
      "step": 895
    },
    {
      "epoch": 0.13513309705150442,
      "grad_norm": 0.24563967703761258,
      "learning_rate": 4.800240364913044e-07,
      "loss": 1.6071,
      "step": 896
    },
    {
      "epoch": 0.13528391524017797,
      "grad_norm": 0.3388244706396415,
      "learning_rate": 4.799800960803698e-07,
      "loss": 1.581,
      "step": 897
    },
    {
      "epoch": 0.1354347334288515,
      "grad_norm": 0.2625542380631137,
      "learning_rate": 4.79936109645434e-07,
      "loss": 1.523,
      "step": 898
    },
    {
      "epoch": 0.13558555161752509,
      "grad_norm": 0.25532417159291754,
      "learning_rate": 4.798920771963731e-07,
      "loss": 1.6067,
      "step": 899
    },
    {
      "epoch": 0.13573636980619863,
      "grad_norm": 0.268227095836578,
      "learning_rate": 4.798479987430737e-07,
      "loss": 1.5463,
      "step": 900
    },
    {
      "epoch": 0.13588718799487218,
      "grad_norm": 0.26815668103982254,
      "learning_rate": 4.798038742954327e-07,
      "loss": 1.586,
      "step": 901
    },
    {
      "epoch": 0.13603800618354572,
      "grad_norm": 0.24824025017149767,
      "learning_rate": 4.797597038633576e-07,
      "loss": 1.5606,
      "step": 902
    },
    {
      "epoch": 0.1361888243722193,
      "grad_norm": 0.26842140212655446,
      "learning_rate": 4.797154874567656e-07,
      "loss": 1.6167,
      "step": 903
    },
    {
      "epoch": 0.13633964256089284,
      "grad_norm": 0.2571476298071355,
      "learning_rate": 4.796712250855846e-07,
      "loss": 1.5371,
      "step": 904
    },
    {
      "epoch": 0.1364904607495664,
      "grad_norm": 0.27375277855080987,
      "learning_rate": 4.796269167597529e-07,
      "loss": 1.5679,
      "step": 905
    },
    {
      "epoch": 0.13664127893823996,
      "grad_norm": 0.3057162574463642,
      "learning_rate": 4.795825624892191e-07,
      "loss": 1.6045,
      "step": 906
    },
    {
      "epoch": 0.1367920971269135,
      "grad_norm": 0.2498937260128442,
      "learning_rate": 4.795381622839418e-07,
      "loss": 1.6371,
      "step": 907
    },
    {
      "epoch": 0.13694291531558705,
      "grad_norm": 0.2862944084345334,
      "learning_rate": 4.794937161538903e-07,
      "loss": 1.5801,
      "step": 908
    },
    {
      "epoch": 0.1370937335042606,
      "grad_norm": 0.26138244883566913,
      "learning_rate": 4.794492241090441e-07,
      "loss": 1.6468,
      "step": 909
    },
    {
      "epoch": 0.13724455169293417,
      "grad_norm": 0.26629757391984443,
      "learning_rate": 4.794046861593928e-07,
      "loss": 1.5691,
      "step": 910
    },
    {
      "epoch": 0.13739536988160772,
      "grad_norm": 0.30809370729255164,
      "learning_rate": 4.793601023149366e-07,
      "loss": 1.4806,
      "step": 911
    },
    {
      "epoch": 0.13754618807028127,
      "grad_norm": 0.46536836965333095,
      "learning_rate": 4.793154725856857e-07,
      "loss": 1.5647,
      "step": 912
    },
    {
      "epoch": 0.13769700625895484,
      "grad_norm": 0.2522687711040882,
      "learning_rate": 4.79270796981661e-07,
      "loss": 1.5733,
      "step": 913
    },
    {
      "epoch": 0.1378478244476284,
      "grad_norm": 0.2690001943708058,
      "learning_rate": 4.792260755128932e-07,
      "loss": 1.59,
      "step": 914
    },
    {
      "epoch": 0.13799864263630193,
      "grad_norm": 0.24545110405555004,
      "learning_rate": 4.791813081894237e-07,
      "loss": 1.6301,
      "step": 915
    },
    {
      "epoch": 0.13814946082497548,
      "grad_norm": 0.26644633088028224,
      "learning_rate": 4.791364950213043e-07,
      "loss": 1.5873,
      "step": 916
    },
    {
      "epoch": 0.13830027901364905,
      "grad_norm": 0.251546512526234,
      "learning_rate": 4.790916360185965e-07,
      "loss": 1.5496,
      "step": 917
    },
    {
      "epoch": 0.1384510972023226,
      "grad_norm": 0.25316144412679326,
      "learning_rate": 4.790467311913727e-07,
      "loss": 1.5564,
      "step": 918
    },
    {
      "epoch": 0.13860191539099614,
      "grad_norm": 0.32773966673237986,
      "learning_rate": 4.790017805497152e-07,
      "loss": 1.5444,
      "step": 919
    },
    {
      "epoch": 0.13875273357966972,
      "grad_norm": 0.24546044641204529,
      "learning_rate": 4.789567841037168e-07,
      "loss": 1.5814,
      "step": 920
    },
    {
      "epoch": 0.13890355176834326,
      "grad_norm": 0.29277978064100285,
      "learning_rate": 4.789117418634804e-07,
      "loss": 1.5894,
      "step": 921
    },
    {
      "epoch": 0.1390543699570168,
      "grad_norm": 0.3549813518300409,
      "learning_rate": 4.788666538391196e-07,
      "loss": 1.5953,
      "step": 922
    },
    {
      "epoch": 0.13920518814569038,
      "grad_norm": 0.2685881754179385,
      "learning_rate": 4.788215200407576e-07,
      "loss": 1.5124,
      "step": 923
    },
    {
      "epoch": 0.13935600633436393,
      "grad_norm": 0.23461190696685494,
      "learning_rate": 4.787763404785284e-07,
      "loss": 1.5263,
      "step": 924
    },
    {
      "epoch": 0.13950682452303748,
      "grad_norm": 0.2769659797431462,
      "learning_rate": 4.787311151625762e-07,
      "loss": 1.6239,
      "step": 925
    },
    {
      "epoch": 0.13965764271171102,
      "grad_norm": 0.25856439910948137,
      "learning_rate": 4.786858441030553e-07,
      "loss": 1.6261,
      "step": 926
    },
    {
      "epoch": 0.1398084609003846,
      "grad_norm": 0.30055130559881404,
      "learning_rate": 4.786405273101304e-07,
      "loss": 1.5665,
      "step": 927
    },
    {
      "epoch": 0.13995927908905814,
      "grad_norm": 0.26524355572498903,
      "learning_rate": 4.785951647939765e-07,
      "loss": 1.5662,
      "step": 928
    },
    {
      "epoch": 0.1401100972777317,
      "grad_norm": 0.2615300648761517,
      "learning_rate": 4.785497565647787e-07,
      "loss": 1.6526,
      "step": 929
    },
    {
      "epoch": 0.14026091546640526,
      "grad_norm": 0.24452930953345128,
      "learning_rate": 4.785043026327327e-07,
      "loss": 1.6293,
      "step": 930
    },
    {
      "epoch": 0.1404117336550788,
      "grad_norm": 0.2509944116710881,
      "learning_rate": 4.784588030080439e-07,
      "loss": 1.5249,
      "step": 931
    },
    {
      "epoch": 0.14056255184375235,
      "grad_norm": 0.28509033083799773,
      "learning_rate": 4.784132577009284e-07,
      "loss": 1.5579,
      "step": 932
    },
    {
      "epoch": 0.1407133700324259,
      "grad_norm": 0.25327312061491164,
      "learning_rate": 4.783676667216125e-07,
      "loss": 1.5977,
      "step": 933
    },
    {
      "epoch": 0.14086418822109947,
      "grad_norm": 0.244814281253799,
      "learning_rate": 4.783220300803329e-07,
      "loss": 1.5698,
      "step": 934
    },
    {
      "epoch": 0.14101500640977302,
      "grad_norm": 0.24512065385327506,
      "learning_rate": 4.78276347787336e-07,
      "loss": 1.581,
      "step": 935
    },
    {
      "epoch": 0.14116582459844657,
      "grad_norm": 0.2407057281097465,
      "learning_rate": 4.78230619852879e-07,
      "loss": 1.573,
      "step": 936
    },
    {
      "epoch": 0.14131664278712014,
      "grad_norm": 0.2957663472009309,
      "learning_rate": 4.781848462872291e-07,
      "loss": 1.6006,
      "step": 937
    },
    {
      "epoch": 0.14146746097579369,
      "grad_norm": 0.2523508100196154,
      "learning_rate": 4.781390271006638e-07,
      "loss": 1.5686,
      "step": 938
    },
    {
      "epoch": 0.14161827916446723,
      "grad_norm": 0.26587967742482643,
      "learning_rate": 4.780931623034708e-07,
      "loss": 1.588,
      "step": 939
    },
    {
      "epoch": 0.14176909735314078,
      "grad_norm": 0.24226402477563086,
      "learning_rate": 4.780472519059482e-07,
      "loss": 1.57,
      "step": 940
    },
    {
      "epoch": 0.14191991554181435,
      "grad_norm": 0.32111658741147964,
      "learning_rate": 4.780012959184042e-07,
      "loss": 1.5811,
      "step": 941
    },
    {
      "epoch": 0.1420707337304879,
      "grad_norm": 0.2603441769850403,
      "learning_rate": 4.779552943511572e-07,
      "loss": 1.5989,
      "step": 942
    },
    {
      "epoch": 0.14222155191916144,
      "grad_norm": 0.24997121269396286,
      "learning_rate": 4.779092472145359e-07,
      "loss": 1.5094,
      "step": 943
    },
    {
      "epoch": 0.14237237010783502,
      "grad_norm": 0.24301016836612505,
      "learning_rate": 4.778631545188791e-07,
      "loss": 1.5627,
      "step": 944
    },
    {
      "epoch": 0.14252318829650856,
      "grad_norm": 0.389051844125457,
      "learning_rate": 4.778170162745363e-07,
      "loss": 1.5494,
      "step": 945
    },
    {
      "epoch": 0.1426740064851821,
      "grad_norm": 0.2547230822653501,
      "learning_rate": 4.777708324918665e-07,
      "loss": 1.5722,
      "step": 946
    },
    {
      "epoch": 0.14282482467385565,
      "grad_norm": 0.24358299425546492,
      "learning_rate": 4.777246031812395e-07,
      "loss": 1.6627,
      "step": 947
    },
    {
      "epoch": 0.14297564286252923,
      "grad_norm": 0.2658494143360813,
      "learning_rate": 4.776783283530351e-07,
      "loss": 1.6202,
      "step": 948
    },
    {
      "epoch": 0.14312646105120277,
      "grad_norm": 0.2876997672428938,
      "learning_rate": 4.776320080176433e-07,
      "loss": 1.6024,
      "step": 949
    },
    {
      "epoch": 0.14327727923987632,
      "grad_norm": 0.36983100061415125,
      "learning_rate": 4.775856421854645e-07,
      "loss": 1.5943,
      "step": 950
    },
    {
      "epoch": 0.1434280974285499,
      "grad_norm": 0.2506670735811908,
      "learning_rate": 4.775392308669091e-07,
      "loss": 1.5817,
      "step": 951
    },
    {
      "epoch": 0.14357891561722344,
      "grad_norm": 0.2371938577597567,
      "learning_rate": 4.774927740723977e-07,
      "loss": 1.5705,
      "step": 952
    },
    {
      "epoch": 0.143729733805897,
      "grad_norm": 0.23952843693509399,
      "learning_rate": 4.774462718123614e-07,
      "loss": 1.4839,
      "step": 953
    },
    {
      "epoch": 0.14388055199457053,
      "grad_norm": 0.23949570105139512,
      "learning_rate": 4.773997240972409e-07,
      "loss": 1.5263,
      "step": 954
    },
    {
      "epoch": 0.1440313701832441,
      "grad_norm": 0.2551375984425037,
      "learning_rate": 4.773531309374881e-07,
      "loss": 1.6084,
      "step": 955
    },
    {
      "epoch": 0.14418218837191765,
      "grad_norm": 0.24541644561947223,
      "learning_rate": 4.773064923435641e-07,
      "loss": 1.5743,
      "step": 956
    },
    {
      "epoch": 0.1443330065605912,
      "grad_norm": 0.25515556428521197,
      "learning_rate": 4.772598083259408e-07,
      "loss": 1.5043,
      "step": 957
    },
    {
      "epoch": 0.14448382474926477,
      "grad_norm": 0.2750279009285979,
      "learning_rate": 4.772130788951e-07,
      "loss": 1.5227,
      "step": 958
    },
    {
      "epoch": 0.14463464293793832,
      "grad_norm": 0.25507085734973817,
      "learning_rate": 4.77166304061534e-07,
      "loss": 1.5771,
      "step": 959
    },
    {
      "epoch": 0.14478546112661186,
      "grad_norm": 0.2620215918819778,
      "learning_rate": 4.771194838357449e-07,
      "loss": 1.559,
      "step": 960
    },
    {
      "epoch": 0.1449362793152854,
      "grad_norm": 0.2733625850021945,
      "learning_rate": 4.770726182282454e-07,
      "loss": 1.5437,
      "step": 961
    },
    {
      "epoch": 0.14508709750395898,
      "grad_norm": 0.28518365346685176,
      "learning_rate": 4.770257072495581e-07,
      "loss": 1.5703,
      "step": 962
    },
    {
      "epoch": 0.14523791569263253,
      "grad_norm": 0.25041762802372103,
      "learning_rate": 4.769787509102158e-07,
      "loss": 1.6308,
      "step": 963
    },
    {
      "epoch": 0.14538873388130608,
      "grad_norm": 0.25820967916794474,
      "learning_rate": 4.769317492207617e-07,
      "loss": 1.5527,
      "step": 964
    },
    {
      "epoch": 0.14553955206997965,
      "grad_norm": 0.334784271673938,
      "learning_rate": 4.76884702191749e-07,
      "loss": 1.4922,
      "step": 965
    },
    {
      "epoch": 0.1456903702586532,
      "grad_norm": 0.24810790355705478,
      "learning_rate": 4.768376098337411e-07,
      "loss": 1.5765,
      "step": 966
    },
    {
      "epoch": 0.14584118844732674,
      "grad_norm": 0.28700061399749227,
      "learning_rate": 4.767904721573117e-07,
      "loss": 1.6496,
      "step": 967
    },
    {
      "epoch": 0.14599200663600032,
      "grad_norm": 0.34732089248600745,
      "learning_rate": 4.767432891730444e-07,
      "loss": 1.5558,
      "step": 968
    },
    {
      "epoch": 0.14614282482467386,
      "grad_norm": 0.2673344082419073,
      "learning_rate": 4.766960608915333e-07,
      "loss": 1.5894,
      "step": 969
    },
    {
      "epoch": 0.1462936430133474,
      "grad_norm": 2.653973193991139,
      "learning_rate": 4.7664878732338253e-07,
      "loss": 1.5334,
      "step": 970
    },
    {
      "epoch": 0.14644446120202095,
      "grad_norm": 0.24993909864096386,
      "learning_rate": 4.7660146847920635e-07,
      "loss": 1.5961,
      "step": 971
    },
    {
      "epoch": 0.14659527939069453,
      "grad_norm": 0.2565497081102889,
      "learning_rate": 4.7655410436962916e-07,
      "loss": 1.6027,
      "step": 972
    },
    {
      "epoch": 0.14674609757936807,
      "grad_norm": 0.254279538292664,
      "learning_rate": 4.765066950052857e-07,
      "loss": 1.6422,
      "step": 973
    },
    {
      "epoch": 0.14689691576804162,
      "grad_norm": 0.2657242795261302,
      "learning_rate": 4.7645924039682066e-07,
      "loss": 1.6159,
      "step": 974
    },
    {
      "epoch": 0.1470477339567152,
      "grad_norm": 0.2551289716846203,
      "learning_rate": 4.7641174055488903e-07,
      "loss": 1.5766,
      "step": 975
    },
    {
      "epoch": 0.14719855214538874,
      "grad_norm": 0.2487164615456287,
      "learning_rate": 4.7636419549015587e-07,
      "loss": 1.6085,
      "step": 976
    },
    {
      "epoch": 0.14734937033406229,
      "grad_norm": 0.24934246137057814,
      "learning_rate": 4.763166052132964e-07,
      "loss": 1.5588,
      "step": 977
    },
    {
      "epoch": 0.14750018852273583,
      "grad_norm": 0.25608798985597103,
      "learning_rate": 4.7626896973499615e-07,
      "loss": 1.5647,
      "step": 978
    },
    {
      "epoch": 0.1476510067114094,
      "grad_norm": 0.24874879011014975,
      "learning_rate": 4.7622128906595054e-07,
      "loss": 1.5121,
      "step": 979
    },
    {
      "epoch": 0.14780182490008295,
      "grad_norm": 0.24432845222015379,
      "learning_rate": 4.761735632168653e-07,
      "loss": 1.6303,
      "step": 980
    },
    {
      "epoch": 0.1479526430887565,
      "grad_norm": 0.252842065429307,
      "learning_rate": 4.761257921984564e-07,
      "loss": 1.59,
      "step": 981
    },
    {
      "epoch": 0.14810346127743007,
      "grad_norm": 0.2514307218624979,
      "learning_rate": 4.760779760214496e-07,
      "loss": 1.6196,
      "step": 982
    },
    {
      "epoch": 0.14825427946610362,
      "grad_norm": 0.2667127561644108,
      "learning_rate": 4.7603011469658126e-07,
      "loss": 1.58,
      "step": 983
    },
    {
      "epoch": 0.14840509765477716,
      "grad_norm": 0.24168970436906176,
      "learning_rate": 4.7598220823459744e-07,
      "loss": 1.5355,
      "step": 984
    },
    {
      "epoch": 0.1485559158434507,
      "grad_norm": 0.25620873924454746,
      "learning_rate": 4.759342566462546e-07,
      "loss": 1.6269,
      "step": 985
    },
    {
      "epoch": 0.14870673403212428,
      "grad_norm": 0.2452078703443453,
      "learning_rate": 4.758862599423194e-07,
      "loss": 1.6141,
      "step": 986
    },
    {
      "epoch": 0.14885755222079783,
      "grad_norm": 0.24225881503880753,
      "learning_rate": 4.758382181335684e-07,
      "loss": 1.5921,
      "step": 987
    },
    {
      "epoch": 0.14900837040947137,
      "grad_norm": 0.2612762223581993,
      "learning_rate": 4.757901312307882e-07,
      "loss": 1.5693,
      "step": 988
    },
    {
      "epoch": 0.14915918859814495,
      "grad_norm": 0.303459353963245,
      "learning_rate": 4.7574199924477605e-07,
      "loss": 1.5342,
      "step": 989
    },
    {
      "epoch": 0.1493100067868185,
      "grad_norm": 0.3088514981989386,
      "learning_rate": 4.7569382218633877e-07,
      "loss": 1.6037,
      "step": 990
    },
    {
      "epoch": 0.14946082497549204,
      "grad_norm": 0.2562445993286538,
      "learning_rate": 4.756456000662935e-07,
      "loss": 1.5493,
      "step": 991
    },
    {
      "epoch": 0.1496116431641656,
      "grad_norm": 0.28232305406411545,
      "learning_rate": 4.755973328954676e-07,
      "loss": 1.6046,
      "step": 992
    },
    {
      "epoch": 0.14976246135283916,
      "grad_norm": 0.23837206086373763,
      "learning_rate": 4.755490206846985e-07,
      "loss": 1.5531,
      "step": 993
    },
    {
      "epoch": 0.1499132795415127,
      "grad_norm": 0.28518372943400533,
      "learning_rate": 4.7550066344483355e-07,
      "loss": 1.5578,
      "step": 994
    },
    {
      "epoch": 0.15006409773018625,
      "grad_norm": 0.2549768269904049,
      "learning_rate": 4.754522611867304e-07,
      "loss": 1.5646,
      "step": 995
    },
    {
      "epoch": 0.15021491591885983,
      "grad_norm": 0.2562706132685018,
      "learning_rate": 4.754038139212569e-07,
      "loss": 1.589,
      "step": 996
    },
    {
      "epoch": 0.15036573410753337,
      "grad_norm": 0.2410268153834008,
      "learning_rate": 4.753553216592907e-07,
      "loss": 1.4844,
      "step": 997
    },
    {
      "epoch": 0.15051655229620692,
      "grad_norm": 0.26281148717796365,
      "learning_rate": 4.7530678441171976e-07,
      "loss": 1.5636,
      "step": 998
    },
    {
      "epoch": 0.15066737048488046,
      "grad_norm": 0.30961820231400505,
      "learning_rate": 4.752582021894422e-07,
      "loss": 1.6057,
      "step": 999
    },
    {
      "epoch": 0.15081818867355404,
      "grad_norm": 0.3321593493732633,
      "learning_rate": 4.752095750033661e-07,
      "loss": 1.6641,
      "step": 1000
    },
    {
      "epoch": 0.15096900686222758,
      "grad_norm": 0.2969182772224481,
      "learning_rate": 4.7516090286440963e-07,
      "loss": 1.6061,
      "step": 1001
    },
    {
      "epoch": 0.15111982505090113,
      "grad_norm": 0.2477958419058681,
      "learning_rate": 4.751121857835012e-07,
      "loss": 1.597,
      "step": 1002
    },
    {
      "epoch": 0.1512706432395747,
      "grad_norm": 0.24983981603131114,
      "learning_rate": 4.7506342377157904e-07,
      "loss": 1.6094,
      "step": 1003
    },
    {
      "epoch": 0.15142146142824825,
      "grad_norm": 0.2543220894110329,
      "learning_rate": 4.750146168395918e-07,
      "loss": 1.5846,
      "step": 1004
    },
    {
      "epoch": 0.1515722796169218,
      "grad_norm": 0.2587477761766272,
      "learning_rate": 4.7496576499849807e-07,
      "loss": 1.56,
      "step": 1005
    },
    {
      "epoch": 0.15172309780559534,
      "grad_norm": 0.24154994402700736,
      "learning_rate": 4.749168682592664e-07,
      "loss": 1.6037,
      "step": 1006
    },
    {
      "epoch": 0.15187391599426892,
      "grad_norm": 0.2508792993278851,
      "learning_rate": 4.748679266328755e-07,
      "loss": 1.5603,
      "step": 1007
    },
    {
      "epoch": 0.15202473418294246,
      "grad_norm": 0.2493414013337419,
      "learning_rate": 4.748189401303144e-07,
      "loss": 1.5396,
      "step": 1008
    },
    {
      "epoch": 0.152175552371616,
      "grad_norm": 0.658800165472276,
      "learning_rate": 4.7476990876258185e-07,
      "loss": 1.5619,
      "step": 1009
    },
    {
      "epoch": 0.15232637056028958,
      "grad_norm": 0.23745478283721239,
      "learning_rate": 4.7472083254068686e-07,
      "loss": 1.5544,
      "step": 1010
    },
    {
      "epoch": 0.15247718874896313,
      "grad_norm": 0.305400220510485,
      "learning_rate": 4.7467171147564835e-07,
      "loss": 1.6674,
      "step": 1011
    },
    {
      "epoch": 0.15262800693763667,
      "grad_norm": 0.2785138739432395,
      "learning_rate": 4.7462254557849557e-07,
      "loss": 1.5855,
      "step": 1012
    },
    {
      "epoch": 0.15277882512631022,
      "grad_norm": 0.2434299468753599,
      "learning_rate": 4.745733348602677e-07,
      "loss": 1.6288,
      "step": 1013
    },
    {
      "epoch": 0.1529296433149838,
      "grad_norm": 0.2451373266031179,
      "learning_rate": 4.745240793320139e-07,
      "loss": 1.5515,
      "step": 1014
    },
    {
      "epoch": 0.15308046150365734,
      "grad_norm": 0.2701367378557065,
      "learning_rate": 4.744747790047935e-07,
      "loss": 1.516,
      "step": 1015
    },
    {
      "epoch": 0.15323127969233089,
      "grad_norm": 0.3157527268124156,
      "learning_rate": 4.7442543388967587e-07,
      "loss": 1.6163,
      "step": 1016
    },
    {
      "epoch": 0.15338209788100446,
      "grad_norm": 0.24232869990901512,
      "learning_rate": 4.7437604399774047e-07,
      "loss": 1.6275,
      "step": 1017
    },
    {
      "epoch": 0.153532916069678,
      "grad_norm": 0.24900214045309305,
      "learning_rate": 4.7432660934007665e-07,
      "loss": 1.5733,
      "step": 1018
    },
    {
      "epoch": 0.15368373425835155,
      "grad_norm": 0.2822160839916749,
      "learning_rate": 4.742771299277841e-07,
      "loss": 1.594,
      "step": 1019
    },
    {
      "epoch": 0.15383455244702512,
      "grad_norm": 0.2733829280739275,
      "learning_rate": 4.742276057719722e-07,
      "loss": 1.57,
      "step": 1020
    },
    {
      "epoch": 0.15398537063569867,
      "grad_norm": 0.2460228836578742,
      "learning_rate": 4.741780368837607e-07,
      "loss": 1.5509,
      "step": 1021
    },
    {
      "epoch": 0.15413618882437222,
      "grad_norm": 0.25260122809000746,
      "learning_rate": 4.7412842327427914e-07,
      "loss": 1.5806,
      "step": 1022
    },
    {
      "epoch": 0.15428700701304576,
      "grad_norm": 0.23750399144418952,
      "learning_rate": 4.7407876495466737e-07,
      "loss": 1.5771,
      "step": 1023
    },
    {
      "epoch": 0.15443782520171934,
      "grad_norm": 0.26808088140916786,
      "learning_rate": 4.7402906193607506e-07,
      "loss": 1.6318,
      "step": 1024
    },
    {
      "epoch": 0.15458864339039288,
      "grad_norm": 0.24828288210625818,
      "learning_rate": 4.73979314229662e-07,
      "loss": 1.5416,
      "step": 1025
    },
    {
      "epoch": 0.15473946157906643,
      "grad_norm": 0.2436099220209187,
      "learning_rate": 4.739295218465979e-07,
      "loss": 1.5476,
      "step": 1026
    },
    {
      "epoch": 0.15489027976774,
      "grad_norm": 0.3083906756582321,
      "learning_rate": 4.7387968479806266e-07,
      "loss": 1.5377,
      "step": 1027
    },
    {
      "epoch": 0.15504109795641355,
      "grad_norm": 0.4550389188653999,
      "learning_rate": 4.738298030952462e-07,
      "loss": 1.6148,
      "step": 1028
    },
    {
      "epoch": 0.1551919161450871,
      "grad_norm": 0.2626493565680759,
      "learning_rate": 4.737798767493484e-07,
      "loss": 1.5929,
      "step": 1029
    },
    {
      "epoch": 0.15534273433376064,
      "grad_norm": 0.25785948030370415,
      "learning_rate": 4.7372990577157917e-07,
      "loss": 1.5832,
      "step": 1030
    },
    {
      "epoch": 0.15549355252243421,
      "grad_norm": 0.2610654083785875,
      "learning_rate": 4.736798901731585e-07,
      "loss": 1.5268,
      "step": 1031
    },
    {
      "epoch": 0.15564437071110776,
      "grad_norm": 0.2810034490504235,
      "learning_rate": 4.736298299653161e-07,
      "loss": 1.577,
      "step": 1032
    },
    {
      "epoch": 0.1557951888997813,
      "grad_norm": 0.5772108063423994,
      "learning_rate": 4.7357972515929227e-07,
      "loss": 1.561,
      "step": 1033
    },
    {
      "epoch": 0.15594600708845488,
      "grad_norm": 0.25471771878994864,
      "learning_rate": 4.7352957576633683e-07,
      "loss": 1.6665,
      "step": 1034
    },
    {
      "epoch": 0.15609682527712843,
      "grad_norm": 0.25694693768929777,
      "learning_rate": 4.7347938179770985e-07,
      "loss": 1.5935,
      "step": 1035
    },
    {
      "epoch": 0.15624764346580197,
      "grad_norm": 0.3104727488339877,
      "learning_rate": 4.734291432646813e-07,
      "loss": 1.4999,
      "step": 1036
    },
    {
      "epoch": 0.15639846165447552,
      "grad_norm": 0.2975832370733511,
      "learning_rate": 4.733788601785311e-07,
      "loss": 1.5515,
      "step": 1037
    },
    {
      "epoch": 0.1565492798431491,
      "grad_norm": 0.27642272849022204,
      "learning_rate": 4.733285325505494e-07,
      "loss": 1.6036,
      "step": 1038
    },
    {
      "epoch": 0.15670009803182264,
      "grad_norm": 0.24721769073668393,
      "learning_rate": 4.7327816039203624e-07,
      "loss": 1.5687,
      "step": 1039
    },
    {
      "epoch": 0.15685091622049618,
      "grad_norm": 0.2956038050337575,
      "learning_rate": 4.732277437143015e-07,
      "loss": 1.5981,
      "step": 1040
    },
    {
      "epoch": 0.15700173440916976,
      "grad_norm": 0.2718469053700773,
      "learning_rate": 4.731772825286653e-07,
      "loss": 1.5874,
      "step": 1041
    },
    {
      "epoch": 0.1571525525978433,
      "grad_norm": 0.25197611213411847,
      "learning_rate": 4.7312677684645755e-07,
      "loss": 1.5553,
      "step": 1042
    },
    {
      "epoch": 0.15730337078651685,
      "grad_norm": 0.29647496304966825,
      "learning_rate": 4.730762266790184e-07,
      "loss": 1.5247,
      "step": 1043
    },
    {
      "epoch": 0.1574541889751904,
      "grad_norm": 0.26208550422646826,
      "learning_rate": 4.730256320376977e-07,
      "loss": 1.5189,
      "step": 1044
    },
    {
      "epoch": 0.15760500716386397,
      "grad_norm": 0.2647759645027192,
      "learning_rate": 4.7297499293385545e-07,
      "loss": 1.6032,
      "step": 1045
    },
    {
      "epoch": 0.15775582535253752,
      "grad_norm": 0.31315008832451746,
      "learning_rate": 4.729243093788617e-07,
      "loss": 1.6115,
      "step": 1046
    },
    {
      "epoch": 0.15790664354121106,
      "grad_norm": 0.2924108623243284,
      "learning_rate": 4.728735813840962e-07,
      "loss": 1.6144,
      "step": 1047
    },
    {
      "epoch": 0.15805746172988464,
      "grad_norm": 0.2616373957240464,
      "learning_rate": 4.72822808960949e-07,
      "loss": 1.5285,
      "step": 1048
    },
    {
      "epoch": 0.15820827991855818,
      "grad_norm": 0.24923023915333234,
      "learning_rate": 4.7277199212082e-07,
      "loss": 1.6548,
      "step": 1049
    },
    {
      "epoch": 0.15835909810723173,
      "grad_norm": 0.2551933398409467,
      "learning_rate": 4.727211308751189e-07,
      "loss": 1.6165,
      "step": 1050
    },
    {
      "epoch": 0.15850991629590527,
      "grad_norm": 0.2593180448603392,
      "learning_rate": 4.726702252352657e-07,
      "loss": 1.5991,
      "step": 1051
    },
    {
      "epoch": 0.15866073448457885,
      "grad_norm": 0.2522277006422934,
      "learning_rate": 4.726192752126902e-07,
      "loss": 1.577,
      "step": 1052
    },
    {
      "epoch": 0.1588115526732524,
      "grad_norm": 0.24196721815033448,
      "learning_rate": 4.72568280818832e-07,
      "loss": 1.5836,
      "step": 1053
    },
    {
      "epoch": 0.15896237086192594,
      "grad_norm": 0.3618317509241435,
      "learning_rate": 4.72517242065141e-07,
      "loss": 1.5203,
      "step": 1054
    },
    {
      "epoch": 0.1591131890505995,
      "grad_norm": 0.25454939085154465,
      "learning_rate": 4.724661589630768e-07,
      "loss": 1.5784,
      "step": 1055
    },
    {
      "epoch": 0.15926400723927306,
      "grad_norm": 0.3109584087928939,
      "learning_rate": 4.7241503152410907e-07,
      "loss": 1.5591,
      "step": 1056
    },
    {
      "epoch": 0.1594148254279466,
      "grad_norm": 0.7106449826146803,
      "learning_rate": 4.723638597597174e-07,
      "loss": 1.5349,
      "step": 1057
    },
    {
      "epoch": 0.15956564361662015,
      "grad_norm": 0.25701180952300506,
      "learning_rate": 4.7231264368139134e-07,
      "loss": 1.5358,
      "step": 1058
    },
    {
      "epoch": 0.15971646180529372,
      "grad_norm": 0.2617384036431046,
      "learning_rate": 4.7226138330063037e-07,
      "loss": 1.6136,
      "step": 1059
    },
    {
      "epoch": 0.15986727999396727,
      "grad_norm": 0.248514514370961,
      "learning_rate": 4.72210078628944e-07,
      "loss": 1.5699,
      "step": 1060
    },
    {
      "epoch": 0.16001809818264082,
      "grad_norm": 0.4566482953295286,
      "learning_rate": 4.7215872967785157e-07,
      "loss": 1.4869,
      "step": 1061
    },
    {
      "epoch": 0.1601689163713144,
      "grad_norm": 0.24761995402178832,
      "learning_rate": 4.7210733645888246e-07,
      "loss": 1.6156,
      "step": 1062
    },
    {
      "epoch": 0.16031973455998794,
      "grad_norm": 0.250070974124865,
      "learning_rate": 4.720558989835758e-07,
      "loss": 1.5712,
      "step": 1063
    },
    {
      "epoch": 0.16047055274866148,
      "grad_norm": 0.2786032458759379,
      "learning_rate": 4.720044172634811e-07,
      "loss": 1.583,
      "step": 1064
    },
    {
      "epoch": 0.16062137093733506,
      "grad_norm": 0.23855863462524987,
      "learning_rate": 4.719528913101572e-07,
      "loss": 1.5799,
      "step": 1065
    },
    {
      "epoch": 0.1607721891260086,
      "grad_norm": 0.32775340260671904,
      "learning_rate": 4.719013211351733e-07,
      "loss": 1.5525,
      "step": 1066
    },
    {
      "epoch": 0.16092300731468215,
      "grad_norm": 0.2502199020901237,
      "learning_rate": 4.718497067501084e-07,
      "loss": 1.5793,
      "step": 1067
    },
    {
      "epoch": 0.1610738255033557,
      "grad_norm": 0.2769500848160153,
      "learning_rate": 4.717980481665515e-07,
      "loss": 1.578,
      "step": 1068
    },
    {
      "epoch": 0.16122464369202927,
      "grad_norm": 0.2811566565102702,
      "learning_rate": 4.7174634539610135e-07,
      "loss": 1.5422,
      "step": 1069
    },
    {
      "epoch": 0.16137546188070281,
      "grad_norm": 0.26690201580123935,
      "learning_rate": 4.7169459845036685e-07,
      "loss": 1.5062,
      "step": 1070
    },
    {
      "epoch": 0.16152628006937636,
      "grad_norm": 0.24278929565756457,
      "learning_rate": 4.716428073409665e-07,
      "loss": 1.5764,
      "step": 1071
    },
    {
      "epoch": 0.16167709825804993,
      "grad_norm": 0.26380707944919146,
      "learning_rate": 4.7159097207952907e-07,
      "loss": 1.5759,
      "step": 1072
    },
    {
      "epoch": 0.16182791644672348,
      "grad_norm": 0.2401141242040836,
      "learning_rate": 4.715390926776931e-07,
      "loss": 1.553,
      "step": 1073
    },
    {
      "epoch": 0.16197873463539703,
      "grad_norm": 0.29678926738553935,
      "learning_rate": 4.7148716914710696e-07,
      "loss": 1.5737,
      "step": 1074
    },
    {
      "epoch": 0.16212955282407057,
      "grad_norm": 0.24725309737718767,
      "learning_rate": 4.7143520149942894e-07,
      "loss": 1.5078,
      "step": 1075
    },
    {
      "epoch": 0.16228037101274415,
      "grad_norm": 0.25803787562110775,
      "learning_rate": 4.713831897463274e-07,
      "loss": 1.5073,
      "step": 1076
    },
    {
      "epoch": 0.1624311892014177,
      "grad_norm": 0.26833123309240886,
      "learning_rate": 4.7133113389948044e-07,
      "loss": 1.5977,
      "step": 1077
    },
    {
      "epoch": 0.16258200739009124,
      "grad_norm": 0.24091943366927152,
      "learning_rate": 4.712790339705761e-07,
      "loss": 1.5915,
      "step": 1078
    },
    {
      "epoch": 0.1627328255787648,
      "grad_norm": 0.2454951827903659,
      "learning_rate": 4.7122688997131244e-07,
      "loss": 1.5016,
      "step": 1079
    },
    {
      "epoch": 0.16288364376743836,
      "grad_norm": 0.327938893492718,
      "learning_rate": 4.711747019133972e-07,
      "loss": 1.5499,
      "step": 1080
    },
    {
      "epoch": 0.1630344619561119,
      "grad_norm": 0.2709383069557141,
      "learning_rate": 4.71122469808548e-07,
      "loss": 1.5399,
      "step": 1081
    },
    {
      "epoch": 0.16318528014478545,
      "grad_norm": 0.2581676243463674,
      "learning_rate": 4.7107019366849276e-07,
      "loss": 1.6087,
      "step": 1082
    },
    {
      "epoch": 0.16333609833345902,
      "grad_norm": 0.3892574896237313,
      "learning_rate": 4.710178735049688e-07,
      "loss": 1.5812,
      "step": 1083
    },
    {
      "epoch": 0.16348691652213257,
      "grad_norm": 0.270203428554805,
      "learning_rate": 4.709655093297236e-07,
      "loss": 1.5764,
      "step": 1084
    },
    {
      "epoch": 0.16363773471080612,
      "grad_norm": 0.28517256024517534,
      "learning_rate": 4.7091310115451437e-07,
      "loss": 1.6103,
      "step": 1085
    },
    {
      "epoch": 0.1637885528994797,
      "grad_norm": 0.27840669121440514,
      "learning_rate": 4.7086064899110837e-07,
      "loss": 1.5659,
      "step": 1086
    },
    {
      "epoch": 0.16393937108815324,
      "grad_norm": 0.2461045836784305,
      "learning_rate": 4.708081528512825e-07,
      "loss": 1.6039,
      "step": 1087
    },
    {
      "epoch": 0.16409018927682678,
      "grad_norm": 0.7459017336078355,
      "learning_rate": 4.707556127468238e-07,
      "loss": 1.6061,
      "step": 1088
    },
    {
      "epoch": 0.16424100746550033,
      "grad_norm": 0.2561621447257331,
      "learning_rate": 4.70703028689529e-07,
      "loss": 1.5939,
      "step": 1089
    },
    {
      "epoch": 0.1643918256541739,
      "grad_norm": 0.2849028517073751,
      "learning_rate": 4.7065040069120474e-07,
      "loss": 1.5602,
      "step": 1090
    },
    {
      "epoch": 0.16454264384284745,
      "grad_norm": 0.24979491525986122,
      "learning_rate": 4.7059772876366755e-07,
      "loss": 1.5781,
      "step": 1091
    },
    {
      "epoch": 0.164693462031521,
      "grad_norm": 0.24168740791476312,
      "learning_rate": 4.705450129187438e-07,
      "loss": 1.5414,
      "step": 1092
    },
    {
      "epoch": 0.16484428022019457,
      "grad_norm": 0.24173643436887862,
      "learning_rate": 4.7049225316826986e-07,
      "loss": 1.4875,
      "step": 1093
    },
    {
      "epoch": 0.1649950984088681,
      "grad_norm": 0.3580945143021086,
      "learning_rate": 4.704394495240916e-07,
      "loss": 1.5554,
      "step": 1094
    },
    {
      "epoch": 0.16514591659754166,
      "grad_norm": 0.27131705981391885,
      "learning_rate": 4.7038660199806514e-07,
      "loss": 1.5368,
      "step": 1095
    },
    {
      "epoch": 0.1652967347862152,
      "grad_norm": 0.2525411683704331,
      "learning_rate": 4.7033371060205626e-07,
      "loss": 1.5617,
      "step": 1096
    },
    {
      "epoch": 0.16544755297488878,
      "grad_norm": 0.2426665694554285,
      "learning_rate": 4.702807753479406e-07,
      "loss": 1.5246,
      "step": 1097
    },
    {
      "epoch": 0.16559837116356232,
      "grad_norm": 0.25346540257856687,
      "learning_rate": 4.702277962476036e-07,
      "loss": 1.5931,
      "step": 1098
    },
    {
      "epoch": 0.16574918935223587,
      "grad_norm": 0.24568700514771802,
      "learning_rate": 4.7017477331294075e-07,
      "loss": 1.6475,
      "step": 1099
    },
    {
      "epoch": 0.16590000754090944,
      "grad_norm": 0.2506897867451099,
      "learning_rate": 4.701217065558572e-07,
      "loss": 1.6134,
      "step": 1100
    },
    {
      "epoch": 0.166050825729583,
      "grad_norm": 0.24960037855627876,
      "learning_rate": 4.7006859598826787e-07,
      "loss": 1.5192,
      "step": 1101
    },
    {
      "epoch": 0.16620164391825654,
      "grad_norm": 0.2545680815817411,
      "learning_rate": 4.700154416220978e-07,
      "loss": 1.6317,
      "step": 1102
    },
    {
      "epoch": 0.16635246210693008,
      "grad_norm": 0.2876698360184517,
      "learning_rate": 4.699622434692816e-07,
      "loss": 1.551,
      "step": 1103
    },
    {
      "epoch": 0.16650328029560366,
      "grad_norm": 0.28949239161364276,
      "learning_rate": 4.6990900154176374e-07,
      "loss": 1.5751,
      "step": 1104
    },
    {
      "epoch": 0.1666540984842772,
      "grad_norm": 0.27888008836701456,
      "learning_rate": 4.698557158514987e-07,
      "loss": 1.5704,
      "step": 1105
    },
    {
      "epoch": 0.16680491667295075,
      "grad_norm": 0.25306042514402155,
      "learning_rate": 4.6980238641045063e-07,
      "loss": 1.6322,
      "step": 1106
    },
    {
      "epoch": 0.16695573486162432,
      "grad_norm": 0.27416420160692634,
      "learning_rate": 4.6974901323059344e-07,
      "loss": 1.5494,
      "step": 1107
    },
    {
      "epoch": 0.16710655305029787,
      "grad_norm": 0.27633683131886433,
      "learning_rate": 4.6969559632391126e-07,
      "loss": 1.5587,
      "step": 1108
    },
    {
      "epoch": 0.16725737123897141,
      "grad_norm": 0.23453200202259325,
      "learning_rate": 4.6964213570239734e-07,
      "loss": 1.5099,
      "step": 1109
    },
    {
      "epoch": 0.167408189427645,
      "grad_norm": 0.25539051826846015,
      "learning_rate": 4.6958863137805555e-07,
      "loss": 1.5002,
      "step": 1110
    },
    {
      "epoch": 0.16755900761631853,
      "grad_norm": 0.24623451238381902,
      "learning_rate": 4.6953508336289884e-07,
      "loss": 1.5422,
      "step": 1111
    },
    {
      "epoch": 0.16770982580499208,
      "grad_norm": 0.24411710831121344,
      "learning_rate": 4.6948149166895047e-07,
      "loss": 1.5739,
      "step": 1112
    },
    {
      "epoch": 0.16786064399366563,
      "grad_norm": 0.7545924295706974,
      "learning_rate": 4.694278563082432e-07,
      "loss": 1.5068,
      "step": 1113
    },
    {
      "epoch": 0.1680114621823392,
      "grad_norm": 0.24330881130548362,
      "learning_rate": 4.693741772928199e-07,
      "loss": 1.538,
      "step": 1114
    },
    {
      "epoch": 0.16816228037101275,
      "grad_norm": 0.25616718301837377,
      "learning_rate": 4.6932045463473296e-07,
      "loss": 1.6199,
      "step": 1115
    },
    {
      "epoch": 0.1683130985596863,
      "grad_norm": 0.247945338635809,
      "learning_rate": 4.6926668834604485e-07,
      "loss": 1.5797,
      "step": 1116
    },
    {
      "epoch": 0.16846391674835987,
      "grad_norm": 0.2753322662696697,
      "learning_rate": 4.692128784388274e-07,
      "loss": 1.5547,
      "step": 1117
    },
    {
      "epoch": 0.1686147349370334,
      "grad_norm": 0.2896067230267288,
      "learning_rate": 4.6915902492516264e-07,
      "loss": 1.6004,
      "step": 1118
    },
    {
      "epoch": 0.16876555312570696,
      "grad_norm": 0.2941113608129385,
      "learning_rate": 4.6910512781714227e-07,
      "loss": 1.5139,
      "step": 1119
    },
    {
      "epoch": 0.1689163713143805,
      "grad_norm": 0.25923390975079347,
      "learning_rate": 4.6905118712686775e-07,
      "loss": 1.532,
      "step": 1120
    },
    {
      "epoch": 0.16906718950305408,
      "grad_norm": 0.24243194337292528,
      "learning_rate": 4.689972028664503e-07,
      "loss": 1.5278,
      "step": 1121
    },
    {
      "epoch": 0.16921800769172762,
      "grad_norm": 0.2608954624809244,
      "learning_rate": 4.689431750480111e-07,
      "loss": 1.5048,
      "step": 1122
    },
    {
      "epoch": 0.16936882588040117,
      "grad_norm": 0.24984681694708277,
      "learning_rate": 4.688891036836807e-07,
      "loss": 1.5663,
      "step": 1123
    },
    {
      "epoch": 0.16951964406907474,
      "grad_norm": 0.3197763053893994,
      "learning_rate": 4.6883498878559996e-07,
      "loss": 1.5756,
      "step": 1124
    },
    {
      "epoch": 0.1696704622577483,
      "grad_norm": 0.24561609481331534,
      "learning_rate": 4.68780830365919e-07,
      "loss": 1.5683,
      "step": 1125
    },
    {
      "epoch": 0.16982128044642184,
      "grad_norm": 0.28356897070283404,
      "learning_rate": 4.687266284367982e-07,
      "loss": 1.5617,
      "step": 1126
    },
    {
      "epoch": 0.16997209863509538,
      "grad_norm": 0.2818725478575507,
      "learning_rate": 4.686723830104073e-07,
      "loss": 1.5588,
      "step": 1127
    },
    {
      "epoch": 0.17012291682376895,
      "grad_norm": 0.24702190052737416,
      "learning_rate": 4.686180940989262e-07,
      "loss": 1.65,
      "step": 1128
    },
    {
      "epoch": 0.1702737350124425,
      "grad_norm": 0.235138614450344,
      "learning_rate": 4.6856376171454407e-07,
      "loss": 1.5549,
      "step": 1129
    },
    {
      "epoch": 0.17042455320111605,
      "grad_norm": 0.2628161495605817,
      "learning_rate": 4.685093858694603e-07,
      "loss": 1.5571,
      "step": 1130
    },
    {
      "epoch": 0.17057537138978962,
      "grad_norm": 0.3027638920075395,
      "learning_rate": 4.684549665758838e-07,
      "loss": 1.5786,
      "step": 1131
    },
    {
      "epoch": 0.17072618957846317,
      "grad_norm": 0.25499204449953544,
      "learning_rate": 4.684005038460332e-07,
      "loss": 1.552,
      "step": 1132
    },
    {
      "epoch": 0.1708770077671367,
      "grad_norm": 0.2709751281073486,
      "learning_rate": 4.6834599769213706e-07,
      "loss": 1.6273,
      "step": 1133
    },
    {
      "epoch": 0.17102782595581026,
      "grad_norm": 0.24112103680979602,
      "learning_rate": 4.682914481264336e-07,
      "loss": 1.595,
      "step": 1134
    },
    {
      "epoch": 0.17117864414448383,
      "grad_norm": 0.24147697799095597,
      "learning_rate": 4.682368551611707e-07,
      "loss": 1.588,
      "step": 1135
    },
    {
      "epoch": 0.17132946233315738,
      "grad_norm": 0.2804421696394345,
      "learning_rate": 4.6818221880860624e-07,
      "loss": 1.5457,
      "step": 1136
    },
    {
      "epoch": 0.17148028052183092,
      "grad_norm": 0.2744210161984986,
      "learning_rate": 4.6812753908100755e-07,
      "loss": 1.5371,
      "step": 1137
    },
    {
      "epoch": 0.1716310987105045,
      "grad_norm": 0.262449872729529,
      "learning_rate": 4.680728159906518e-07,
      "loss": 1.5695,
      "step": 1138
    },
    {
      "epoch": 0.17178191689917804,
      "grad_norm": 0.24442008725733502,
      "learning_rate": 4.680180495498259e-07,
      "loss": 1.6474,
      "step": 1139
    },
    {
      "epoch": 0.1719327350878516,
      "grad_norm": 0.2704581218611592,
      "learning_rate": 4.6796323977082673e-07,
      "loss": 1.5642,
      "step": 1140
    },
    {
      "epoch": 0.17208355327652514,
      "grad_norm": 0.25987740785314206,
      "learning_rate": 4.679083866659604e-07,
      "loss": 1.5364,
      "step": 1141
    },
    {
      "epoch": 0.1722343714651987,
      "grad_norm": 0.25998785665147006,
      "learning_rate": 4.6785349024754316e-07,
      "loss": 1.5816,
      "step": 1142
    },
    {
      "epoch": 0.17238518965387226,
      "grad_norm": 0.25789553553819966,
      "learning_rate": 4.677985505279009e-07,
      "loss": 1.5617,
      "step": 1143
    },
    {
      "epoch": 0.1725360078425458,
      "grad_norm": 0.28434801761107686,
      "learning_rate": 4.677435675193691e-07,
      "loss": 1.5721,
      "step": 1144
    },
    {
      "epoch": 0.17268682603121938,
      "grad_norm": 0.23661054651052396,
      "learning_rate": 4.6768854123429315e-07,
      "loss": 1.5955,
      "step": 1145
    },
    {
      "epoch": 0.17283764421989292,
      "grad_norm": 0.2629669952322758,
      "learning_rate": 4.676334716850279e-07,
      "loss": 1.6193,
      "step": 1146
    },
    {
      "epoch": 0.17298846240856647,
      "grad_norm": 0.26201173668106353,
      "learning_rate": 4.675783588839381e-07,
      "loss": 1.5898,
      "step": 1147
    },
    {
      "epoch": 0.17313928059724001,
      "grad_norm": 0.24778101315185597,
      "learning_rate": 4.675232028433983e-07,
      "loss": 1.5255,
      "step": 1148
    },
    {
      "epoch": 0.1732900987859136,
      "grad_norm": 0.24200358612342673,
      "learning_rate": 4.674680035757925e-07,
      "loss": 1.5684,
      "step": 1149
    },
    {
      "epoch": 0.17344091697458713,
      "grad_norm": 0.25224952718930654,
      "learning_rate": 4.6741276109351475e-07,
      "loss": 1.5593,
      "step": 1150
    },
    {
      "epoch": 0.17359173516326068,
      "grad_norm": 0.3384173809607694,
      "learning_rate": 4.6735747540896834e-07,
      "loss": 1.5833,
      "step": 1151
    },
    {
      "epoch": 0.17374255335193425,
      "grad_norm": 0.2524479053804888,
      "learning_rate": 4.673021465345667e-07,
      "loss": 1.5164,
      "step": 1152
    },
    {
      "epoch": 0.1738933715406078,
      "grad_norm": 0.28287920840389524,
      "learning_rate": 4.6724677448273266e-07,
      "loss": 1.5673,
      "step": 1153
    },
    {
      "epoch": 0.17404418972928135,
      "grad_norm": 0.2591593514038448,
      "learning_rate": 4.6719135926589894e-07,
      "loss": 1.6015,
      "step": 1154
    },
    {
      "epoch": 0.1741950079179549,
      "grad_norm": 0.24488990280488418,
      "learning_rate": 4.671359008965077e-07,
      "loss": 1.5585,
      "step": 1155
    },
    {
      "epoch": 0.17434582610662847,
      "grad_norm": 0.4787738584719552,
      "learning_rate": 4.6708039938701125e-07,
      "loss": 1.5247,
      "step": 1156
    },
    {
      "epoch": 0.174496644295302,
      "grad_norm": 0.25065181476230003,
      "learning_rate": 4.6702485474987106e-07,
      "loss": 1.5832,
      "step": 1157
    },
    {
      "epoch": 0.17464746248397556,
      "grad_norm": 0.2562710623733999,
      "learning_rate": 4.669692669975587e-07,
      "loss": 1.5467,
      "step": 1158
    },
    {
      "epoch": 0.17479828067264913,
      "grad_norm": 0.25990482455141345,
      "learning_rate": 4.6691363614255495e-07,
      "loss": 1.5911,
      "step": 1159
    },
    {
      "epoch": 0.17494909886132268,
      "grad_norm": 1.1721035851483081,
      "learning_rate": 4.6685796219735085e-07,
      "loss": 1.6056,
      "step": 1160
    },
    {
      "epoch": 0.17509991704999622,
      "grad_norm": 0.4569980453387458,
      "learning_rate": 4.6680224517444677e-07,
      "loss": 1.536,
      "step": 1161
    },
    {
      "epoch": 0.1752507352386698,
      "grad_norm": 0.26157794262597983,
      "learning_rate": 4.6674648508635265e-07,
      "loss": 1.5472,
      "step": 1162
    },
    {
      "epoch": 0.17540155342734334,
      "grad_norm": 0.5498314042925766,
      "learning_rate": 4.6669068194558845e-07,
      "loss": 1.5841,
      "step": 1163
    },
    {
      "epoch": 0.1755523716160169,
      "grad_norm": 0.25302157568776995,
      "learning_rate": 4.666348357646835e-07,
      "loss": 1.5322,
      "step": 1164
    },
    {
      "epoch": 0.17570318980469044,
      "grad_norm": 0.2614995286773213,
      "learning_rate": 4.665789465561768e-07,
      "loss": 1.6134,
      "step": 1165
    },
    {
      "epoch": 0.175854007993364,
      "grad_norm": 0.26952547735276366,
      "learning_rate": 4.665230143326172e-07,
      "loss": 1.552,
      "step": 1166
    },
    {
      "epoch": 0.17600482618203755,
      "grad_norm": 0.24225918451469886,
      "learning_rate": 4.6646703910656323e-07,
      "loss": 1.5364,
      "step": 1167
    },
    {
      "epoch": 0.1761556443707111,
      "grad_norm": 0.25182051284504964,
      "learning_rate": 4.664110208905828e-07,
      "loss": 1.6144,
      "step": 1168
    },
    {
      "epoch": 0.17630646255938467,
      "grad_norm": 0.2730450933795206,
      "learning_rate": 4.663549596972536e-07,
      "loss": 1.6151,
      "step": 1169
    },
    {
      "epoch": 0.17645728074805822,
      "grad_norm": 0.2557698545135914,
      "learning_rate": 4.6629885553916317e-07,
      "loss": 1.5636,
      "step": 1170
    },
    {
      "epoch": 0.17660809893673177,
      "grad_norm": 0.2650047668822396,
      "learning_rate": 4.6624270842890837e-07,
      "loss": 1.4867,
      "step": 1171
    },
    {
      "epoch": 0.1767589171254053,
      "grad_norm": 0.2763224194988131,
      "learning_rate": 4.6618651837909594e-07,
      "loss": 1.5732,
      "step": 1172
    },
    {
      "epoch": 0.1769097353140789,
      "grad_norm": 0.24415437934266487,
      "learning_rate": 4.661302854023422e-07,
      "loss": 1.5438,
      "step": 1173
    },
    {
      "epoch": 0.17706055350275243,
      "grad_norm": 0.2499587713726338,
      "learning_rate": 4.6607400951127303e-07,
      "loss": 1.6016,
      "step": 1174
    },
    {
      "epoch": 0.17721137169142598,
      "grad_norm": 0.25085110818024176,
      "learning_rate": 4.6601769071852404e-07,
      "loss": 1.5574,
      "step": 1175
    },
    {
      "epoch": 0.17736218988009955,
      "grad_norm": 0.28681072029014526,
      "learning_rate": 4.6596132903674045e-07,
      "loss": 1.6581,
      "step": 1176
    },
    {
      "epoch": 0.1775130080687731,
      "grad_norm": 0.24192430815712648,
      "learning_rate": 4.6590492447857703e-07,
      "loss": 1.5414,
      "step": 1177
    },
    {
      "epoch": 0.17766382625744664,
      "grad_norm": 0.24901022969070188,
      "learning_rate": 4.6584847705669836e-07,
      "loss": 1.6126,
      "step": 1178
    },
    {
      "epoch": 0.1778146444461202,
      "grad_norm": 0.2693770304622494,
      "learning_rate": 4.657919867837785e-07,
      "loss": 1.6803,
      "step": 1179
    },
    {
      "epoch": 0.17796546263479376,
      "grad_norm": 0.27253275460786325,
      "learning_rate": 4.657354536725011e-07,
      "loss": 1.5254,
      "step": 1180
    },
    {
      "epoch": 0.1781162808234673,
      "grad_norm": 0.28893955719910797,
      "learning_rate": 4.656788777355594e-07,
      "loss": 1.6278,
      "step": 1181
    },
    {
      "epoch": 0.17826709901214086,
      "grad_norm": 0.25047897466701746,
      "learning_rate": 4.656222589856566e-07,
      "loss": 1.6276,
      "step": 1182
    },
    {
      "epoch": 0.17841791720081443,
      "grad_norm": 0.4119551108041056,
      "learning_rate": 4.6556559743550503e-07,
      "loss": 1.5875,
      "step": 1183
    },
    {
      "epoch": 0.17856873538948798,
      "grad_norm": 0.25089661197311985,
      "learning_rate": 4.6550889309782706e-07,
      "loss": 1.6185,
      "step": 1184
    },
    {
      "epoch": 0.17871955357816152,
      "grad_norm": 0.26179055051036315,
      "learning_rate": 4.6545214598535425e-07,
      "loss": 1.6112,
      "step": 1185
    },
    {
      "epoch": 0.17887037176683507,
      "grad_norm": 0.26197715582258846,
      "learning_rate": 4.653953561108282e-07,
      "loss": 1.5322,
      "step": 1186
    },
    {
      "epoch": 0.17902118995550864,
      "grad_norm": 0.26613377733800503,
      "learning_rate": 4.653385234869997e-07,
      "loss": 1.5794,
      "step": 1187
    },
    {
      "epoch": 0.1791720081441822,
      "grad_norm": 0.27556089428854624,
      "learning_rate": 4.652816481266294e-07,
      "loss": 1.5722,
      "step": 1188
    },
    {
      "epoch": 0.17932282633285573,
      "grad_norm": 0.24865359633810788,
      "learning_rate": 4.6522473004248753e-07,
      "loss": 1.6073,
      "step": 1189
    },
    {
      "epoch": 0.1794736445215293,
      "grad_norm": 0.257044327015407,
      "learning_rate": 4.6516776924735377e-07,
      "loss": 1.5689,
      "step": 1190
    },
    {
      "epoch": 0.17962446271020285,
      "grad_norm": 0.26973739450249506,
      "learning_rate": 4.651107657540176e-07,
      "loss": 1.6174,
      "step": 1191
    },
    {
      "epoch": 0.1797752808988764,
      "grad_norm": 0.2868057332040489,
      "learning_rate": 4.650537195752778e-07,
      "loss": 1.6308,
      "step": 1192
    },
    {
      "epoch": 0.17992609908754995,
      "grad_norm": 0.3133909161724188,
      "learning_rate": 4.6499663072394297e-07,
      "loss": 1.6096,
      "step": 1193
    },
    {
      "epoch": 0.18007691727622352,
      "grad_norm": 0.24387235006379238,
      "learning_rate": 4.649394992128313e-07,
      "loss": 1.6036,
      "step": 1194
    },
    {
      "epoch": 0.18022773546489707,
      "grad_norm": 0.25947348393779196,
      "learning_rate": 4.648823250547703e-07,
      "loss": 1.5708,
      "step": 1195
    },
    {
      "epoch": 0.1803785536535706,
      "grad_norm": 0.2523203292864254,
      "learning_rate": 4.648251082625975e-07,
      "loss": 1.5055,
      "step": 1196
    },
    {
      "epoch": 0.18052937184224419,
      "grad_norm": 0.2652017675976741,
      "learning_rate": 4.647678488491594e-07,
      "loss": 1.5491,
      "step": 1197
    },
    {
      "epoch": 0.18068019003091773,
      "grad_norm": 0.26212973358226843,
      "learning_rate": 4.6471054682731265e-07,
      "loss": 1.5581,
      "step": 1198
    },
    {
      "epoch": 0.18083100821959128,
      "grad_norm": 0.25981702383470334,
      "learning_rate": 4.6465320220992313e-07,
      "loss": 1.5524,
      "step": 1199
    },
    {
      "epoch": 0.18098182640826482,
      "grad_norm": 0.2526846428134517,
      "learning_rate": 4.6459581500986643e-07,
      "loss": 1.5485,
      "step": 1200
    },
    {
      "epoch": 0.1811326445969384,
      "grad_norm": 0.26353334465494704,
      "learning_rate": 4.6453838524002766e-07,
      "loss": 1.5731,
      "step": 1201
    },
    {
      "epoch": 0.18128346278561194,
      "grad_norm": 0.243742854481381,
      "learning_rate": 4.644809129133014e-07,
      "loss": 1.5511,
      "step": 1202
    },
    {
      "epoch": 0.1814342809742855,
      "grad_norm": 0.26479887844214733,
      "learning_rate": 4.644233980425919e-07,
      "loss": 1.6034,
      "step": 1203
    },
    {
      "epoch": 0.18158509916295906,
      "grad_norm": 0.2691086005875682,
      "learning_rate": 4.643658406408129e-07,
      "loss": 1.6087,
      "step": 1204
    },
    {
      "epoch": 0.1817359173516326,
      "grad_norm": 0.2697227597209044,
      "learning_rate": 4.643082407208878e-07,
      "loss": 1.5873,
      "step": 1205
    },
    {
      "epoch": 0.18188673554030615,
      "grad_norm": 0.24146757915863593,
      "learning_rate": 4.642505982957493e-07,
      "loss": 1.5692,
      "step": 1206
    },
    {
      "epoch": 0.18203755372897973,
      "grad_norm": 0.2532703897938818,
      "learning_rate": 4.6419291337834e-07,
      "loss": 1.5492,
      "step": 1207
    },
    {
      "epoch": 0.18218837191765327,
      "grad_norm": 1.700524666935398,
      "learning_rate": 4.641351859816117e-07,
      "loss": 1.5766,
      "step": 1208
    },
    {
      "epoch": 0.18233919010632682,
      "grad_norm": 0.25003078506309195,
      "learning_rate": 4.6407741611852586e-07,
      "loss": 1.6277,
      "step": 1209
    },
    {
      "epoch": 0.18249000829500037,
      "grad_norm": 0.25690734785468633,
      "learning_rate": 4.640196038020536e-07,
      "loss": 1.5871,
      "step": 1210
    },
    {
      "epoch": 0.18264082648367394,
      "grad_norm": 0.24526265142746212,
      "learning_rate": 4.6396174904517536e-07,
      "loss": 1.5454,
      "step": 1211
    },
    {
      "epoch": 0.1827916446723475,
      "grad_norm": 0.29939366930004707,
      "learning_rate": 4.639038518608814e-07,
      "loss": 1.5375,
      "step": 1212
    },
    {
      "epoch": 0.18294246286102103,
      "grad_norm": 0.25046227529852316,
      "learning_rate": 4.6384591226217116e-07,
      "loss": 1.6298,
      "step": 1213
    },
    {
      "epoch": 0.1830932810496946,
      "grad_norm": 0.45612339086244286,
      "learning_rate": 4.6378793026205384e-07,
      "loss": 1.5488,
      "step": 1214
    },
    {
      "epoch": 0.18324409923836815,
      "grad_norm": 0.24891897568860974,
      "learning_rate": 4.6372990587354805e-07,
      "loss": 1.5294,
      "step": 1215
    },
    {
      "epoch": 0.1833949174270417,
      "grad_norm": 0.24876130723764878,
      "learning_rate": 4.6367183910968195e-07,
      "loss": 1.6293,
      "step": 1216
    },
    {
      "epoch": 0.18354573561571524,
      "grad_norm": 0.2574961763836179,
      "learning_rate": 4.636137299834932e-07,
      "loss": 1.5755,
      "step": 1217
    },
    {
      "epoch": 0.18369655380438882,
      "grad_norm": 0.24504201176849688,
      "learning_rate": 4.6355557850802914e-07,
      "loss": 1.5819,
      "step": 1218
    },
    {
      "epoch": 0.18384737199306236,
      "grad_norm": 0.25803038007649165,
      "learning_rate": 4.6349738469634624e-07,
      "loss": 1.52,
      "step": 1219
    },
    {
      "epoch": 0.1839981901817359,
      "grad_norm": 0.37056793420273376,
      "learning_rate": 4.634391485615109e-07,
      "loss": 1.5206,
      "step": 1220
    },
    {
      "epoch": 0.18414900837040948,
      "grad_norm": 0.23740715237675503,
      "learning_rate": 4.633808701165988e-07,
      "loss": 1.519,
      "step": 1221
    },
    {
      "epoch": 0.18429982655908303,
      "grad_norm": 0.256097921253572,
      "learning_rate": 4.63322549374695e-07,
      "loss": 1.6017,
      "step": 1222
    },
    {
      "epoch": 0.18445064474775658,
      "grad_norm": 0.24119311605941288,
      "learning_rate": 4.632641863488944e-07,
      "loss": 1.5682,
      "step": 1223
    },
    {
      "epoch": 0.18460146293643012,
      "grad_norm": 0.2398564783834201,
      "learning_rate": 4.63205781052301e-07,
      "loss": 1.5638,
      "step": 1224
    },
    {
      "epoch": 0.1847522811251037,
      "grad_norm": 0.3588221629085808,
      "learning_rate": 4.6314733349802867e-07,
      "loss": 1.5388,
      "step": 1225
    },
    {
      "epoch": 0.18490309931377724,
      "grad_norm": 0.4161033662354953,
      "learning_rate": 4.630888436992005e-07,
      "loss": 1.639,
      "step": 1226
    },
    {
      "epoch": 0.1850539175024508,
      "grad_norm": 0.24893733198183843,
      "learning_rate": 4.6303031166894914e-07,
      "loss": 1.5986,
      "step": 1227
    },
    {
      "epoch": 0.18520473569112436,
      "grad_norm": 0.25661111198037884,
      "learning_rate": 4.629717374204168e-07,
      "loss": 1.5917,
      "step": 1228
    },
    {
      "epoch": 0.1853555538797979,
      "grad_norm": 0.23584522799787572,
      "learning_rate": 4.6291312096675507e-07,
      "loss": 1.5144,
      "step": 1229
    },
    {
      "epoch": 0.18550637206847145,
      "grad_norm": 0.35466238153603863,
      "learning_rate": 4.6285446232112494e-07,
      "loss": 1.6056,
      "step": 1230
    },
    {
      "epoch": 0.185657190257145,
      "grad_norm": 0.2761629325029236,
      "learning_rate": 4.6279576149669717e-07,
      "loss": 1.6149,
      "step": 1231
    },
    {
      "epoch": 0.18580800844581857,
      "grad_norm": 0.24320364119864546,
      "learning_rate": 4.6273701850665176e-07,
      "loss": 1.5697,
      "step": 1232
    },
    {
      "epoch": 0.18595882663449212,
      "grad_norm": 0.28039155604174815,
      "learning_rate": 4.6267823336417817e-07,
      "loss": 1.5901,
      "step": 1233
    },
    {
      "epoch": 0.18610964482316567,
      "grad_norm": 0.3016051065041896,
      "learning_rate": 4.626194060824754e-07,
      "loss": 1.5678,
      "step": 1234
    },
    {
      "epoch": 0.18626046301183924,
      "grad_norm": 0.3978296304643922,
      "learning_rate": 4.6256053667475187e-07,
      "loss": 1.6408,
      "step": 1235
    },
    {
      "epoch": 0.18641128120051279,
      "grad_norm": 0.2618642345392666,
      "learning_rate": 4.625016251542255e-07,
      "loss": 1.5403,
      "step": 1236
    },
    {
      "epoch": 0.18656209938918633,
      "grad_norm": 0.28061797524066334,
      "learning_rate": 4.6244267153412363e-07,
      "loss": 1.6287,
      "step": 1237
    },
    {
      "epoch": 0.18671291757785988,
      "grad_norm": 0.2539669193134728,
      "learning_rate": 4.6238367582768304e-07,
      "loss": 1.571,
      "step": 1238
    },
    {
      "epoch": 0.18686373576653345,
      "grad_norm": 0.2510040514153634,
      "learning_rate": 4.6232463804815003e-07,
      "loss": 1.5547,
      "step": 1239
    },
    {
      "epoch": 0.187014553955207,
      "grad_norm": 0.25538695923375226,
      "learning_rate": 4.622655582087803e-07,
      "loss": 1.5653,
      "step": 1240
    },
    {
      "epoch": 0.18716537214388054,
      "grad_norm": 0.34510024150041224,
      "learning_rate": 4.62206436322839e-07,
      "loss": 1.6462,
      "step": 1241
    },
    {
      "epoch": 0.18731619033255412,
      "grad_norm": 0.25761472217381176,
      "learning_rate": 4.6214727240360055e-07,
      "loss": 1.6034,
      "step": 1242
    },
    {
      "epoch": 0.18746700852122766,
      "grad_norm": 0.2602442266228183,
      "learning_rate": 4.6208806646434923e-07,
      "loss": 1.6195,
      "step": 1243
    },
    {
      "epoch": 0.1876178267099012,
      "grad_norm": 0.3121342337827262,
      "learning_rate": 4.620288185183784e-07,
      "loss": 1.5173,
      "step": 1244
    },
    {
      "epoch": 0.18776864489857475,
      "grad_norm": 0.3112698054707776,
      "learning_rate": 4.619695285789909e-07,
      "loss": 1.5601,
      "step": 1245
    },
    {
      "epoch": 0.18791946308724833,
      "grad_norm": 0.6375963358679859,
      "learning_rate": 4.619101966594991e-07,
      "loss": 1.5594,
      "step": 1246
    },
    {
      "epoch": 0.18807028127592187,
      "grad_norm": 0.24807667638603587,
      "learning_rate": 4.6185082277322465e-07,
      "loss": 1.5797,
      "step": 1247
    },
    {
      "epoch": 0.18822109946459542,
      "grad_norm": 0.2747542330598371,
      "learning_rate": 4.617914069334989e-07,
      "loss": 1.5427,
      "step": 1248
    },
    {
      "epoch": 0.188371917653269,
      "grad_norm": 0.24848149709894782,
      "learning_rate": 4.617319491536623e-07,
      "loss": 1.5991,
      "step": 1249
    },
    {
      "epoch": 0.18852273584194254,
      "grad_norm": 0.25170883005078193,
      "learning_rate": 4.616724494470648e-07,
      "loss": 1.6421,
      "step": 1250
    },
    {
      "epoch": 0.1886735540306161,
      "grad_norm": 0.2679738449785433,
      "learning_rate": 4.616129078270659e-07,
      "loss": 1.5997,
      "step": 1251
    },
    {
      "epoch": 0.18882437221928966,
      "grad_norm": 0.2577440112764784,
      "learning_rate": 4.6155332430703456e-07,
      "loss": 1.5468,
      "step": 1252
    },
    {
      "epoch": 0.1889751904079632,
      "grad_norm": 0.2444515358662243,
      "learning_rate": 4.614936989003487e-07,
      "loss": 1.606,
      "step": 1253
    },
    {
      "epoch": 0.18912600859663675,
      "grad_norm": 0.3533797144661275,
      "learning_rate": 4.614340316203963e-07,
      "loss": 1.5285,
      "step": 1254
    },
    {
      "epoch": 0.1892768267853103,
      "grad_norm": 0.24621271282013607,
      "learning_rate": 4.613743224805741e-07,
      "loss": 1.5976,
      "step": 1255
    },
    {
      "epoch": 0.18942764497398387,
      "grad_norm": 0.25113303023350053,
      "learning_rate": 4.613145714942888e-07,
      "loss": 1.5832,
      "step": 1256
    },
    {
      "epoch": 0.18957846316265742,
      "grad_norm": 0.25694879956941913,
      "learning_rate": 4.6125477867495603e-07,
      "loss": 1.5284,
      "step": 1257
    },
    {
      "epoch": 0.18972928135133096,
      "grad_norm": 0.2501853930979283,
      "learning_rate": 4.6119494403600123e-07,
      "loss": 1.6257,
      "step": 1258
    },
    {
      "epoch": 0.18988009954000454,
      "grad_norm": 0.2657571741616329,
      "learning_rate": 4.611350675908588e-07,
      "loss": 1.55,
      "step": 1259
    },
    {
      "epoch": 0.19003091772867808,
      "grad_norm": 0.24410102087236463,
      "learning_rate": 4.6107514935297285e-07,
      "loss": 1.5801,
      "step": 1260
    },
    {
      "epoch": 0.19018173591735163,
      "grad_norm": 0.24922626928952427,
      "learning_rate": 4.6101518933579674e-07,
      "loss": 1.5757,
      "step": 1261
    },
    {
      "epoch": 0.19033255410602518,
      "grad_norm": 0.2651313594595526,
      "learning_rate": 4.609551875527933e-07,
      "loss": 1.6483,
      "step": 1262
    },
    {
      "epoch": 0.19048337229469875,
      "grad_norm": 0.25817997842493584,
      "learning_rate": 4.6089514401743466e-07,
      "loss": 1.5612,
      "step": 1263
    },
    {
      "epoch": 0.1906341904833723,
      "grad_norm": 0.3206507643473013,
      "learning_rate": 4.608350587432023e-07,
      "loss": 1.5033,
      "step": 1264
    },
    {
      "epoch": 0.19078500867204584,
      "grad_norm": 0.23999627909004798,
      "learning_rate": 4.6077493174358715e-07,
      "loss": 1.5554,
      "step": 1265
    },
    {
      "epoch": 0.19093582686071942,
      "grad_norm": 0.24365895893821485,
      "learning_rate": 4.607147630320894e-07,
      "loss": 1.5166,
      "step": 1266
    },
    {
      "epoch": 0.19108664504939296,
      "grad_norm": 0.25204093453281395,
      "learning_rate": 4.6065455262221884e-07,
      "loss": 1.5883,
      "step": 1267
    },
    {
      "epoch": 0.1912374632380665,
      "grad_norm": 0.24931428905056033,
      "learning_rate": 4.605943005274943e-07,
      "loss": 1.4928,
      "step": 1268
    },
    {
      "epoch": 0.19138828142674005,
      "grad_norm": 0.27467672421431577,
      "learning_rate": 4.605340067614442e-07,
      "loss": 1.5958,
      "step": 1269
    },
    {
      "epoch": 0.19153909961541363,
      "grad_norm": 0.24682085106543605,
      "learning_rate": 4.6047367133760623e-07,
      "loss": 1.5171,
      "step": 1270
    },
    {
      "epoch": 0.19168991780408717,
      "grad_norm": 0.2553685144848191,
      "learning_rate": 4.604132942695275e-07,
      "loss": 1.5606,
      "step": 1271
    },
    {
      "epoch": 0.19184073599276072,
      "grad_norm": 0.2537348129260827,
      "learning_rate": 4.603528755707642e-07,
      "loss": 1.5597,
      "step": 1272
    },
    {
      "epoch": 0.1919915541814343,
      "grad_norm": 0.25380935336276694,
      "learning_rate": 4.6029241525488246e-07,
      "loss": 1.521,
      "step": 1273
    },
    {
      "epoch": 0.19214237237010784,
      "grad_norm": 0.25771573527963487,
      "learning_rate": 4.6023191333545707e-07,
      "loss": 1.5038,
      "step": 1274
    },
    {
      "epoch": 0.19229319055878139,
      "grad_norm": 0.24367657069761992,
      "learning_rate": 4.6017136982607265e-07,
      "loss": 1.5537,
      "step": 1275
    },
    {
      "epoch": 0.19244400874745493,
      "grad_norm": 0.252004375801695,
      "learning_rate": 4.601107847403229e-07,
      "loss": 1.5619,
      "step": 1276
    },
    {
      "epoch": 0.1925948269361285,
      "grad_norm": 0.24426652256256345,
      "learning_rate": 4.6005015809181094e-07,
      "loss": 1.6252,
      "step": 1277
    },
    {
      "epoch": 0.19274564512480205,
      "grad_norm": 0.24547818197943544,
      "learning_rate": 4.5998948989414934e-07,
      "loss": 1.5503,
      "step": 1278
    },
    {
      "epoch": 0.1928964633134756,
      "grad_norm": 0.24177262746334108,
      "learning_rate": 4.599287801609596e-07,
      "loss": 1.5638,
      "step": 1279
    },
    {
      "epoch": 0.19304728150214917,
      "grad_norm": 0.2613304195476814,
      "learning_rate": 4.5986802890587304e-07,
      "loss": 1.5172,
      "step": 1280
    },
    {
      "epoch": 0.19319809969082272,
      "grad_norm": 0.25227676166352114,
      "learning_rate": 4.598072361425301e-07,
      "loss": 1.5577,
      "step": 1281
    },
    {
      "epoch": 0.19334891787949626,
      "grad_norm": 0.257336503929896,
      "learning_rate": 4.5974640188458043e-07,
      "loss": 1.5726,
      "step": 1282
    },
    {
      "epoch": 0.1934997360681698,
      "grad_norm": 0.25307676147124225,
      "learning_rate": 4.596855261456831e-07,
      "loss": 1.572,
      "step": 1283
    },
    {
      "epoch": 0.19365055425684338,
      "grad_norm": 0.3182845915121795,
      "learning_rate": 4.596246089395066e-07,
      "loss": 1.6012,
      "step": 1284
    },
    {
      "epoch": 0.19380137244551693,
      "grad_norm": 0.27830626585061713,
      "learning_rate": 4.595636502797284e-07,
      "loss": 1.6253,
      "step": 1285
    },
    {
      "epoch": 0.19395219063419047,
      "grad_norm": 0.2625158798232183,
      "learning_rate": 4.595026501800357e-07,
      "loss": 1.5688,
      "step": 1286
    },
    {
      "epoch": 0.19410300882286405,
      "grad_norm": 0.257794216429428,
      "learning_rate": 4.5944160865412474e-07,
      "loss": 1.5646,
      "step": 1287
    },
    {
      "epoch": 0.1942538270115376,
      "grad_norm": 0.24249718273407012,
      "learning_rate": 4.593805257157011e-07,
      "loss": 1.568,
      "step": 1288
    },
    {
      "epoch": 0.19440464520021114,
      "grad_norm": 0.2717055951556258,
      "learning_rate": 4.5931940137847964e-07,
      "loss": 1.5762,
      "step": 1289
    },
    {
      "epoch": 0.1945554633888847,
      "grad_norm": 0.251473646820601,
      "learning_rate": 4.592582356561846e-07,
      "loss": 1.6225,
      "step": 1290
    },
    {
      "epoch": 0.19470628157755826,
      "grad_norm": 0.2530523964470091,
      "learning_rate": 4.5919702856254947e-07,
      "loss": 1.4989,
      "step": 1291
    },
    {
      "epoch": 0.1948570997662318,
      "grad_norm": 0.24090357138386617,
      "learning_rate": 4.5913578011131703e-07,
      "loss": 1.5378,
      "step": 1292
    },
    {
      "epoch": 0.19500791795490535,
      "grad_norm": 0.2529652593382991,
      "learning_rate": 4.5907449031623936e-07,
      "loss": 1.6045,
      "step": 1293
    },
    {
      "epoch": 0.19515873614357893,
      "grad_norm": 0.24554505644207056,
      "learning_rate": 4.5901315919107777e-07,
      "loss": 1.5979,
      "step": 1294
    },
    {
      "epoch": 0.19530955433225247,
      "grad_norm": 0.2462215351501105,
      "learning_rate": 4.589517867496028e-07,
      "loss": 1.6437,
      "step": 1295
    },
    {
      "epoch": 0.19546037252092602,
      "grad_norm": 0.25811189385629774,
      "learning_rate": 4.588903730055945e-07,
      "loss": 1.5753,
      "step": 1296
    },
    {
      "epoch": 0.19561119070959956,
      "grad_norm": 0.26564361953917165,
      "learning_rate": 4.5882891797284193e-07,
      "loss": 1.4945,
      "step": 1297
    },
    {
      "epoch": 0.19576200889827314,
      "grad_norm": 0.26828284814969955,
      "learning_rate": 4.587674216651437e-07,
      "loss": 1.5596,
      "step": 1298
    },
    {
      "epoch": 0.19591282708694668,
      "grad_norm": 0.2518716180925786,
      "learning_rate": 4.587058840963073e-07,
      "loss": 1.5467,
      "step": 1299
    },
    {
      "epoch": 0.19606364527562023,
      "grad_norm": 0.24883212949633934,
      "learning_rate": 4.5864430528014994e-07,
      "loss": 1.5684,
      "step": 1300
    },
    {
      "epoch": 0.1962144634642938,
      "grad_norm": 0.5995044319510145,
      "learning_rate": 4.585826852304976e-07,
      "loss": 1.6714,
      "step": 1301
    },
    {
      "epoch": 0.19636528165296735,
      "grad_norm": 0.2528807116883068,
      "learning_rate": 4.585210239611859e-07,
      "loss": 1.49,
      "step": 1302
    },
    {
      "epoch": 0.1965160998416409,
      "grad_norm": 0.2484612567828872,
      "learning_rate": 4.584593214860597e-07,
      "loss": 1.5581,
      "step": 1303
    },
    {
      "epoch": 0.19666691803031447,
      "grad_norm": 0.2520265580853553,
      "learning_rate": 4.583975778189728e-07,
      "loss": 1.5489,
      "step": 1304
    },
    {
      "epoch": 0.19681773621898802,
      "grad_norm": 0.24330409597101163,
      "learning_rate": 4.583357929737886e-07,
      "loss": 1.5461,
      "step": 1305
    },
    {
      "epoch": 0.19696855440766156,
      "grad_norm": 0.2450801329502804,
      "learning_rate": 4.582739669643796e-07,
      "loss": 1.542,
      "step": 1306
    },
    {
      "epoch": 0.1971193725963351,
      "grad_norm": 0.24670229706809432,
      "learning_rate": 4.5821209980462737e-07,
      "loss": 1.5479,
      "step": 1307
    },
    {
      "epoch": 0.19727019078500868,
      "grad_norm": 0.265775400162428,
      "learning_rate": 4.581501915084231e-07,
      "loss": 1.6353,
      "step": 1308
    },
    {
      "epoch": 0.19742100897368223,
      "grad_norm": 0.2451961552667391,
      "learning_rate": 4.580882420896669e-07,
      "loss": 1.624,
      "step": 1309
    },
    {
      "epoch": 0.19757182716235577,
      "grad_norm": 0.24175200633283164,
      "learning_rate": 4.580262515622682e-07,
      "loss": 1.5441,
      "step": 1310
    },
    {
      "epoch": 0.19772264535102935,
      "grad_norm": 0.266058836376841,
      "learning_rate": 4.579642199401457e-07,
      "loss": 1.5569,
      "step": 1311
    },
    {
      "epoch": 0.1978734635397029,
      "grad_norm": 0.24367772871318977,
      "learning_rate": 4.5790214723722735e-07,
      "loss": 1.5444,
      "step": 1312
    },
    {
      "epoch": 0.19802428172837644,
      "grad_norm": 0.2669581598378095,
      "learning_rate": 4.5784003346745026e-07,
      "loss": 1.5663,
      "step": 1313
    },
    {
      "epoch": 0.19817509991704998,
      "grad_norm": 0.27868587914293996,
      "learning_rate": 4.5777787864476073e-07,
      "loss": 1.5606,
      "step": 1314
    },
    {
      "epoch": 0.19832591810572356,
      "grad_norm": 0.2531823451378717,
      "learning_rate": 4.577156827831144e-07,
      "loss": 1.5303,
      "step": 1315
    },
    {
      "epoch": 0.1984767362943971,
      "grad_norm": 0.26456851904105433,
      "learning_rate": 4.57653445896476e-07,
      "loss": 1.5881,
      "step": 1316
    },
    {
      "epoch": 0.19862755448307065,
      "grad_norm": 0.24625701695281055,
      "learning_rate": 4.575911679988195e-07,
      "loss": 1.5938,
      "step": 1317
    },
    {
      "epoch": 0.19877837267174422,
      "grad_norm": 0.2705797450523775,
      "learning_rate": 4.575288491041282e-07,
      "loss": 1.5614,
      "step": 1318
    },
    {
      "epoch": 0.19892919086041777,
      "grad_norm": 0.3298997867142449,
      "learning_rate": 4.5746648922639444e-07,
      "loss": 1.5986,
      "step": 1319
    },
    {
      "epoch": 0.19908000904909132,
      "grad_norm": 0.260750951121751,
      "learning_rate": 4.574040883796198e-07,
      "loss": 1.5308,
      "step": 1320
    },
    {
      "epoch": 0.19923082723776486,
      "grad_norm": 0.25912363854098036,
      "learning_rate": 4.5734164657781515e-07,
      "loss": 1.542,
      "step": 1321
    },
    {
      "epoch": 0.19938164542643844,
      "grad_norm": 0.25176792843992035,
      "learning_rate": 4.5727916383500043e-07,
      "loss": 1.56,
      "step": 1322
    },
    {
      "epoch": 0.19953246361511198,
      "grad_norm": 0.2843038062909802,
      "learning_rate": 4.572166401652049e-07,
      "loss": 1.5762,
      "step": 1323
    },
    {
      "epoch": 0.19968328180378553,
      "grad_norm": 0.3716912797456824,
      "learning_rate": 4.571540755824669e-07,
      "loss": 1.535,
      "step": 1324
    },
    {
      "epoch": 0.1998340999924591,
      "grad_norm": 0.2509797256223773,
      "learning_rate": 4.570914701008341e-07,
      "loss": 1.494,
      "step": 1325
    },
    {
      "epoch": 0.19998491818113265,
      "grad_norm": 0.24197474440474273,
      "learning_rate": 4.5702882373436314e-07,
      "loss": 1.6091,
      "step": 1326
    },
    {
      "epoch": 0.2001357363698062,
      "grad_norm": 0.27322255667328965,
      "learning_rate": 4.5696613649711994e-07,
      "loss": 1.5679,
      "step": 1327
    },
    {
      "epoch": 0.20028655455847974,
      "grad_norm": 0.2574972467885414,
      "learning_rate": 4.569034084031797e-07,
      "loss": 1.5417,
      "step": 1328
    },
    {
      "epoch": 0.20043737274715331,
      "grad_norm": 0.2476096783459321,
      "learning_rate": 4.5684063946662663e-07,
      "loss": 1.5555,
      "step": 1329
    },
    {
      "epoch": 0.20058819093582686,
      "grad_norm": 0.2543742745683159,
      "learning_rate": 4.5677782970155425e-07,
      "loss": 1.561,
      "step": 1330
    },
    {
      "epoch": 0.2007390091245004,
      "grad_norm": 0.2588741933427589,
      "learning_rate": 4.567149791220651e-07,
      "loss": 1.5036,
      "step": 1331
    },
    {
      "epoch": 0.20088982731317398,
      "grad_norm": 0.24985352822264073,
      "learning_rate": 4.56652087742271e-07,
      "loss": 1.551,
      "step": 1332
    },
    {
      "epoch": 0.20104064550184753,
      "grad_norm": 1.0017865312383014,
      "learning_rate": 4.56589155576293e-07,
      "loss": 1.5412,
      "step": 1333
    },
    {
      "epoch": 0.20119146369052107,
      "grad_norm": 0.24359154642673883,
      "learning_rate": 4.56526182638261e-07,
      "loss": 1.5443,
      "step": 1334
    },
    {
      "epoch": 0.20134228187919462,
      "grad_norm": 0.25702345630740436,
      "learning_rate": 4.5646316894231447e-07,
      "loss": 1.6313,
      "step": 1335
    },
    {
      "epoch": 0.2014931000678682,
      "grad_norm": 0.25280312061914906,
      "learning_rate": 4.5640011450260174e-07,
      "loss": 1.5685,
      "step": 1336
    },
    {
      "epoch": 0.20164391825654174,
      "grad_norm": 0.28617712038991966,
      "learning_rate": 4.5633701933328026e-07,
      "loss": 1.5567,
      "step": 1337
    },
    {
      "epoch": 0.20179473644521528,
      "grad_norm": 0.2513625985066753,
      "learning_rate": 4.562738834485168e-07,
      "loss": 1.5928,
      "step": 1338
    },
    {
      "epoch": 0.20194555463388886,
      "grad_norm": 0.2583391580247105,
      "learning_rate": 4.5621070686248733e-07,
      "loss": 1.6059,
      "step": 1339
    },
    {
      "epoch": 0.2020963728225624,
      "grad_norm": 0.43483098079460936,
      "learning_rate": 4.561474895893767e-07,
      "loss": 1.5168,
      "step": 1340
    },
    {
      "epoch": 0.20224719101123595,
      "grad_norm": 0.2652926044977916,
      "learning_rate": 4.56084231643379e-07,
      "loss": 1.5608,
      "step": 1341
    },
    {
      "epoch": 0.2023980091999095,
      "grad_norm": 0.24995728221583127,
      "learning_rate": 4.5602093303869757e-07,
      "loss": 1.6096,
      "step": 1342
    },
    {
      "epoch": 0.20254882738858307,
      "grad_norm": 0.24987711335700388,
      "learning_rate": 4.5595759378954473e-07,
      "loss": 1.5823,
      "step": 1343
    },
    {
      "epoch": 0.20269964557725662,
      "grad_norm": 0.33269507721172,
      "learning_rate": 4.558942139101421e-07,
      "loss": 1.5985,
      "step": 1344
    },
    {
      "epoch": 0.20285046376593016,
      "grad_norm": 0.24662871345387807,
      "learning_rate": 4.558307934147202e-07,
      "loss": 1.597,
      "step": 1345
    },
    {
      "epoch": 0.20300128195460374,
      "grad_norm": 0.25277950197978016,
      "learning_rate": 4.557673323175188e-07,
      "loss": 1.6153,
      "step": 1346
    },
    {
      "epoch": 0.20315210014327728,
      "grad_norm": 0.25791182532812484,
      "learning_rate": 4.557038306327867e-07,
      "loss": 1.5121,
      "step": 1347
    },
    {
      "epoch": 0.20330291833195083,
      "grad_norm": 0.3057793602717895,
      "learning_rate": 4.5564028837478196e-07,
      "loss": 1.5631,
      "step": 1348
    },
    {
      "epoch": 0.2034537365206244,
      "grad_norm": 0.24552413290619537,
      "learning_rate": 4.5557670555777174e-07,
      "loss": 1.5521,
      "step": 1349
    },
    {
      "epoch": 0.20360455470929795,
      "grad_norm": 0.2430317899241359,
      "learning_rate": 4.5551308219603205e-07,
      "loss": 1.5716,
      "step": 1350
    },
    {
      "epoch": 0.2037553728979715,
      "grad_norm": 0.2505642792908967,
      "learning_rate": 4.5544941830384837e-07,
      "loss": 1.5883,
      "step": 1351
    },
    {
      "epoch": 0.20390619108664504,
      "grad_norm": 0.2659737294912798,
      "learning_rate": 4.5538571389551496e-07,
      "loss": 1.5785,
      "step": 1352
    },
    {
      "epoch": 0.2040570092753186,
      "grad_norm": 0.3284157122003667,
      "learning_rate": 4.5532196898533545e-07,
      "loss": 1.564,
      "step": 1353
    },
    {
      "epoch": 0.20420782746399216,
      "grad_norm": 0.2492202242651036,
      "learning_rate": 4.5525818358762227e-07,
      "loss": 1.5739,
      "step": 1354
    },
    {
      "epoch": 0.2043586456526657,
      "grad_norm": 0.24554436767224772,
      "learning_rate": 4.551943577166972e-07,
      "loss": 1.5291,
      "step": 1355
    },
    {
      "epoch": 0.20450946384133928,
      "grad_norm": 0.27509833380478466,
      "learning_rate": 4.5513049138689094e-07,
      "loss": 1.5675,
      "step": 1356
    },
    {
      "epoch": 0.20466028203001282,
      "grad_norm": 0.2542901920012441,
      "learning_rate": 4.5506658461254337e-07,
      "loss": 1.5154,
      "step": 1357
    },
    {
      "epoch": 0.20481110021868637,
      "grad_norm": 0.2491163040776305,
      "learning_rate": 4.5500263740800356e-07,
      "loss": 1.6534,
      "step": 1358
    },
    {
      "epoch": 0.20496191840735992,
      "grad_norm": 0.25258441943888266,
      "learning_rate": 4.5493864978762925e-07,
      "loss": 1.5954,
      "step": 1359
    },
    {
      "epoch": 0.2051127365960335,
      "grad_norm": 0.3901512693311432,
      "learning_rate": 4.548746217657878e-07,
      "loss": 1.5297,
      "step": 1360
    },
    {
      "epoch": 0.20526355478470704,
      "grad_norm": 0.26586117055017966,
      "learning_rate": 4.5481055335685517e-07,
      "loss": 1.5197,
      "step": 1361
    },
    {
      "epoch": 0.20541437297338058,
      "grad_norm": 0.24156859903027042,
      "learning_rate": 4.547464445752166e-07,
      "loss": 1.602,
      "step": 1362
    },
    {
      "epoch": 0.20556519116205416,
      "grad_norm": 0.288296122042675,
      "learning_rate": 4.546822954352664e-07,
      "loss": 1.5139,
      "step": 1363
    },
    {
      "epoch": 0.2057160093507277,
      "grad_norm": 0.27571681432133305,
      "learning_rate": 4.546181059514081e-07,
      "loss": 1.6343,
      "step": 1364
    },
    {
      "epoch": 0.20586682753940125,
      "grad_norm": 0.26007909076181435,
      "learning_rate": 4.545538761380539e-07,
      "loss": 1.5911,
      "step": 1365
    },
    {
      "epoch": 0.2060176457280748,
      "grad_norm": 0.25431514373043707,
      "learning_rate": 4.5448960600962525e-07,
      "loss": 1.5176,
      "step": 1366
    },
    {
      "epoch": 0.20616846391674837,
      "grad_norm": 0.25302059966266965,
      "learning_rate": 4.544252955805528e-07,
      "loss": 1.5248,
      "step": 1367
    },
    {
      "epoch": 0.20631928210542191,
      "grad_norm": 0.24800911294965144,
      "learning_rate": 4.5436094486527606e-07,
      "loss": 1.5458,
      "step": 1368
    },
    {
      "epoch": 0.20647010029409546,
      "grad_norm": 0.24503612947593298,
      "learning_rate": 4.542965538782436e-07,
      "loss": 1.5657,
      "step": 1369
    },
    {
      "epoch": 0.20662091848276903,
      "grad_norm": 0.2658313769865448,
      "learning_rate": 4.5423212263391314e-07,
      "loss": 1.5449,
      "step": 1370
    },
    {
      "epoch": 0.20677173667144258,
      "grad_norm": 0.26167357035395145,
      "learning_rate": 4.541676511467513e-07,
      "loss": 1.542,
      "step": 1371
    },
    {
      "epoch": 0.20692255486011613,
      "grad_norm": 0.27849759106683974,
      "learning_rate": 4.541031394312338e-07,
      "loss": 1.4999,
      "step": 1372
    },
    {
      "epoch": 0.20707337304878967,
      "grad_norm": 0.2630077162730558,
      "learning_rate": 4.540385875018455e-07,
      "loss": 1.4677,
      "step": 1373
    },
    {
      "epoch": 0.20722419123746325,
      "grad_norm": 0.2518083771440917,
      "learning_rate": 4.539739953730801e-07,
      "loss": 1.5819,
      "step": 1374
    },
    {
      "epoch": 0.2073750094261368,
      "grad_norm": 0.23858991358920195,
      "learning_rate": 4.5390936305944036e-07,
      "loss": 1.5444,
      "step": 1375
    },
    {
      "epoch": 0.20752582761481034,
      "grad_norm": 0.2731108237392616,
      "learning_rate": 4.5384469057543836e-07,
      "loss": 1.5505,
      "step": 1376
    },
    {
      "epoch": 0.2076766458034839,
      "grad_norm": 0.23783912526258008,
      "learning_rate": 4.537799779355947e-07,
      "loss": 1.5205,
      "step": 1377
    },
    {
      "epoch": 0.20782746399215746,
      "grad_norm": 0.26817501745430766,
      "learning_rate": 4.537152251544393e-07,
      "loss": 1.555,
      "step": 1378
    },
    {
      "epoch": 0.207978282180831,
      "grad_norm": 0.31211585887200594,
      "learning_rate": 4.5365043224651117e-07,
      "loss": 1.578,
      "step": 1379
    },
    {
      "epoch": 0.20812910036950455,
      "grad_norm": 0.3369507015026358,
      "learning_rate": 4.5358559922635806e-07,
      "loss": 1.5843,
      "step": 1380
    },
    {
      "epoch": 0.20827991855817812,
      "grad_norm": 0.2468479801615849,
      "learning_rate": 4.5352072610853704e-07,
      "loss": 1.5821,
      "step": 1381
    },
    {
      "epoch": 0.20843073674685167,
      "grad_norm": 0.9366577594549299,
      "learning_rate": 4.534558129076138e-07,
      "loss": 1.5913,
      "step": 1382
    },
    {
      "epoch": 0.20858155493552522,
      "grad_norm": 0.2636202380918008,
      "learning_rate": 4.533908596381634e-07,
      "loss": 1.5372,
      "step": 1383
    },
    {
      "epoch": 0.2087323731241988,
      "grad_norm": 0.2512541914978473,
      "learning_rate": 4.5332586631476976e-07,
      "loss": 1.6358,
      "step": 1384
    },
    {
      "epoch": 0.20888319131287233,
      "grad_norm": 0.2633196962037907,
      "learning_rate": 4.532608329520256e-07,
      "loss": 1.5599,
      "step": 1385
    },
    {
      "epoch": 0.20903400950154588,
      "grad_norm": 0.24785371601498915,
      "learning_rate": 4.53195759564533e-07,
      "loss": 1.7057,
      "step": 1386
    },
    {
      "epoch": 0.20918482769021943,
      "grad_norm": 0.25019900140013046,
      "learning_rate": 4.531306461669027e-07,
      "loss": 1.5962,
      "step": 1387
    },
    {
      "epoch": 0.209335645878893,
      "grad_norm": 0.2633845347548641,
      "learning_rate": 4.530654927737546e-07,
      "loss": 1.6358,
      "step": 1388
    },
    {
      "epoch": 0.20948646406756655,
      "grad_norm": 0.2744402674265749,
      "learning_rate": 4.5300029939971757e-07,
      "loss": 1.5347,
      "step": 1389
    },
    {
      "epoch": 0.2096372822562401,
      "grad_norm": 0.24882511692788956,
      "learning_rate": 4.5293506605942934e-07,
      "loss": 1.5777,
      "step": 1390
    },
    {
      "epoch": 0.20978810044491367,
      "grad_norm": 0.25709926694621,
      "learning_rate": 4.528697927675367e-07,
      "loss": 1.5273,
      "step": 1391
    },
    {
      "epoch": 0.2099389186335872,
      "grad_norm": 0.26502976739542417,
      "learning_rate": 4.528044795386955e-07,
      "loss": 1.5539,
      "step": 1392
    },
    {
      "epoch": 0.21008973682226076,
      "grad_norm": 0.2520151211005096,
      "learning_rate": 4.5273912638757043e-07,
      "loss": 1.5474,
      "step": 1393
    },
    {
      "epoch": 0.21024055501093433,
      "grad_norm": 0.2386463581533624,
      "learning_rate": 4.5267373332883506e-07,
      "loss": 1.576,
      "step": 1394
    },
    {
      "epoch": 0.21039137319960788,
      "grad_norm": 0.24899071043384616,
      "learning_rate": 4.526083003771722e-07,
      "loss": 1.5386,
      "step": 1395
    },
    {
      "epoch": 0.21054219138828142,
      "grad_norm": 0.2770613415174753,
      "learning_rate": 4.5254282754727337e-07,
      "loss": 1.6591,
      "step": 1396
    },
    {
      "epoch": 0.21069300957695497,
      "grad_norm": 0.25022938621669893,
      "learning_rate": 4.524773148538391e-07,
      "loss": 1.5791,
      "step": 1397
    },
    {
      "epoch": 0.21084382776562854,
      "grad_norm": 0.27685533621155517,
      "learning_rate": 4.52411762311579e-07,
      "loss": 1.5734,
      "step": 1398
    },
    {
      "epoch": 0.2109946459543021,
      "grad_norm": 0.25338547389921595,
      "learning_rate": 4.523461699352115e-07,
      "loss": 1.5792,
      "step": 1399
    },
    {
      "epoch": 0.21114546414297564,
      "grad_norm": 0.24111278814349393,
      "learning_rate": 4.522805377394638e-07,
      "loss": 1.513,
      "step": 1400
    },
    {
      "epoch": 0.2112962823316492,
      "grad_norm": 0.24924887701465612,
      "learning_rate": 4.522148657390725e-07,
      "loss": 1.6394,
      "step": 1401
    },
    {
      "epoch": 0.21144710052032276,
      "grad_norm": 0.2564505137840794,
      "learning_rate": 4.521491539487827e-07,
      "loss": 1.5265,
      "step": 1402
    },
    {
      "epoch": 0.2115979187089963,
      "grad_norm": 0.2926745104258473,
      "learning_rate": 4.5208340238334876e-07,
      "loss": 1.5805,
      "step": 1403
    },
    {
      "epoch": 0.21174873689766985,
      "grad_norm": 0.2494023731334625,
      "learning_rate": 4.520176110575337e-07,
      "loss": 1.5697,
      "step": 1404
    },
    {
      "epoch": 0.21189955508634342,
      "grad_norm": 0.2569437009834236,
      "learning_rate": 4.519517799861097e-07,
      "loss": 1.5759,
      "step": 1405
    },
    {
      "epoch": 0.21205037327501697,
      "grad_norm": 0.2476924207888695,
      "learning_rate": 4.518859091838576e-07,
      "loss": 1.6233,
      "step": 1406
    },
    {
      "epoch": 0.21220119146369051,
      "grad_norm": 0.2532654188168525,
      "learning_rate": 4.5181999866556744e-07,
      "loss": 1.6218,
      "step": 1407
    },
    {
      "epoch": 0.2123520096523641,
      "grad_norm": 0.25466059140149927,
      "learning_rate": 4.5175404844603805e-07,
      "loss": 1.5523,
      "step": 1408
    },
    {
      "epoch": 0.21250282784103763,
      "grad_norm": 0.5170922925364997,
      "learning_rate": 4.5168805854007714e-07,
      "loss": 1.5398,
      "step": 1409
    },
    {
      "epoch": 0.21265364602971118,
      "grad_norm": 0.25832194865975405,
      "learning_rate": 4.516220289625013e-07,
      "loss": 1.5365,
      "step": 1410
    },
    {
      "epoch": 0.21280446421838473,
      "grad_norm": 0.2668076476122713,
      "learning_rate": 4.5155595972813616e-07,
      "loss": 1.574,
      "step": 1411
    },
    {
      "epoch": 0.2129552824070583,
      "grad_norm": 0.30558559464052965,
      "learning_rate": 4.514898508518162e-07,
      "loss": 1.5048,
      "step": 1412
    },
    {
      "epoch": 0.21310610059573185,
      "grad_norm": 0.24986675140765469,
      "learning_rate": 4.514237023483848e-07,
      "loss": 1.594,
      "step": 1413
    },
    {
      "epoch": 0.2132569187844054,
      "grad_norm": 0.2967956906681153,
      "learning_rate": 4.513575142326942e-07,
      "loss": 1.6564,
      "step": 1414
    },
    {
      "epoch": 0.21340773697307897,
      "grad_norm": 0.25236193419130576,
      "learning_rate": 4.512912865196055e-07,
      "loss": 1.53,
      "step": 1415
    },
    {
      "epoch": 0.2135585551617525,
      "grad_norm": 0.23697946220883495,
      "learning_rate": 4.512250192239888e-07,
      "loss": 1.5201,
      "step": 1416
    },
    {
      "epoch": 0.21370937335042606,
      "grad_norm": 0.25188233784849035,
      "learning_rate": 4.511587123607231e-07,
      "loss": 1.5729,
      "step": 1417
    },
    {
      "epoch": 0.2138601915390996,
      "grad_norm": 0.24316938283711606,
      "learning_rate": 4.5109236594469616e-07,
      "loss": 1.581,
      "step": 1418
    },
    {
      "epoch": 0.21401100972777318,
      "grad_norm": 0.258347771069871,
      "learning_rate": 4.510259799908047e-07,
      "loss": 1.5459,
      "step": 1419
    },
    {
      "epoch": 0.21416182791644672,
      "grad_norm": 0.2553160362376422,
      "learning_rate": 4.509595545139542e-07,
      "loss": 1.593,
      "step": 1420
    },
    {
      "epoch": 0.21431264610512027,
      "grad_norm": 0.25322419596260776,
      "learning_rate": 4.508930895290593e-07,
      "loss": 1.5785,
      "step": 1421
    },
    {
      "epoch": 0.21446346429379384,
      "grad_norm": 0.24474866831756664,
      "learning_rate": 4.5082658505104323e-07,
      "loss": 1.6553,
      "step": 1422
    },
    {
      "epoch": 0.2146142824824674,
      "grad_norm": 0.23931984187269134,
      "learning_rate": 4.5076004109483813e-07,
      "loss": 1.6013,
      "step": 1423
    },
    {
      "epoch": 0.21476510067114093,
      "grad_norm": 0.27588477665124317,
      "learning_rate": 4.5069345767538507e-07,
      "loss": 1.5221,
      "step": 1424
    },
    {
      "epoch": 0.21491591885981448,
      "grad_norm": 0.2628541784750753,
      "learning_rate": 4.5062683480763407e-07,
      "loss": 1.5947,
      "step": 1425
    },
    {
      "epoch": 0.21506673704848805,
      "grad_norm": 0.24678121117596793,
      "learning_rate": 4.505601725065439e-07,
      "loss": 1.5461,
      "step": 1426
    },
    {
      "epoch": 0.2152175552371616,
      "grad_norm": 0.26435666630384597,
      "learning_rate": 4.50493470787082e-07,
      "loss": 1.57,
      "step": 1427
    },
    {
      "epoch": 0.21536837342583515,
      "grad_norm": 0.3012053269260454,
      "learning_rate": 4.50426729664225e-07,
      "loss": 1.5197,
      "step": 1428
    },
    {
      "epoch": 0.21551919161450872,
      "grad_norm": 0.25146246237406045,
      "learning_rate": 4.503599491529582e-07,
      "loss": 1.5976,
      "step": 1429
    },
    {
      "epoch": 0.21567000980318227,
      "grad_norm": 0.25224063497089616,
      "learning_rate": 4.502931292682758e-07,
      "loss": 1.5875,
      "step": 1430
    },
    {
      "epoch": 0.2158208279918558,
      "grad_norm": 0.25331435975433114,
      "learning_rate": 4.5022627002518075e-07,
      "loss": 1.6048,
      "step": 1431
    },
    {
      "epoch": 0.21597164618052936,
      "grad_norm": 0.25813299552290325,
      "learning_rate": 4.50159371438685e-07,
      "loss": 1.5472,
      "step": 1432
    },
    {
      "epoch": 0.21612246436920293,
      "grad_norm": 0.2404868997906578,
      "learning_rate": 4.500924335238091e-07,
      "loss": 1.6059,
      "step": 1433
    },
    {
      "epoch": 0.21627328255787648,
      "grad_norm": 0.2613901399699812,
      "learning_rate": 4.500254562955826e-07,
      "loss": 1.6188,
      "step": 1434
    },
    {
      "epoch": 0.21642410074655002,
      "grad_norm": 0.24683355561448922,
      "learning_rate": 4.4995843976904386e-07,
      "loss": 1.5956,
      "step": 1435
    },
    {
      "epoch": 0.2165749189352236,
      "grad_norm": 0.25137075226570926,
      "learning_rate": 4.498913839592401e-07,
      "loss": 1.566,
      "step": 1436
    },
    {
      "epoch": 0.21672573712389714,
      "grad_norm": 0.282624118284134,
      "learning_rate": 4.498242888812272e-07,
      "loss": 1.6292,
      "step": 1437
    },
    {
      "epoch": 0.2168765553125707,
      "grad_norm": 0.2990794129963659,
      "learning_rate": 4.4975715455007e-07,
      "loss": 1.5933,
      "step": 1438
    },
    {
      "epoch": 0.21702737350124424,
      "grad_norm": 0.2980897205504419,
      "learning_rate": 4.496899809808421e-07,
      "loss": 1.6262,
      "step": 1439
    },
    {
      "epoch": 0.2171781916899178,
      "grad_norm": 0.23491385301471046,
      "learning_rate": 4.4962276818862593e-07,
      "loss": 1.5114,
      "step": 1440
    },
    {
      "epoch": 0.21732900987859136,
      "grad_norm": 0.24169981231441703,
      "learning_rate": 4.495555161885128e-07,
      "loss": 1.5901,
      "step": 1441
    },
    {
      "epoch": 0.2174798280672649,
      "grad_norm": 0.24587673073324526,
      "learning_rate": 4.494882249956026e-07,
      "loss": 1.5081,
      "step": 1442
    },
    {
      "epoch": 0.21763064625593848,
      "grad_norm": 0.6741389015324207,
      "learning_rate": 4.4942089462500414e-07,
      "loss": 1.5852,
      "step": 1443
    },
    {
      "epoch": 0.21778146444461202,
      "grad_norm": 0.30447488077669177,
      "learning_rate": 4.4935352509183527e-07,
      "loss": 1.5588,
      "step": 1444
    },
    {
      "epoch": 0.21793228263328557,
      "grad_norm": 0.2687597512504896,
      "learning_rate": 4.4928611641122223e-07,
      "loss": 1.591,
      "step": 1445
    },
    {
      "epoch": 0.21808310082195914,
      "grad_norm": 0.24939306790272178,
      "learning_rate": 4.492186685983004e-07,
      "loss": 1.5418,
      "step": 1446
    },
    {
      "epoch": 0.2182339190106327,
      "grad_norm": 0.24246473944579597,
      "learning_rate": 4.4915118166821346e-07,
      "loss": 1.5475,
      "step": 1447
    },
    {
      "epoch": 0.21838473719930623,
      "grad_norm": 0.25160094471473154,
      "learning_rate": 4.4908365563611456e-07,
      "loss": 1.5889,
      "step": 1448
    },
    {
      "epoch": 0.21853555538797978,
      "grad_norm": 0.24502512931851553,
      "learning_rate": 4.49016090517165e-07,
      "loss": 1.6334,
      "step": 1449
    },
    {
      "epoch": 0.21868637357665335,
      "grad_norm": 0.24632534584312824,
      "learning_rate": 4.489484863265352e-07,
      "loss": 1.5815,
      "step": 1450
    },
    {
      "epoch": 0.2188371917653269,
      "grad_norm": 0.7228259896607148,
      "learning_rate": 4.4888084307940433e-07,
      "loss": 1.5779,
      "step": 1451
    },
    {
      "epoch": 0.21898800995400045,
      "grad_norm": 0.28835413330730547,
      "learning_rate": 4.488131607909602e-07,
      "loss": 1.5069,
      "step": 1452
    },
    {
      "epoch": 0.21913882814267402,
      "grad_norm": 0.32849414491868345,
      "learning_rate": 4.4874543947639946e-07,
      "loss": 1.5055,
      "step": 1453
    },
    {
      "epoch": 0.21928964633134757,
      "grad_norm": 0.25614128263544494,
      "learning_rate": 4.486776791509276e-07,
      "loss": 1.6218,
      "step": 1454
    },
    {
      "epoch": 0.2194404645200211,
      "grad_norm": 0.2788075708337781,
      "learning_rate": 4.486098798297587e-07,
      "loss": 1.5488,
      "step": 1455
    },
    {
      "epoch": 0.21959128270869466,
      "grad_norm": 0.25557639661216225,
      "learning_rate": 4.4854204152811563e-07,
      "loss": 1.6018,
      "step": 1456
    },
    {
      "epoch": 0.21974210089736823,
      "grad_norm": 0.2688433402158743,
      "learning_rate": 4.484741642612302e-07,
      "loss": 1.5651,
      "step": 1457
    },
    {
      "epoch": 0.21989291908604178,
      "grad_norm": 0.2744743880824375,
      "learning_rate": 4.484062480443427e-07,
      "loss": 1.553,
      "step": 1458
    },
    {
      "epoch": 0.22004373727471532,
      "grad_norm": 0.2620242735199598,
      "learning_rate": 4.483382928927025e-07,
      "loss": 1.5042,
      "step": 1459
    },
    {
      "epoch": 0.2201945554633889,
      "grad_norm": 0.24288976128343828,
      "learning_rate": 4.482702988215672e-07,
      "loss": 1.5489,
      "step": 1460
    },
    {
      "epoch": 0.22034537365206244,
      "grad_norm": 0.2594367330284345,
      "learning_rate": 4.482022658462037e-07,
      "loss": 1.4735,
      "step": 1461
    },
    {
      "epoch": 0.220496191840736,
      "grad_norm": 0.3167292658690441,
      "learning_rate": 4.4813419398188736e-07,
      "loss": 1.5435,
      "step": 1462
    },
    {
      "epoch": 0.22064701002940953,
      "grad_norm": 0.2519629922479481,
      "learning_rate": 4.480660832439022e-07,
      "loss": 1.5606,
      "step": 1463
    },
    {
      "epoch": 0.2207978282180831,
      "grad_norm": 0.31160173778972206,
      "learning_rate": 4.4799793364754113e-07,
      "loss": 1.5988,
      "step": 1464
    },
    {
      "epoch": 0.22094864640675665,
      "grad_norm": 0.3017651985325454,
      "learning_rate": 4.479297452081056e-07,
      "loss": 1.579,
      "step": 1465
    },
    {
      "epoch": 0.2210994645954302,
      "grad_norm": 0.2439919574688214,
      "learning_rate": 4.478615179409061e-07,
      "loss": 1.5537,
      "step": 1466
    },
    {
      "epoch": 0.22125028278410377,
      "grad_norm": 0.32076252021976376,
      "learning_rate": 4.4779325186126157e-07,
      "loss": 1.4989,
      "step": 1467
    },
    {
      "epoch": 0.22140110097277732,
      "grad_norm": 0.26932891039293677,
      "learning_rate": 4.4772494698449965e-07,
      "loss": 1.5191,
      "step": 1468
    },
    {
      "epoch": 0.22155191916145087,
      "grad_norm": 0.2633568107974307,
      "learning_rate": 4.476566033259568e-07,
      "loss": 1.5494,
      "step": 1469
    },
    {
      "epoch": 0.2217027373501244,
      "grad_norm": 0.28810828200999133,
      "learning_rate": 4.475882209009782e-07,
      "loss": 1.5609,
      "step": 1470
    },
    {
      "epoch": 0.221853555538798,
      "grad_norm": 0.3317982309128049,
      "learning_rate": 4.475197997249177e-07,
      "loss": 1.5672,
      "step": 1471
    },
    {
      "epoch": 0.22200437372747153,
      "grad_norm": 0.2538882922500279,
      "learning_rate": 4.4745133981313785e-07,
      "loss": 1.524,
      "step": 1472
    },
    {
      "epoch": 0.22215519191614508,
      "grad_norm": 0.27207734764137315,
      "learning_rate": 4.4738284118100984e-07,
      "loss": 1.5714,
      "step": 1473
    },
    {
      "epoch": 0.22230601010481865,
      "grad_norm": 0.24683524078912186,
      "learning_rate": 4.473143038439137e-07,
      "loss": 1.5806,
      "step": 1474
    },
    {
      "epoch": 0.2224568282934922,
      "grad_norm": 0.2855048578394907,
      "learning_rate": 4.4724572781723787e-07,
      "loss": 1.5427,
      "step": 1475
    },
    {
      "epoch": 0.22260764648216574,
      "grad_norm": 0.2897639699473296,
      "learning_rate": 4.471771131163799e-07,
      "loss": 1.5443,
      "step": 1476
    },
    {
      "epoch": 0.2227584646708393,
      "grad_norm": 0.2550600619785885,
      "learning_rate": 4.4710845975674565e-07,
      "loss": 1.5153,
      "step": 1477
    },
    {
      "epoch": 0.22290928285951286,
      "grad_norm": 0.25232906029764135,
      "learning_rate": 4.4703976775374986e-07,
      "loss": 1.4903,
      "step": 1478
    },
    {
      "epoch": 0.2230601010481864,
      "grad_norm": 0.25328137332083733,
      "learning_rate": 4.469710371228158e-07,
      "loss": 1.5976,
      "step": 1479
    },
    {
      "epoch": 0.22321091923685996,
      "grad_norm": 0.27936930653293984,
      "learning_rate": 4.469022678793756e-07,
      "loss": 1.6363,
      "step": 1480
    },
    {
      "epoch": 0.22336173742553353,
      "grad_norm": 0.2548072627596578,
      "learning_rate": 4.4683346003887e-07,
      "loss": 1.584,
      "step": 1481
    },
    {
      "epoch": 0.22351255561420708,
      "grad_norm": 0.251863851633017,
      "learning_rate": 4.4676461361674816e-07,
      "loss": 1.5825,
      "step": 1482
    },
    {
      "epoch": 0.22366337380288062,
      "grad_norm": 0.2607448940880632,
      "learning_rate": 4.466957286284683e-07,
      "loss": 1.5988,
      "step": 1483
    },
    {
      "epoch": 0.22381419199155417,
      "grad_norm": 0.2569225382965004,
      "learning_rate": 4.466268050894969e-07,
      "loss": 1.5543,
      "step": 1484
    },
    {
      "epoch": 0.22396501018022774,
      "grad_norm": 0.24952522775379582,
      "learning_rate": 4.4655784301530955e-07,
      "loss": 1.5792,
      "step": 1485
    },
    {
      "epoch": 0.2241158283689013,
      "grad_norm": 0.2651653229474127,
      "learning_rate": 4.4648884242139005e-07,
      "loss": 1.5456,
      "step": 1486
    },
    {
      "epoch": 0.22426664655757483,
      "grad_norm": 0.24259453226169617,
      "learning_rate": 4.464198033232312e-07,
      "loss": 1.5347,
      "step": 1487
    },
    {
      "epoch": 0.2244174647462484,
      "grad_norm": 0.26490694394792824,
      "learning_rate": 4.4635072573633416e-07,
      "loss": 1.5912,
      "step": 1488
    },
    {
      "epoch": 0.22456828293492195,
      "grad_norm": 0.2866525843062583,
      "learning_rate": 4.4628160967620886e-07,
      "loss": 1.614,
      "step": 1489
    },
    {
      "epoch": 0.2247191011235955,
      "grad_norm": 0.2682453651065939,
      "learning_rate": 4.46212455158374e-07,
      "loss": 1.5625,
      "step": 1490
    },
    {
      "epoch": 0.22486991931226907,
      "grad_norm": 0.2419121075971261,
      "learning_rate": 4.4614326219835666e-07,
      "loss": 1.5419,
      "step": 1491
    },
    {
      "epoch": 0.22502073750094262,
      "grad_norm": 0.267993400660638,
      "learning_rate": 4.460740308116928e-07,
      "loss": 1.6074,
      "step": 1492
    },
    {
      "epoch": 0.22517155568961617,
      "grad_norm": 0.32575575615746094,
      "learning_rate": 4.4600476101392677e-07,
      "loss": 1.5516,
      "step": 1493
    },
    {
      "epoch": 0.2253223738782897,
      "grad_norm": 0.2483260022994849,
      "learning_rate": 4.4593545282061156e-07,
      "loss": 1.6733,
      "step": 1494
    },
    {
      "epoch": 0.22547319206696328,
      "grad_norm": 0.27530216464083385,
      "learning_rate": 4.458661062473091e-07,
      "loss": 1.606,
      "step": 1495
    },
    {
      "epoch": 0.22562401025563683,
      "grad_norm": 0.2443319973678457,
      "learning_rate": 4.457967213095896e-07,
      "loss": 1.5532,
      "step": 1496
    },
    {
      "epoch": 0.22577482844431038,
      "grad_norm": 0.2427356152328098,
      "learning_rate": 4.45727298023032e-07,
      "loss": 1.5055,
      "step": 1497
    },
    {
      "epoch": 0.22592564663298395,
      "grad_norm": 0.25507158062659324,
      "learning_rate": 4.4565783640322386e-07,
      "loss": 1.557,
      "step": 1498
    },
    {
      "epoch": 0.2260764648216575,
      "grad_norm": 0.2620037370610053,
      "learning_rate": 4.455883364657614e-07,
      "loss": 1.6873,
      "step": 1499
    },
    {
      "epoch": 0.22622728301033104,
      "grad_norm": 0.2470852857188749,
      "learning_rate": 4.455187982262493e-07,
      "loss": 1.5153,
      "step": 1500
    },
    {
      "epoch": 0.2263781011990046,
      "grad_norm": 0.2479219594273459,
      "learning_rate": 4.4544922170030084e-07,
      "loss": 1.5514,
      "step": 1501
    },
    {
      "epoch": 0.22652891938767816,
      "grad_norm": 0.2582843145931511,
      "learning_rate": 4.4537960690353823e-07,
      "loss": 1.4804,
      "step": 1502
    },
    {
      "epoch": 0.2266797375763517,
      "grad_norm": 0.2683250070621692,
      "learning_rate": 4.4530995385159173e-07,
      "loss": 1.6919,
      "step": 1503
    },
    {
      "epoch": 0.22683055576502525,
      "grad_norm": 0.23998718465092833,
      "learning_rate": 4.4524026256010073e-07,
      "loss": 1.6152,
      "step": 1504
    },
    {
      "epoch": 0.22698137395369883,
      "grad_norm": 0.3137984877545782,
      "learning_rate": 4.4517053304471273e-07,
      "loss": 1.6059,
      "step": 1505
    },
    {
      "epoch": 0.22713219214237237,
      "grad_norm": 0.2569777072612162,
      "learning_rate": 4.4510076532108425e-07,
      "loss": 1.5924,
      "step": 1506
    },
    {
      "epoch": 0.22728301033104592,
      "grad_norm": 0.24520541445209215,
      "learning_rate": 4.4503095940488e-07,
      "loss": 1.546,
      "step": 1507
    },
    {
      "epoch": 0.22743382851971947,
      "grad_norm": 0.27660005451994707,
      "learning_rate": 4.4496111531177356e-07,
      "loss": 1.6151,
      "step": 1508
    },
    {
      "epoch": 0.22758464670839304,
      "grad_norm": 0.4193956155170342,
      "learning_rate": 4.448912330574469e-07,
      "loss": 1.5805,
      "step": 1509
    },
    {
      "epoch": 0.2277354648970666,
      "grad_norm": 0.24382335659336823,
      "learning_rate": 4.4482131265759067e-07,
      "loss": 1.6322,
      "step": 1510
    },
    {
      "epoch": 0.22788628308574013,
      "grad_norm": 0.2621205660842912,
      "learning_rate": 4.44751354127904e-07,
      "loss": 1.581,
      "step": 1511
    },
    {
      "epoch": 0.2280371012744137,
      "grad_norm": 0.2654196652049394,
      "learning_rate": 4.446813574840946e-07,
      "loss": 1.5895,
      "step": 1512
    },
    {
      "epoch": 0.22818791946308725,
      "grad_norm": 0.24947378314897425,
      "learning_rate": 4.446113227418787e-07,
      "loss": 1.5185,
      "step": 1513
    },
    {
      "epoch": 0.2283387376517608,
      "grad_norm": 0.25239730996688686,
      "learning_rate": 4.4454124991698136e-07,
      "loss": 1.5731,
      "step": 1514
    },
    {
      "epoch": 0.22848955584043434,
      "grad_norm": 0.27285743016093594,
      "learning_rate": 4.4447113902513577e-07,
      "loss": 1.597,
      "step": 1515
    },
    {
      "epoch": 0.22864037402910792,
      "grad_norm": 0.2492947708196186,
      "learning_rate": 4.44400990082084e-07,
      "loss": 1.6109,
      "step": 1516
    },
    {
      "epoch": 0.22879119221778146,
      "grad_norm": 0.2475032437299696,
      "learning_rate": 4.4433080310357636e-07,
      "loss": 1.5192,
      "step": 1517
    },
    {
      "epoch": 0.228942010406455,
      "grad_norm": 0.24904705840941427,
      "learning_rate": 4.4426057810537187e-07,
      "loss": 1.5728,
      "step": 1518
    },
    {
      "epoch": 0.22909282859512858,
      "grad_norm": 0.48411826079897535,
      "learning_rate": 4.441903151032383e-07,
      "loss": 1.6649,
      "step": 1519
    },
    {
      "epoch": 0.22924364678380213,
      "grad_norm": 0.2454018555263281,
      "learning_rate": 4.441200141129516e-07,
      "loss": 1.5949,
      "step": 1520
    },
    {
      "epoch": 0.22939446497247568,
      "grad_norm": 0.2366798276780095,
      "learning_rate": 4.440496751502965e-07,
      "loss": 1.5925,
      "step": 1521
    },
    {
      "epoch": 0.22954528316114922,
      "grad_norm": 0.2410733842493999,
      "learning_rate": 4.4397929823106596e-07,
      "loss": 1.5542,
      "step": 1522
    },
    {
      "epoch": 0.2296961013498228,
      "grad_norm": 0.5869512944596569,
      "learning_rate": 4.4390888337106177e-07,
      "loss": 1.6107,
      "step": 1523
    },
    {
      "epoch": 0.22984691953849634,
      "grad_norm": 0.2569931933365196,
      "learning_rate": 4.4383843058609413e-07,
      "loss": 1.5418,
      "step": 1524
    },
    {
      "epoch": 0.2299977377271699,
      "grad_norm": 0.29199340682139846,
      "learning_rate": 4.4376793989198165e-07,
      "loss": 1.5371,
      "step": 1525
    },
    {
      "epoch": 0.23014855591584346,
      "grad_norm": 0.3161682392737767,
      "learning_rate": 4.4369741130455156e-07,
      "loss": 1.587,
      "step": 1526
    },
    {
      "epoch": 0.230299374104517,
      "grad_norm": 0.25244665931894156,
      "learning_rate": 4.4362684483963967e-07,
      "loss": 1.5758,
      "step": 1527
    },
    {
      "epoch": 0.23045019229319055,
      "grad_norm": 0.3287435741615946,
      "learning_rate": 4.435562405130902e-07,
      "loss": 1.5638,
      "step": 1528
    },
    {
      "epoch": 0.2306010104818641,
      "grad_norm": 0.26030938077603366,
      "learning_rate": 4.4348559834075577e-07,
      "loss": 1.5588,
      "step": 1529
    },
    {
      "epoch": 0.23075182867053767,
      "grad_norm": 0.2529517007869986,
      "learning_rate": 4.434149183384977e-07,
      "loss": 1.4988,
      "step": 1530
    },
    {
      "epoch": 0.23090264685921122,
      "grad_norm": 0.23807690258336311,
      "learning_rate": 4.433442005221857e-07,
      "loss": 1.5728,
      "step": 1531
    },
    {
      "epoch": 0.23105346504788477,
      "grad_norm": 0.2980673816915895,
      "learning_rate": 4.432734449076979e-07,
      "loss": 1.5573,
      "step": 1532
    },
    {
      "epoch": 0.23120428323655834,
      "grad_norm": 0.24343236440687027,
      "learning_rate": 4.4320265151092115e-07,
      "loss": 1.5945,
      "step": 1533
    },
    {
      "epoch": 0.23135510142523188,
      "grad_norm": 0.2438373996327165,
      "learning_rate": 4.4313182034775044e-07,
      "loss": 1.5857,
      "step": 1534
    },
    {
      "epoch": 0.23150591961390543,
      "grad_norm": 0.2847928399924164,
      "learning_rate": 4.4306095143408957e-07,
      "loss": 1.5607,
      "step": 1535
    },
    {
      "epoch": 0.231656737802579,
      "grad_norm": 0.24168986587532496,
      "learning_rate": 4.429900447858506e-07,
      "loss": 1.5662,
      "step": 1536
    },
    {
      "epoch": 0.23180755599125255,
      "grad_norm": 0.24619379089962148,
      "learning_rate": 4.429191004189542e-07,
      "loss": 1.5442,
      "step": 1537
    },
    {
      "epoch": 0.2319583741799261,
      "grad_norm": 0.24484056814337715,
      "learning_rate": 4.4284811834932933e-07,
      "loss": 1.6062,
      "step": 1538
    },
    {
      "epoch": 0.23210919236859964,
      "grad_norm": 0.2489517175275712,
      "learning_rate": 4.4277709859291375e-07,
      "loss": 1.5339,
      "step": 1539
    },
    {
      "epoch": 0.23226001055727322,
      "grad_norm": 0.252822674318724,
      "learning_rate": 4.4270604116565324e-07,
      "loss": 1.6146,
      "step": 1540
    },
    {
      "epoch": 0.23241082874594676,
      "grad_norm": 0.25335946947980326,
      "learning_rate": 4.4263494608350234e-07,
      "loss": 1.5978,
      "step": 1541
    },
    {
      "epoch": 0.2325616469346203,
      "grad_norm": 0.3251146437009607,
      "learning_rate": 4.4256381336242393e-07,
      "loss": 1.5488,
      "step": 1542
    },
    {
      "epoch": 0.23271246512329388,
      "grad_norm": 0.2703768118135377,
      "learning_rate": 4.424926430183895e-07,
      "loss": 1.5846,
      "step": 1543
    },
    {
      "epoch": 0.23286328331196743,
      "grad_norm": 0.27268227347666524,
      "learning_rate": 4.424214350673788e-07,
      "loss": 1.577,
      "step": 1544
    },
    {
      "epoch": 0.23301410150064097,
      "grad_norm": 0.24783995137465384,
      "learning_rate": 4.4235018952538006e-07,
      "loss": 1.559,
      "step": 1545
    },
    {
      "epoch": 0.23316491968931452,
      "grad_norm": 0.26967822095437516,
      "learning_rate": 4.4227890640839e-07,
      "loss": 1.5483,
      "step": 1546
    },
    {
      "epoch": 0.2333157378779881,
      "grad_norm": 0.25574816910228887,
      "learning_rate": 4.422075857324137e-07,
      "loss": 1.6082,
      "step": 1547
    },
    {
      "epoch": 0.23346655606666164,
      "grad_norm": 0.25014337543770127,
      "learning_rate": 4.4213622751346474e-07,
      "loss": 1.5987,
      "step": 1548
    },
    {
      "epoch": 0.2336173742553352,
      "grad_norm": 0.25473300230673646,
      "learning_rate": 4.420648317675653e-07,
      "loss": 1.5779,
      "step": 1549
    },
    {
      "epoch": 0.23376819244400876,
      "grad_norm": 0.24752771791929504,
      "learning_rate": 4.4199339851074556e-07,
      "loss": 1.5996,
      "step": 1550
    },
    {
      "epoch": 0.2339190106326823,
      "grad_norm": 0.24099919228942546,
      "learning_rate": 4.419219277590445e-07,
      "loss": 1.5386,
      "step": 1551
    },
    {
      "epoch": 0.23406982882135585,
      "grad_norm": 0.4893165747025458,
      "learning_rate": 4.4185041952850937e-07,
      "loss": 1.5356,
      "step": 1552
    },
    {
      "epoch": 0.2342206470100294,
      "grad_norm": 0.2659578973957303,
      "learning_rate": 4.4177887383519584e-07,
      "loss": 1.6222,
      "step": 1553
    },
    {
      "epoch": 0.23437146519870297,
      "grad_norm": 0.24979724219614716,
      "learning_rate": 4.41707290695168e-07,
      "loss": 1.5928,
      "step": 1554
    },
    {
      "epoch": 0.23452228338737652,
      "grad_norm": 0.27578137794704516,
      "learning_rate": 4.416356701244983e-07,
      "loss": 1.662,
      "step": 1555
    },
    {
      "epoch": 0.23467310157605006,
      "grad_norm": 0.2680631467952242,
      "learning_rate": 4.4156401213926777e-07,
      "loss": 1.6505,
      "step": 1556
    },
    {
      "epoch": 0.23482391976472364,
      "grad_norm": 0.2700553313518092,
      "learning_rate": 4.414923167555655e-07,
      "loss": 1.6536,
      "step": 1557
    },
    {
      "epoch": 0.23497473795339718,
      "grad_norm": 0.25150863617644587,
      "learning_rate": 4.4142058398948937e-07,
      "loss": 1.6113,
      "step": 1558
    },
    {
      "epoch": 0.23512555614207073,
      "grad_norm": 0.23728026868984647,
      "learning_rate": 4.413488138571455e-07,
      "loss": 1.5653,
      "step": 1559
    },
    {
      "epoch": 0.23527637433074428,
      "grad_norm": 0.8685979934961994,
      "learning_rate": 4.4127700637464827e-07,
      "loss": 1.6315,
      "step": 1560
    },
    {
      "epoch": 0.23542719251941785,
      "grad_norm": 0.2725852994301352,
      "learning_rate": 4.412051615581206e-07,
      "loss": 1.5659,
      "step": 1561
    },
    {
      "epoch": 0.2355780107080914,
      "grad_norm": 0.2548194551778707,
      "learning_rate": 4.4113327942369373e-07,
      "loss": 1.5474,
      "step": 1562
    },
    {
      "epoch": 0.23572882889676494,
      "grad_norm": 1.5574091156237442,
      "learning_rate": 4.410613599875073e-07,
      "loss": 1.5811,
      "step": 1563
    },
    {
      "epoch": 0.23587964708543852,
      "grad_norm": 0.2492067176832404,
      "learning_rate": 4.4098940326570935e-07,
      "loss": 1.5303,
      "step": 1564
    },
    {
      "epoch": 0.23603046527411206,
      "grad_norm": 0.24230255657856714,
      "learning_rate": 4.4091740927445607e-07,
      "loss": 1.576,
      "step": 1565
    },
    {
      "epoch": 0.2361812834627856,
      "grad_norm": 0.24894191425084267,
      "learning_rate": 4.408453780299125e-07,
      "loss": 1.5438,
      "step": 1566
    },
    {
      "epoch": 0.23633210165145915,
      "grad_norm": 0.2392475719036145,
      "learning_rate": 4.4077330954825153e-07,
      "loss": 1.4873,
      "step": 1567
    },
    {
      "epoch": 0.23648291984013273,
      "grad_norm": 0.2584213398407718,
      "learning_rate": 4.407012038456547e-07,
      "loss": 1.6108,
      "step": 1568
    },
    {
      "epoch": 0.23663373802880627,
      "grad_norm": 0.24981877332816438,
      "learning_rate": 4.40629060938312e-07,
      "loss": 1.4808,
      "step": 1569
    },
    {
      "epoch": 0.23678455621747982,
      "grad_norm": 0.2670772171175633,
      "learning_rate": 4.405568808424213e-07,
      "loss": 1.5513,
      "step": 1570
    },
    {
      "epoch": 0.2369353744061534,
      "grad_norm": 0.2561177002350499,
      "learning_rate": 4.404846635741894e-07,
      "loss": 1.5232,
      "step": 1571
    },
    {
      "epoch": 0.23708619259482694,
      "grad_norm": 0.24947164268052072,
      "learning_rate": 4.4041240914983114e-07,
      "loss": 1.5567,
      "step": 1572
    },
    {
      "epoch": 0.23723701078350048,
      "grad_norm": 0.46657843250411435,
      "learning_rate": 4.403401175855695e-07,
      "loss": 1.5458,
      "step": 1573
    },
    {
      "epoch": 0.23738782897217403,
      "grad_norm": 0.24195526178643315,
      "learning_rate": 4.4026778889763637e-07,
      "loss": 1.5594,
      "step": 1574
    },
    {
      "epoch": 0.2375386471608476,
      "grad_norm": 0.27600359901541593,
      "learning_rate": 4.401954231022714e-07,
      "loss": 1.5401,
      "step": 1575
    },
    {
      "epoch": 0.23768946534952115,
      "grad_norm": 0.24704062385590803,
      "learning_rate": 4.40123020215723e-07,
      "loss": 1.5771,
      "step": 1576
    },
    {
      "epoch": 0.2378402835381947,
      "grad_norm": 0.259061251052292,
      "learning_rate": 4.4005058025424765e-07,
      "loss": 1.5045,
      "step": 1577
    },
    {
      "epoch": 0.23799110172686827,
      "grad_norm": 0.25213420085896426,
      "learning_rate": 4.399781032341102e-07,
      "loss": 1.6264,
      "step": 1578
    },
    {
      "epoch": 0.23814191991554182,
      "grad_norm": 0.24235696486526614,
      "learning_rate": 4.399055891715839e-07,
      "loss": 1.554,
      "step": 1579
    },
    {
      "epoch": 0.23829273810421536,
      "grad_norm": 0.24894393406909265,
      "learning_rate": 4.398330380829502e-07,
      "loss": 1.5122,
      "step": 1580
    },
    {
      "epoch": 0.23844355629288894,
      "grad_norm": 0.2522058131129758,
      "learning_rate": 4.3976044998449904e-07,
      "loss": 1.5451,
      "step": 1581
    },
    {
      "epoch": 0.23859437448156248,
      "grad_norm": 0.2481532556382654,
      "learning_rate": 4.3968782489252845e-07,
      "loss": 1.5803,
      "step": 1582
    },
    {
      "epoch": 0.23874519267023603,
      "grad_norm": 0.25947523180652005,
      "learning_rate": 4.3961516282334497e-07,
      "loss": 1.6562,
      "step": 1583
    },
    {
      "epoch": 0.23889601085890957,
      "grad_norm": 0.47998219527827213,
      "learning_rate": 4.3954246379326326e-07,
      "loss": 1.5688,
      "step": 1584
    },
    {
      "epoch": 0.23904682904758315,
      "grad_norm": 0.24000681895761397,
      "learning_rate": 4.394697278186065e-07,
      "loss": 1.6005,
      "step": 1585
    },
    {
      "epoch": 0.2391976472362567,
      "grad_norm": 0.24719599930872269,
      "learning_rate": 4.393969549157059e-07,
      "loss": 1.5753,
      "step": 1586
    },
    {
      "epoch": 0.23934846542493024,
      "grad_norm": 0.2651440718103725,
      "learning_rate": 4.393241451009012e-07,
      "loss": 1.5798,
      "step": 1587
    },
    {
      "epoch": 0.23949928361360381,
      "grad_norm": 0.2573708129158974,
      "learning_rate": 4.392512983905403e-07,
      "loss": 1.6199,
      "step": 1588
    },
    {
      "epoch": 0.23965010180227736,
      "grad_norm": 0.24350260889665962,
      "learning_rate": 4.3917841480097935e-07,
      "loss": 1.5163,
      "step": 1589
    },
    {
      "epoch": 0.2398009199909509,
      "grad_norm": 0.24904924118751787,
      "learning_rate": 4.391054943485828e-07,
      "loss": 1.6708,
      "step": 1590
    },
    {
      "epoch": 0.23995173817962445,
      "grad_norm": 0.23762750588913742,
      "learning_rate": 4.3903253704972356e-07,
      "loss": 1.6214,
      "step": 1591
    },
    {
      "epoch": 0.24010255636829803,
      "grad_norm": 0.2564142889769098,
      "learning_rate": 4.3895954292078256e-07,
      "loss": 1.6223,
      "step": 1592
    },
    {
      "epoch": 0.24025337455697157,
      "grad_norm": 0.30659021520772384,
      "learning_rate": 4.388865119781492e-07,
      "loss": 1.6608,
      "step": 1593
    },
    {
      "epoch": 0.24040419274564512,
      "grad_norm": 0.2545411842270761,
      "learning_rate": 4.38813444238221e-07,
      "loss": 1.5091,
      "step": 1594
    },
    {
      "epoch": 0.2405550109343187,
      "grad_norm": 0.2433107911193599,
      "learning_rate": 4.3874033971740375e-07,
      "loss": 1.5504,
      "step": 1595
    },
    {
      "epoch": 0.24070582912299224,
      "grad_norm": 0.25584836780205455,
      "learning_rate": 4.386671984321116e-07,
      "loss": 1.533,
      "step": 1596
    },
    {
      "epoch": 0.24085664731166578,
      "grad_norm": 0.24865548225791834,
      "learning_rate": 4.3859402039876684e-07,
      "loss": 1.5324,
      "step": 1597
    },
    {
      "epoch": 0.24100746550033933,
      "grad_norm": 0.24093033741997524,
      "learning_rate": 4.3852080563380023e-07,
      "loss": 1.585,
      "step": 1598
    },
    {
      "epoch": 0.2411582836890129,
      "grad_norm": 0.24709009545789049,
      "learning_rate": 4.3844755415365043e-07,
      "loss": 1.5321,
      "step": 1599
    },
    {
      "epoch": 0.24130910187768645,
      "grad_norm": 0.2503465791645773,
      "learning_rate": 4.3837426597476455e-07,
      "loss": 1.5112,
      "step": 1600
    },
    {
      "epoch": 0.24145992006636,
      "grad_norm": 0.2509426828595389,
      "learning_rate": 4.383009411135981e-07,
      "loss": 1.5806,
      "step": 1601
    },
    {
      "epoch": 0.24161073825503357,
      "grad_norm": 0.24265146376897662,
      "learning_rate": 4.3822757958661445e-07,
      "loss": 1.5811,
      "step": 1602
    },
    {
      "epoch": 0.24176155644370712,
      "grad_norm": 0.2733494998765217,
      "learning_rate": 4.3815418141028546e-07,
      "loss": 1.5979,
      "step": 1603
    },
    {
      "epoch": 0.24191237463238066,
      "grad_norm": 0.2451539073765318,
      "learning_rate": 4.380807466010912e-07,
      "loss": 1.6545,
      "step": 1604
    },
    {
      "epoch": 0.2420631928210542,
      "grad_norm": 4.374804279747756,
      "learning_rate": 4.3800727517551984e-07,
      "loss": 1.6669,
      "step": 1605
    },
    {
      "epoch": 0.24221401100972778,
      "grad_norm": 0.25491735405130056,
      "learning_rate": 4.37933767150068e-07,
      "loss": 1.6174,
      "step": 1606
    },
    {
      "epoch": 0.24236482919840133,
      "grad_norm": 0.2633247174467053,
      "learning_rate": 4.378602225412402e-07,
      "loss": 1.5471,
      "step": 1607
    },
    {
      "epoch": 0.24251564738707487,
      "grad_norm": 0.2533193105973062,
      "learning_rate": 4.3778664136554945e-07,
      "loss": 1.5456,
      "step": 1608
    },
    {
      "epoch": 0.24266646557574845,
      "grad_norm": 0.27718357103998414,
      "learning_rate": 4.3771302363951684e-07,
      "loss": 1.5974,
      "step": 1609
    },
    {
      "epoch": 0.242817283764422,
      "grad_norm": 0.2525200412087624,
      "learning_rate": 4.3763936937967165e-07,
      "loss": 1.5636,
      "step": 1610
    },
    {
      "epoch": 0.24296810195309554,
      "grad_norm": 0.2527191841200574,
      "learning_rate": 4.375656786025515e-07,
      "loss": 1.5813,
      "step": 1611
    },
    {
      "epoch": 0.24311892014176908,
      "grad_norm": 0.2424044029783404,
      "learning_rate": 4.3749195132470205e-07,
      "loss": 1.5572,
      "step": 1612
    },
    {
      "epoch": 0.24326973833044266,
      "grad_norm": 0.26674412181726,
      "learning_rate": 4.374181875626772e-07,
      "loss": 1.6054,
      "step": 1613
    },
    {
      "epoch": 0.2434205565191162,
      "grad_norm": 0.24507553143528535,
      "learning_rate": 4.373443873330392e-07,
      "loss": 1.4877,
      "step": 1614
    },
    {
      "epoch": 0.24357137470778975,
      "grad_norm": 0.24946117678679283,
      "learning_rate": 4.3727055065235823e-07,
      "loss": 1.5787,
      "step": 1615
    },
    {
      "epoch": 0.24372219289646332,
      "grad_norm": 0.2555671451600455,
      "learning_rate": 4.3719667753721285e-07,
      "loss": 1.6166,
      "step": 1616
    },
    {
      "epoch": 0.24387301108513687,
      "grad_norm": 0.25242635778884653,
      "learning_rate": 4.3712276800418955e-07,
      "loss": 1.5867,
      "step": 1617
    },
    {
      "epoch": 0.24402382927381042,
      "grad_norm": 0.2548187267229543,
      "learning_rate": 4.3704882206988337e-07,
      "loss": 1.54,
      "step": 1618
    },
    {
      "epoch": 0.24417464746248396,
      "grad_norm": 0.2608444111849341,
      "learning_rate": 4.3697483975089715e-07,
      "loss": 1.5985,
      "step": 1619
    },
    {
      "epoch": 0.24432546565115754,
      "grad_norm": 0.26605893611604164,
      "learning_rate": 4.369008210638423e-07,
      "loss": 1.5876,
      "step": 1620
    },
    {
      "epoch": 0.24447628383983108,
      "grad_norm": 0.2556804828535995,
      "learning_rate": 4.3682676602533803e-07,
      "loss": 1.6201,
      "step": 1621
    },
    {
      "epoch": 0.24462710202850463,
      "grad_norm": 0.24406546279014849,
      "learning_rate": 4.3675267465201187e-07,
      "loss": 1.5356,
      "step": 1622
    },
    {
      "epoch": 0.2447779202171782,
      "grad_norm": 0.24947181715050515,
      "learning_rate": 4.366785469604994e-07,
      "loss": 1.5389,
      "step": 1623
    },
    {
      "epoch": 0.24492873840585175,
      "grad_norm": 0.25390938677924657,
      "learning_rate": 4.366043829674446e-07,
      "loss": 1.5537,
      "step": 1624
    },
    {
      "epoch": 0.2450795565945253,
      "grad_norm": 0.3393992031538395,
      "learning_rate": 4.365301826894994e-07,
      "loss": 1.5805,
      "step": 1625
    },
    {
      "epoch": 0.24523037478319884,
      "grad_norm": 0.25072179195252253,
      "learning_rate": 4.3645594614332395e-07,
      "loss": 1.5496,
      "step": 1626
    },
    {
      "epoch": 0.2453811929718724,
      "grad_norm": 0.24083494853172305,
      "learning_rate": 4.3638167334558635e-07,
      "loss": 1.5589,
      "step": 1627
    },
    {
      "epoch": 0.24553201116054596,
      "grad_norm": 0.25949341928748476,
      "learning_rate": 4.363073643129632e-07,
      "loss": 1.5614,
      "step": 1628
    },
    {
      "epoch": 0.2456828293492195,
      "grad_norm": 0.2411908488354143,
      "learning_rate": 4.36233019062139e-07,
      "loss": 1.5887,
      "step": 1629
    },
    {
      "epoch": 0.24583364753789308,
      "grad_norm": 0.25684724445685997,
      "learning_rate": 4.361586376098064e-07,
      "loss": 1.5917,
      "step": 1630
    },
    {
      "epoch": 0.24598446572656663,
      "grad_norm": 0.24380831232286737,
      "learning_rate": 4.3608421997266607e-07,
      "loss": 1.4848,
      "step": 1631
    },
    {
      "epoch": 0.24613528391524017,
      "grad_norm": 0.2483656328460639,
      "learning_rate": 4.3600976616742723e-07,
      "loss": 1.5928,
      "step": 1632
    },
    {
      "epoch": 0.24628610210391375,
      "grad_norm": 0.2540050298497444,
      "learning_rate": 4.359352762108067e-07,
      "loss": 1.5672,
      "step": 1633
    },
    {
      "epoch": 0.2464369202925873,
      "grad_norm": 0.24384300478168583,
      "learning_rate": 4.3586075011952964e-07,
      "loss": 1.5376,
      "step": 1634
    },
    {
      "epoch": 0.24658773848126084,
      "grad_norm": 0.24073011406086497,
      "learning_rate": 4.357861879103295e-07,
      "loss": 1.5768,
      "step": 1635
    },
    {
      "epoch": 0.24673855666993438,
      "grad_norm": 0.2514451360463635,
      "learning_rate": 4.3571158959994753e-07,
      "loss": 1.6003,
      "step": 1636
    },
    {
      "epoch": 0.24688937485860796,
      "grad_norm": 0.23943815733721524,
      "learning_rate": 4.356369552051333e-07,
      "loss": 1.5712,
      "step": 1637
    },
    {
      "epoch": 0.2470401930472815,
      "grad_norm": 0.2527623225302482,
      "learning_rate": 4.355622847426443e-07,
      "loss": 1.6135,
      "step": 1638
    },
    {
      "epoch": 0.24719101123595505,
      "grad_norm": 0.24943866281182905,
      "learning_rate": 4.3548757822924623e-07,
      "loss": 1.5667,
      "step": 1639
    },
    {
      "epoch": 0.24734182942462862,
      "grad_norm": 0.2543994544070957,
      "learning_rate": 4.35412835681713e-07,
      "loss": 1.6605,
      "step": 1640
    },
    {
      "epoch": 0.24749264761330217,
      "grad_norm": 0.49854934631647785,
      "learning_rate": 4.353380571168264e-07,
      "loss": 1.5982,
      "step": 1641
    },
    {
      "epoch": 0.24764346580197572,
      "grad_norm": 0.24169392277694388,
      "learning_rate": 4.352632425513764e-07,
      "loss": 1.5547,
      "step": 1642
    },
    {
      "epoch": 0.24779428399064926,
      "grad_norm": 0.24571868645006378,
      "learning_rate": 4.3518839200216116e-07,
      "loss": 1.5299,
      "step": 1643
    },
    {
      "epoch": 0.24794510217932283,
      "grad_norm": 0.24052959132029506,
      "learning_rate": 4.3511350548598663e-07,
      "loss": 1.6275,
      "step": 1644
    },
    {
      "epoch": 0.24809592036799638,
      "grad_norm": 0.2746076044352863,
      "learning_rate": 4.350385830196671e-07,
      "loss": 1.5231,
      "step": 1645
    },
    {
      "epoch": 0.24824673855666993,
      "grad_norm": 0.27728208574784535,
      "learning_rate": 4.349636246200249e-07,
      "loss": 1.5258,
      "step": 1646
    },
    {
      "epoch": 0.2483975567453435,
      "grad_norm": 0.2421299350714505,
      "learning_rate": 4.348886303038903e-07,
      "loss": 1.5285,
      "step": 1647
    },
    {
      "epoch": 0.24854837493401705,
      "grad_norm": 0.25132360998606307,
      "learning_rate": 4.348136000881017e-07,
      "loss": 1.6063,
      "step": 1648
    },
    {
      "epoch": 0.2486991931226906,
      "grad_norm": 0.2749557018517846,
      "learning_rate": 4.347385339895056e-07,
      "loss": 1.5415,
      "step": 1649
    },
    {
      "epoch": 0.24885001131136414,
      "grad_norm": 0.24158905852084034,
      "learning_rate": 4.346634320249565e-07,
      "loss": 1.5754,
      "step": 1650
    },
    {
      "epoch": 0.2490008295000377,
      "grad_norm": 0.3022825523735689,
      "learning_rate": 4.34588294211317e-07,
      "loss": 1.6446,
      "step": 1651
    },
    {
      "epoch": 0.24915164768871126,
      "grad_norm": 0.24616181996664016,
      "learning_rate": 4.3451312056545773e-07,
      "loss": 1.5699,
      "step": 1652
    },
    {
      "epoch": 0.2493024658773848,
      "grad_norm": 0.4045887878159248,
      "learning_rate": 4.3443791110425743e-07,
      "loss": 1.5591,
      "step": 1653
    },
    {
      "epoch": 0.24945328406605838,
      "grad_norm": 0.24005335829357874,
      "learning_rate": 4.343626658446026e-07,
      "loss": 1.5355,
      "step": 1654
    },
    {
      "epoch": 0.24960410225473192,
      "grad_norm": 0.2955294522291913,
      "learning_rate": 4.342873848033883e-07,
      "loss": 1.5653,
      "step": 1655
    },
    {
      "epoch": 0.24975492044340547,
      "grad_norm": 0.24029762289547038,
      "learning_rate": 4.3421206799751703e-07,
      "loss": 1.5152,
      "step": 1656
    },
    {
      "epoch": 0.24990573863207902,
      "grad_norm": 0.27040839936026695,
      "learning_rate": 4.341367154438998e-07,
      "loss": 1.562,
      "step": 1657
    },
    {
      "epoch": 0.25005655682075256,
      "grad_norm": 0.2490552430466942,
      "learning_rate": 4.340613271594553e-07,
      "loss": 1.5682,
      "step": 1658
    },
    {
      "epoch": 0.25020737500942614,
      "grad_norm": 0.3085223876082704,
      "learning_rate": 4.339859031611105e-07,
      "loss": 1.536,
      "step": 1659
    },
    {
      "epoch": 0.2503581931980997,
      "grad_norm": 0.24622530190986353,
      "learning_rate": 4.3391044346580027e-07,
      "loss": 1.5449,
      "step": 1660
    },
    {
      "epoch": 0.25050901138677323,
      "grad_norm": 0.4929943024064872,
      "learning_rate": 4.338349480904675e-07,
      "loss": 1.5288,
      "step": 1661
    },
    {
      "epoch": 0.2506598295754468,
      "grad_norm": 0.23007190138349864,
      "learning_rate": 4.337594170520631e-07,
      "loss": 1.479,
      "step": 1662
    },
    {
      "epoch": 0.2508106477641204,
      "grad_norm": 0.23837469219630214,
      "learning_rate": 4.33683850367546e-07,
      "loss": 1.5825,
      "step": 1663
    },
    {
      "epoch": 0.2509614659527939,
      "grad_norm": 0.2719929045506161,
      "learning_rate": 4.3360824805388313e-07,
      "loss": 1.5336,
      "step": 1664
    },
    {
      "epoch": 0.25111228414146747,
      "grad_norm": 0.27297958723610233,
      "learning_rate": 4.335326101280493e-07,
      "loss": 1.5636,
      "step": 1665
    },
    {
      "epoch": 0.25126310233014104,
      "grad_norm": 0.25677317032601776,
      "learning_rate": 4.3345693660702766e-07,
      "loss": 1.5933,
      "step": 1666
    },
    {
      "epoch": 0.25141392051881456,
      "grad_norm": 0.24753988356481488,
      "learning_rate": 4.3338122750780884e-07,
      "loss": 1.5948,
      "step": 1667
    },
    {
      "epoch": 0.25156473870748813,
      "grad_norm": 0.2550844565839159,
      "learning_rate": 4.3330548284739185e-07,
      "loss": 1.5243,
      "step": 1668
    },
    {
      "epoch": 0.25171555689616165,
      "grad_norm": 0.26171046048706154,
      "learning_rate": 4.332297026427837e-07,
      "loss": 1.4997,
      "step": 1669
    },
    {
      "epoch": 0.2518663750848352,
      "grad_norm": 0.25282478277315307,
      "learning_rate": 4.3315388691099913e-07,
      "loss": 1.5097,
      "step": 1670
    },
    {
      "epoch": 0.2520171932735088,
      "grad_norm": 0.2517793759803933,
      "learning_rate": 4.33078035669061e-07,
      "loss": 1.6149,
      "step": 1671
    },
    {
      "epoch": 0.2521680114621823,
      "grad_norm": 0.26008369217999744,
      "learning_rate": 4.3300214893400013e-07,
      "loss": 1.5947,
      "step": 1672
    },
    {
      "epoch": 0.2523188296508559,
      "grad_norm": 0.2683378898504414,
      "learning_rate": 4.3292622672285524e-07,
      "loss": 1.616,
      "step": 1673
    },
    {
      "epoch": 0.25246964783952947,
      "grad_norm": 0.23969156913581993,
      "learning_rate": 4.3285026905267315e-07,
      "loss": 1.5356,
      "step": 1674
    },
    {
      "epoch": 0.252620466028203,
      "grad_norm": 0.24086368898263558,
      "learning_rate": 4.3277427594050856e-07,
      "loss": 1.5343,
      "step": 1675
    },
    {
      "epoch": 0.25277128421687656,
      "grad_norm": 0.30730558248028683,
      "learning_rate": 4.326982474034241e-07,
      "loss": 1.6644,
      "step": 1676
    },
    {
      "epoch": 0.25292210240555013,
      "grad_norm": 0.24569205171743508,
      "learning_rate": 4.3262218345849046e-07,
      "loss": 1.5736,
      "step": 1677
    },
    {
      "epoch": 0.25307292059422365,
      "grad_norm": 0.25356805089460793,
      "learning_rate": 4.3254608412278614e-07,
      "loss": 1.4976,
      "step": 1678
    },
    {
      "epoch": 0.2532237387828972,
      "grad_norm": 0.2530609741816862,
      "learning_rate": 4.324699494133977e-07,
      "loss": 1.6002,
      "step": 1679
    },
    {
      "epoch": 0.2533745569715708,
      "grad_norm": 0.24265654867594905,
      "learning_rate": 4.3239377934741953e-07,
      "loss": 1.5675,
      "step": 1680
    },
    {
      "epoch": 0.2535253751602443,
      "grad_norm": 0.6132678415327332,
      "learning_rate": 4.323175739419542e-07,
      "loss": 1.5932,
      "step": 1681
    },
    {
      "epoch": 0.2536761933489179,
      "grad_norm": 0.2531688134086436,
      "learning_rate": 4.322413332141118e-07,
      "loss": 1.535,
      "step": 1682
    },
    {
      "epoch": 0.2538270115375914,
      "grad_norm": 0.2429492542086216,
      "learning_rate": 4.321650571810109e-07,
      "loss": 1.5977,
      "step": 1683
    },
    {
      "epoch": 0.253977829726265,
      "grad_norm": 0.2594783662561257,
      "learning_rate": 4.320887458597773e-07,
      "loss": 1.6983,
      "step": 1684
    },
    {
      "epoch": 0.25412864791493855,
      "grad_norm": 0.2556315585430283,
      "learning_rate": 4.3201239926754545e-07,
      "loss": 1.5241,
      "step": 1685
    },
    {
      "epoch": 0.2542794661036121,
      "grad_norm": 0.26542964134981467,
      "learning_rate": 4.3193601742145717e-07,
      "loss": 1.6081,
      "step": 1686
    },
    {
      "epoch": 0.25443028429228565,
      "grad_norm": 2.0009433719878618,
      "learning_rate": 4.318596003386626e-07,
      "loss": 1.4759,
      "step": 1687
    },
    {
      "epoch": 0.2545811024809592,
      "grad_norm": 0.25061467648599284,
      "learning_rate": 4.3178314803631945e-07,
      "loss": 1.5928,
      "step": 1688
    },
    {
      "epoch": 0.25473192066963274,
      "grad_norm": 0.4110198564177995,
      "learning_rate": 4.317066605315935e-07,
      "loss": 1.5477,
      "step": 1689
    },
    {
      "epoch": 0.2548827388583063,
      "grad_norm": 0.24508704358544142,
      "learning_rate": 4.3163013784165855e-07,
      "loss": 1.5367,
      "step": 1690
    },
    {
      "epoch": 0.2550335570469799,
      "grad_norm": 0.24949565040040414,
      "learning_rate": 4.31553579983696e-07,
      "loss": 1.5979,
      "step": 1691
    },
    {
      "epoch": 0.2551843752356534,
      "grad_norm": 0.2678625870670917,
      "learning_rate": 4.314769869748954e-07,
      "loss": 1.5595,
      "step": 1692
    },
    {
      "epoch": 0.255335193424327,
      "grad_norm": 0.6388075209881965,
      "learning_rate": 4.3140035883245407e-07,
      "loss": 1.5449,
      "step": 1693
    },
    {
      "epoch": 0.25548601161300055,
      "grad_norm": 0.2502282071079168,
      "learning_rate": 4.313236955735774e-07,
      "loss": 1.5923,
      "step": 1694
    },
    {
      "epoch": 0.25563682980167407,
      "grad_norm": 0.3161482244689952,
      "learning_rate": 4.312469972154784e-07,
      "loss": 1.5193,
      "step": 1695
    },
    {
      "epoch": 0.25578764799034764,
      "grad_norm": 0.24365107691376364,
      "learning_rate": 4.31170263775378e-07,
      "loss": 1.6498,
      "step": 1696
    },
    {
      "epoch": 0.25593846617902116,
      "grad_norm": 0.24672334626379835,
      "learning_rate": 4.3109349527050543e-07,
      "loss": 1.5974,
      "step": 1697
    },
    {
      "epoch": 0.25608928436769474,
      "grad_norm": 0.24377888393499542,
      "learning_rate": 4.3101669171809703e-07,
      "loss": 1.5205,
      "step": 1698
    },
    {
      "epoch": 0.2562401025563683,
      "grad_norm": 0.24090813425503693,
      "learning_rate": 4.309398531353977e-07,
      "loss": 1.6216,
      "step": 1699
    },
    {
      "epoch": 0.25639092074504183,
      "grad_norm": 0.27017836011261337,
      "learning_rate": 4.3086297953965987e-07,
      "loss": 1.5731,
      "step": 1700
    },
    {
      "epoch": 0.2565417389337154,
      "grad_norm": 0.24395037223746055,
      "learning_rate": 4.307860709481439e-07,
      "loss": 1.5249,
      "step": 1701
    },
    {
      "epoch": 0.256692557122389,
      "grad_norm": 0.2517507323798212,
      "learning_rate": 4.30709127378118e-07,
      "loss": 1.6236,
      "step": 1702
    },
    {
      "epoch": 0.2568433753110625,
      "grad_norm": 0.2564053341185109,
      "learning_rate": 4.306321488468583e-07,
      "loss": 1.5775,
      "step": 1703
    },
    {
      "epoch": 0.25699419349973607,
      "grad_norm": 0.24955879072954779,
      "learning_rate": 4.3055513537164856e-07,
      "loss": 1.5523,
      "step": 1704
    },
    {
      "epoch": 0.25714501168840964,
      "grad_norm": 0.24512985002257073,
      "learning_rate": 4.304780869697807e-07,
      "loss": 1.5565,
      "step": 1705
    },
    {
      "epoch": 0.25729582987708316,
      "grad_norm": 0.23499944533232645,
      "learning_rate": 4.3040100365855443e-07,
      "loss": 1.5503,
      "step": 1706
    },
    {
      "epoch": 0.25744664806575673,
      "grad_norm": 0.2566829771396666,
      "learning_rate": 4.3032388545527694e-07,
      "loss": 1.5505,
      "step": 1707
    },
    {
      "epoch": 0.2575974662544303,
      "grad_norm": 0.24099414263993085,
      "learning_rate": 4.302467323772637e-07,
      "loss": 1.5022,
      "step": 1708
    },
    {
      "epoch": 0.2577482844431038,
      "grad_norm": 0.24235084453227237,
      "learning_rate": 4.3016954444183767e-07,
      "loss": 1.5593,
      "step": 1709
    },
    {
      "epoch": 0.2578991026317774,
      "grad_norm": 0.4725987508836748,
      "learning_rate": 4.300923216663299e-07,
      "loss": 1.5162,
      "step": 1710
    },
    {
      "epoch": 0.258049920820451,
      "grad_norm": 0.2661850939605425,
      "learning_rate": 4.3001506406807913e-07,
      "loss": 1.5795,
      "step": 1711
    },
    {
      "epoch": 0.2582007390091245,
      "grad_norm": 0.2515271806280741,
      "learning_rate": 4.29937771664432e-07,
      "loss": 1.62,
      "step": 1712
    },
    {
      "epoch": 0.25835155719779807,
      "grad_norm": 0.24030278635856814,
      "learning_rate": 4.2986044447274273e-07,
      "loss": 1.6004,
      "step": 1713
    },
    {
      "epoch": 0.2585023753864716,
      "grad_norm": 0.26136403601121966,
      "learning_rate": 4.2978308251037376e-07,
      "loss": 1.5062,
      "step": 1714
    },
    {
      "epoch": 0.25865319357514516,
      "grad_norm": 0.25022243826164337,
      "learning_rate": 4.297056857946949e-07,
      "loss": 1.5679,
      "step": 1715
    },
    {
      "epoch": 0.25880401176381873,
      "grad_norm": 0.238469988701716,
      "learning_rate": 4.2962825434308406e-07,
      "loss": 1.6042,
      "step": 1716
    },
    {
      "epoch": 0.25895482995249225,
      "grad_norm": 0.2593833044846852,
      "learning_rate": 4.295507881729269e-07,
      "loss": 1.5535,
      "step": 1717
    },
    {
      "epoch": 0.2591056481411658,
      "grad_norm": 0.25348258747119135,
      "learning_rate": 4.2947328730161674e-07,
      "loss": 1.5699,
      "step": 1718
    },
    {
      "epoch": 0.2592564663298394,
      "grad_norm": 0.2515133377116682,
      "learning_rate": 4.2939575174655484e-07,
      "loss": 1.621,
      "step": 1719
    },
    {
      "epoch": 0.2594072845185129,
      "grad_norm": 0.30768284674184804,
      "learning_rate": 4.2931818152515016e-07,
      "loss": 1.5851,
      "step": 1720
    },
    {
      "epoch": 0.2595581027071865,
      "grad_norm": 0.24333286855773978,
      "learning_rate": 4.292405766548195e-07,
      "loss": 1.6054,
      "step": 1721
    },
    {
      "epoch": 0.25970892089586006,
      "grad_norm": 0.2450978395943358,
      "learning_rate": 4.291629371529874e-07,
      "loss": 1.602,
      "step": 1722
    },
    {
      "epoch": 0.2598597390845336,
      "grad_norm": 0.3485855472756898,
      "learning_rate": 4.2908526303708613e-07,
      "loss": 1.5424,
      "step": 1723
    },
    {
      "epoch": 0.26001055727320715,
      "grad_norm": 0.24658756574185361,
      "learning_rate": 4.29007554324556e-07,
      "loss": 1.5697,
      "step": 1724
    },
    {
      "epoch": 0.26016137546188073,
      "grad_norm": 0.24562117457221536,
      "learning_rate": 4.2892981103284457e-07,
      "loss": 1.4849,
      "step": 1725
    },
    {
      "epoch": 0.26031219365055425,
      "grad_norm": 0.2919432368155901,
      "learning_rate": 4.2885203317940766e-07,
      "loss": 1.6167,
      "step": 1726
    },
    {
      "epoch": 0.2604630118392278,
      "grad_norm": 0.2379643579981179,
      "learning_rate": 4.287742207817087e-07,
      "loss": 1.5943,
      "step": 1727
    },
    {
      "epoch": 0.26061383002790134,
      "grad_norm": 0.27234287100086396,
      "learning_rate": 4.286963738572187e-07,
      "loss": 1.5307,
      "step": 1728
    },
    {
      "epoch": 0.2607646482165749,
      "grad_norm": 0.23794033215323887,
      "learning_rate": 4.286184924234167e-07,
      "loss": 1.5633,
      "step": 1729
    },
    {
      "epoch": 0.2609154664052485,
      "grad_norm": 0.24534584613674915,
      "learning_rate": 4.285405764977892e-07,
      "loss": 1.5393,
      "step": 1730
    },
    {
      "epoch": 0.261066284593922,
      "grad_norm": 0.24622183805830222,
      "learning_rate": 4.2846262609783083e-07,
      "loss": 1.5555,
      "step": 1731
    },
    {
      "epoch": 0.2612171027825956,
      "grad_norm": 0.2473139121117716,
      "learning_rate": 4.283846412410435e-07,
      "loss": 1.558,
      "step": 1732
    },
    {
      "epoch": 0.26136792097126915,
      "grad_norm": 0.2503546090239365,
      "learning_rate": 4.2830662194493707e-07,
      "loss": 1.5865,
      "step": 1733
    },
    {
      "epoch": 0.26151873915994267,
      "grad_norm": 0.24329780657648137,
      "learning_rate": 4.2822856822702935e-07,
      "loss": 1.5058,
      "step": 1734
    },
    {
      "epoch": 0.26166955734861624,
      "grad_norm": 4.367126003159523,
      "learning_rate": 4.281504801048455e-07,
      "loss": 1.5286,
      "step": 1735
    },
    {
      "epoch": 0.2618203755372898,
      "grad_norm": 0.24525797410772415,
      "learning_rate": 4.2807235759591867e-07,
      "loss": 1.5335,
      "step": 1736
    },
    {
      "epoch": 0.26197119372596334,
      "grad_norm": 0.25112320477786404,
      "learning_rate": 4.279942007177896e-07,
      "loss": 1.5799,
      "step": 1737
    },
    {
      "epoch": 0.2621220119146369,
      "grad_norm": 0.25664626588007966,
      "learning_rate": 4.279160094880067e-07,
      "loss": 1.5918,
      "step": 1738
    },
    {
      "epoch": 0.2622728301033105,
      "grad_norm": 0.24580238927716838,
      "learning_rate": 4.278377839241263e-07,
      "loss": 1.5439,
      "step": 1739
    },
    {
      "epoch": 0.262423648291984,
      "grad_norm": 0.24261438363071613,
      "learning_rate": 4.2775952404371223e-07,
      "loss": 1.5971,
      "step": 1740
    },
    {
      "epoch": 0.2625744664806576,
      "grad_norm": 0.2479096132206566,
      "learning_rate": 4.276812298643363e-07,
      "loss": 1.5357,
      "step": 1741
    },
    {
      "epoch": 0.2627252846693311,
      "grad_norm": 0.24731093696997355,
      "learning_rate": 4.276029014035776e-07,
      "loss": 1.5494,
      "step": 1742
    },
    {
      "epoch": 0.26287610285800467,
      "grad_norm": 0.2493656978258735,
      "learning_rate": 4.275245386790233e-07,
      "loss": 1.567,
      "step": 1743
    },
    {
      "epoch": 0.26302692104667824,
      "grad_norm": 0.2580901877251635,
      "learning_rate": 4.27446141708268e-07,
      "loss": 1.6034,
      "step": 1744
    },
    {
      "epoch": 0.26317773923535176,
      "grad_norm": 0.2507352232466474,
      "learning_rate": 4.273677105089143e-07,
      "loss": 1.6311,
      "step": 1745
    },
    {
      "epoch": 0.26332855742402533,
      "grad_norm": 0.2702052480530884,
      "learning_rate": 4.27289245098572e-07,
      "loss": 1.5518,
      "step": 1746
    },
    {
      "epoch": 0.2634793756126989,
      "grad_norm": 0.23892961225211057,
      "learning_rate": 4.2721074549485907e-07,
      "loss": 1.5522,
      "step": 1747
    },
    {
      "epoch": 0.2636301938013724,
      "grad_norm": 2.167013539159077,
      "learning_rate": 4.271322117154009e-07,
      "loss": 1.5226,
      "step": 1748
    },
    {
      "epoch": 0.263781011990046,
      "grad_norm": 0.28077461392710434,
      "learning_rate": 4.2705364377783066e-07,
      "loss": 1.5568,
      "step": 1749
    },
    {
      "epoch": 0.2639318301787196,
      "grad_norm": 0.24046928533669656,
      "learning_rate": 4.26975041699789e-07,
      "loss": 1.5687,
      "step": 1750
    },
    {
      "epoch": 0.2640826483673931,
      "grad_norm": 0.25338651586610195,
      "learning_rate": 4.268964054989246e-07,
      "loss": 1.5549,
      "step": 1751
    },
    {
      "epoch": 0.26423346655606667,
      "grad_norm": 0.2631495053728446,
      "learning_rate": 4.2681773519289343e-07,
      "loss": 1.5547,
      "step": 1752
    },
    {
      "epoch": 0.26438428474474024,
      "grad_norm": 0.4099460775581195,
      "learning_rate": 4.267390307993592e-07,
      "loss": 1.5794,
      "step": 1753
    },
    {
      "epoch": 0.26453510293341376,
      "grad_norm": 0.2541266674653664,
      "learning_rate": 4.2666029233599355e-07,
      "loss": 1.5892,
      "step": 1754
    },
    {
      "epoch": 0.26468592112208733,
      "grad_norm": 0.25026334883655993,
      "learning_rate": 4.2658151982047534e-07,
      "loss": 1.5842,
      "step": 1755
    },
    {
      "epoch": 0.2648367393107609,
      "grad_norm": 0.2397883491629975,
      "learning_rate": 4.2650271327049146e-07,
      "loss": 1.589,
      "step": 1756
    },
    {
      "epoch": 0.2649875574994344,
      "grad_norm": 0.27441100966125137,
      "learning_rate": 4.2642387270373614e-07,
      "loss": 1.617,
      "step": 1757
    },
    {
      "epoch": 0.265138375688108,
      "grad_norm": 0.23639191026103049,
      "learning_rate": 4.263449981379115e-07,
      "loss": 1.517,
      "step": 1758
    },
    {
      "epoch": 0.2652891938767815,
      "grad_norm": 0.2699309144334069,
      "learning_rate": 4.262660895907271e-07,
      "loss": 1.6319,
      "step": 1759
    },
    {
      "epoch": 0.2654400120654551,
      "grad_norm": 0.2445276874634692,
      "learning_rate": 4.2618714707990015e-07,
      "loss": 1.5247,
      "step": 1760
    },
    {
      "epoch": 0.26559083025412866,
      "grad_norm": 0.25528482755111215,
      "learning_rate": 4.2610817062315576e-07,
      "loss": 1.5657,
      "step": 1761
    },
    {
      "epoch": 0.2657416484428022,
      "grad_norm": 0.27608493129247075,
      "learning_rate": 4.2602916023822625e-07,
      "loss": 1.5573,
      "step": 1762
    },
    {
      "epoch": 0.26589246663147575,
      "grad_norm": 0.2757447732664978,
      "learning_rate": 4.259501159428518e-07,
      "loss": 1.5814,
      "step": 1763
    },
    {
      "epoch": 0.26604328482014933,
      "grad_norm": 0.260561028735259,
      "learning_rate": 4.2587103775478025e-07,
      "loss": 1.641,
      "step": 1764
    },
    {
      "epoch": 0.26619410300882285,
      "grad_norm": 0.26360915418753517,
      "learning_rate": 4.257919256917668e-07,
      "loss": 1.5619,
      "step": 1765
    },
    {
      "epoch": 0.2663449211974964,
      "grad_norm": 0.2590646826792949,
      "learning_rate": 4.257127797715744e-07,
      "loss": 1.5823,
      "step": 1766
    },
    {
      "epoch": 0.26649573938617,
      "grad_norm": 0.2832432560993926,
      "learning_rate": 4.256336000119738e-07,
      "loss": 1.5921,
      "step": 1767
    },
    {
      "epoch": 0.2666465575748435,
      "grad_norm": 0.2638014243597412,
      "learning_rate": 4.255543864307431e-07,
      "loss": 1.5678,
      "step": 1768
    },
    {
      "epoch": 0.2667973757635171,
      "grad_norm": 0.5282768986208761,
      "learning_rate": 4.2547513904566796e-07,
      "loss": 1.676,
      "step": 1769
    },
    {
      "epoch": 0.26694819395219066,
      "grad_norm": 0.2510764884728056,
      "learning_rate": 4.253958578745418e-07,
      "loss": 1.5789,
      "step": 1770
    },
    {
      "epoch": 0.2670990121408642,
      "grad_norm": 0.25717667833243285,
      "learning_rate": 4.2531654293516556e-07,
      "loss": 1.5463,
      "step": 1771
    },
    {
      "epoch": 0.26724983032953775,
      "grad_norm": 0.28771594463963907,
      "learning_rate": 4.2523719424534765e-07,
      "loss": 1.5332,
      "step": 1772
    },
    {
      "epoch": 0.26740064851821127,
      "grad_norm": 0.2608554228521943,
      "learning_rate": 4.251578118229043e-07,
      "loss": 1.6115,
      "step": 1773
    },
    {
      "epoch": 0.26755146670688484,
      "grad_norm": 0.26476231982666265,
      "learning_rate": 4.2507839568565914e-07,
      "loss": 1.6156,
      "step": 1774
    },
    {
      "epoch": 0.2677022848955584,
      "grad_norm": 0.2543292846190255,
      "learning_rate": 4.249989458514433e-07,
      "loss": 1.545,
      "step": 1775
    },
    {
      "epoch": 0.26785310308423194,
      "grad_norm": 0.24208442426546845,
      "learning_rate": 4.249194623380957e-07,
      "loss": 1.5592,
      "step": 1776
    },
    {
      "epoch": 0.2680039212729055,
      "grad_norm": 0.24415812731000183,
      "learning_rate": 4.248399451634627e-07,
      "loss": 1.5207,
      "step": 1777
    },
    {
      "epoch": 0.2681547394615791,
      "grad_norm": 0.3047787053782849,
      "learning_rate": 4.247603943453982e-07,
      "loss": 1.5716,
      "step": 1778
    },
    {
      "epoch": 0.2683055576502526,
      "grad_norm": 0.24046780380561628,
      "learning_rate": 4.246808099017636e-07,
      "loss": 1.5271,
      "step": 1779
    },
    {
      "epoch": 0.2684563758389262,
      "grad_norm": 0.24969244713149716,
      "learning_rate": 4.246011918504281e-07,
      "loss": 1.55,
      "step": 1780
    },
    {
      "epoch": 0.26860719402759975,
      "grad_norm": 0.2506095835537955,
      "learning_rate": 4.24521540209268e-07,
      "loss": 1.5318,
      "step": 1781
    },
    {
      "epoch": 0.26875801221627327,
      "grad_norm": 0.2591988512992042,
      "learning_rate": 4.2444185499616777e-07,
      "loss": 1.5703,
      "step": 1782
    },
    {
      "epoch": 0.26890883040494684,
      "grad_norm": 0.25705381864614046,
      "learning_rate": 4.243621362290187e-07,
      "loss": 1.5615,
      "step": 1783
    },
    {
      "epoch": 0.2690596485936204,
      "grad_norm": 0.3141365686214411,
      "learning_rate": 4.2428238392572014e-07,
      "loss": 1.6309,
      "step": 1784
    },
    {
      "epoch": 0.26921046678229393,
      "grad_norm": 0.272157584434689,
      "learning_rate": 4.242025981041789e-07,
      "loss": 1.5744,
      "step": 1785
    },
    {
      "epoch": 0.2693612849709675,
      "grad_norm": 0.5842370568199183,
      "learning_rate": 4.2412277878230913e-07,
      "loss": 1.5694,
      "step": 1786
    },
    {
      "epoch": 0.269512103159641,
      "grad_norm": 0.23365515570093115,
      "learning_rate": 4.240429259780325e-07,
      "loss": 1.5745,
      "step": 1787
    },
    {
      "epoch": 0.2696629213483146,
      "grad_norm": 0.25109599306290425,
      "learning_rate": 4.2396303970927845e-07,
      "loss": 1.5455,
      "step": 1788
    },
    {
      "epoch": 0.2698137395369882,
      "grad_norm": 0.2501155339934986,
      "learning_rate": 4.238831199939837e-07,
      "loss": 1.5548,
      "step": 1789
    },
    {
      "epoch": 0.2699645577256617,
      "grad_norm": 0.2648447708666371,
      "learning_rate": 4.2380316685009256e-07,
      "loss": 1.6197,
      "step": 1790
    },
    {
      "epoch": 0.27011537591433527,
      "grad_norm": 0.2644931125098561,
      "learning_rate": 4.237231802955568e-07,
      "loss": 1.5647,
      "step": 1791
    },
    {
      "epoch": 0.27026619410300884,
      "grad_norm": 0.30971024477150005,
      "learning_rate": 4.236431603483357e-07,
      "loss": 1.6217,
      "step": 1792
    },
    {
      "epoch": 0.27041701229168236,
      "grad_norm": 0.26442207873245427,
      "learning_rate": 4.2356310702639626e-07,
      "loss": 1.5302,
      "step": 1793
    },
    {
      "epoch": 0.27056783048035593,
      "grad_norm": 0.2529394164738092,
      "learning_rate": 4.234830203477126e-07,
      "loss": 1.6298,
      "step": 1794
    },
    {
      "epoch": 0.2707186486690295,
      "grad_norm": 0.24635054290731914,
      "learning_rate": 4.2340290033026654e-07,
      "loss": 1.5682,
      "step": 1795
    },
    {
      "epoch": 0.270869466857703,
      "grad_norm": 0.2509760402983561,
      "learning_rate": 4.233227469920474e-07,
      "loss": 1.5543,
      "step": 1796
    },
    {
      "epoch": 0.2710202850463766,
      "grad_norm": 0.24652129894552294,
      "learning_rate": 4.23242560351052e-07,
      "loss": 1.5764,
      "step": 1797
    },
    {
      "epoch": 0.27117110323505017,
      "grad_norm": 0.24762894141823835,
      "learning_rate": 4.231623404252845e-07,
      "loss": 1.5256,
      "step": 1798
    },
    {
      "epoch": 0.2713219214237237,
      "grad_norm": 0.3218613310706804,
      "learning_rate": 4.230820872327565e-07,
      "loss": 1.5912,
      "step": 1799
    },
    {
      "epoch": 0.27147273961239726,
      "grad_norm": 0.25064303095633134,
      "learning_rate": 4.2300180079148726e-07,
      "loss": 1.5687,
      "step": 1800
    },
    {
      "epoch": 0.27162355780107084,
      "grad_norm": 0.2535822235378952,
      "learning_rate": 4.229214811195036e-07,
      "loss": 1.5286,
      "step": 1801
    },
    {
      "epoch": 0.27177437598974435,
      "grad_norm": 0.4217560376619295,
      "learning_rate": 4.228411282348394e-07,
      "loss": 1.526,
      "step": 1802
    },
    {
      "epoch": 0.27192519417841793,
      "grad_norm": 0.246226672236237,
      "learning_rate": 4.227607421555363e-07,
      "loss": 1.6019,
      "step": 1803
    },
    {
      "epoch": 0.27207601236709145,
      "grad_norm": 0.2568678331029727,
      "learning_rate": 4.2268032289964336e-07,
      "loss": 1.5312,
      "step": 1804
    },
    {
      "epoch": 0.272226830555765,
      "grad_norm": 0.24918046149982423,
      "learning_rate": 4.22599870485217e-07,
      "loss": 1.5058,
      "step": 1805
    },
    {
      "epoch": 0.2723776487444386,
      "grad_norm": 0.24894166724012773,
      "learning_rate": 4.2251938493032105e-07,
      "loss": 1.5814,
      "step": 1806
    },
    {
      "epoch": 0.2725284669331121,
      "grad_norm": 0.25993011394727705,
      "learning_rate": 4.2243886625302704e-07,
      "loss": 1.5733,
      "step": 1807
    },
    {
      "epoch": 0.2726792851217857,
      "grad_norm": 0.27306733568102737,
      "learning_rate": 4.2235831447141365e-07,
      "loss": 1.5702,
      "step": 1808
    },
    {
      "epoch": 0.27283010331045926,
      "grad_norm": 0.25989108875404365,
      "learning_rate": 4.2227772960356705e-07,
      "loss": 1.5981,
      "step": 1809
    },
    {
      "epoch": 0.2729809214991328,
      "grad_norm": 0.26149823478332435,
      "learning_rate": 4.2219711166758103e-07,
      "loss": 1.5908,
      "step": 1810
    },
    {
      "epoch": 0.27313173968780635,
      "grad_norm": 0.26026579015433193,
      "learning_rate": 4.2211646068155653e-07,
      "loss": 1.5659,
      "step": 1811
    },
    {
      "epoch": 0.2732825578764799,
      "grad_norm": 0.2786267119059909,
      "learning_rate": 4.220357766636021e-07,
      "loss": 1.4934,
      "step": 1812
    },
    {
      "epoch": 0.27343337606515344,
      "grad_norm": 0.24873854300957907,
      "learning_rate": 4.2195505963183366e-07,
      "loss": 1.5366,
      "step": 1813
    },
    {
      "epoch": 0.273584194253827,
      "grad_norm": 0.2450054811517773,
      "learning_rate": 4.218743096043746e-07,
      "loss": 1.5071,
      "step": 1814
    },
    {
      "epoch": 0.2737350124425006,
      "grad_norm": 0.24452395711635164,
      "learning_rate": 4.217935265993555e-07,
      "loss": 1.635,
      "step": 1815
    },
    {
      "epoch": 0.2738858306311741,
      "grad_norm": 0.26799091446406315,
      "learning_rate": 4.217127106349146e-07,
      "loss": 1.5429,
      "step": 1816
    },
    {
      "epoch": 0.2740366488198477,
      "grad_norm": 0.2529320540608072,
      "learning_rate": 4.216318617291974e-07,
      "loss": 1.5548,
      "step": 1817
    },
    {
      "epoch": 0.2741874670085212,
      "grad_norm": 0.3240342155926816,
      "learning_rate": 4.2155097990035686e-07,
      "loss": 1.4814,
      "step": 1818
    },
    {
      "epoch": 0.2743382851971948,
      "grad_norm": 0.25928057871799937,
      "learning_rate": 4.2147006516655336e-07,
      "loss": 1.6202,
      "step": 1819
    },
    {
      "epoch": 0.27448910338586835,
      "grad_norm": 0.2637460831085882,
      "learning_rate": 4.2138911754595447e-07,
      "loss": 1.5331,
      "step": 1820
    },
    {
      "epoch": 0.27463992157454187,
      "grad_norm": 0.40213844229973383,
      "learning_rate": 4.213081370567354e-07,
      "loss": 1.5896,
      "step": 1821
    },
    {
      "epoch": 0.27479073976321544,
      "grad_norm": 0.2635155410550008,
      "learning_rate": 4.2122712371707857e-07,
      "loss": 1.5273,
      "step": 1822
    },
    {
      "epoch": 0.274941557951889,
      "grad_norm": 0.251759146208383,
      "learning_rate": 4.2114607754517385e-07,
      "loss": 1.5721,
      "step": 1823
    },
    {
      "epoch": 0.27509237614056253,
      "grad_norm": 0.2390934984695804,
      "learning_rate": 4.210649985592186e-07,
      "loss": 1.5836,
      "step": 1824
    },
    {
      "epoch": 0.2752431943292361,
      "grad_norm": 0.24859484293238596,
      "learning_rate": 4.2098388677741707e-07,
      "loss": 1.6303,
      "step": 1825
    },
    {
      "epoch": 0.2753940125179097,
      "grad_norm": 0.3789595946124548,
      "learning_rate": 4.2090274221798166e-07,
      "loss": 1.7072,
      "step": 1826
    },
    {
      "epoch": 0.2755448307065832,
      "grad_norm": 0.2602972456576551,
      "learning_rate": 4.2082156489913134e-07,
      "loss": 1.6016,
      "step": 1827
    },
    {
      "epoch": 0.2756956488952568,
      "grad_norm": 0.2697359521626207,
      "learning_rate": 4.207403548390929e-07,
      "loss": 1.5317,
      "step": 1828
    },
    {
      "epoch": 0.27584646708393035,
      "grad_norm": 0.2568251490608846,
      "learning_rate": 4.2065911205610044e-07,
      "loss": 1.4874,
      "step": 1829
    },
    {
      "epoch": 0.27599728527260387,
      "grad_norm": 0.2511200614946822,
      "learning_rate": 4.2057783656839517e-07,
      "loss": 1.6139,
      "step": 1830
    },
    {
      "epoch": 0.27614810346127744,
      "grad_norm": 0.24320173272634724,
      "learning_rate": 4.20496528394226e-07,
      "loss": 1.6247,
      "step": 1831
    },
    {
      "epoch": 0.27629892164995096,
      "grad_norm": 0.28999189566860223,
      "learning_rate": 4.2041518755184886e-07,
      "loss": 1.5557,
      "step": 1832
    },
    {
      "epoch": 0.27644973983862453,
      "grad_norm": 0.24945551894216397,
      "learning_rate": 4.203338140595271e-07,
      "loss": 1.6079,
      "step": 1833
    },
    {
      "epoch": 0.2766005580272981,
      "grad_norm": 0.24195153170569705,
      "learning_rate": 4.2025240793553155e-07,
      "loss": 1.5163,
      "step": 1834
    },
    {
      "epoch": 0.2767513762159716,
      "grad_norm": 0.258263074146948,
      "learning_rate": 4.201709691981402e-07,
      "loss": 1.4975,
      "step": 1835
    },
    {
      "epoch": 0.2769021944046452,
      "grad_norm": 0.3224877359796225,
      "learning_rate": 4.2008949786563837e-07,
      "loss": 1.5915,
      "step": 1836
    },
    {
      "epoch": 0.27705301259331877,
      "grad_norm": 0.24634862739392877,
      "learning_rate": 4.200079939563188e-07,
      "loss": 1.5729,
      "step": 1837
    },
    {
      "epoch": 0.2772038307819923,
      "grad_norm": 0.25703866871533637,
      "learning_rate": 4.1992645748848156e-07,
      "loss": 1.5443,
      "step": 1838
    },
    {
      "epoch": 0.27735464897066586,
      "grad_norm": 0.9585138467259852,
      "learning_rate": 4.198448884804338e-07,
      "loss": 1.5301,
      "step": 1839
    },
    {
      "epoch": 0.27750546715933944,
      "grad_norm": 0.26208245329281477,
      "learning_rate": 4.197632869504903e-07,
      "loss": 1.569,
      "step": 1840
    },
    {
      "epoch": 0.27765628534801295,
      "grad_norm": 0.2513114884655622,
      "learning_rate": 4.196816529169729e-07,
      "loss": 1.605,
      "step": 1841
    },
    {
      "epoch": 0.27780710353668653,
      "grad_norm": 0.2530136474115613,
      "learning_rate": 4.1959998639821076e-07,
      "loss": 1.5764,
      "step": 1842
    },
    {
      "epoch": 0.2779579217253601,
      "grad_norm": 0.2757870606707001,
      "learning_rate": 4.195182874125405e-07,
      "loss": 1.6069,
      "step": 1843
    },
    {
      "epoch": 0.2781087399140336,
      "grad_norm": 0.25075678759563796,
      "learning_rate": 4.1943655597830585e-07,
      "loss": 1.4765,
      "step": 1844
    },
    {
      "epoch": 0.2782595581027072,
      "grad_norm": 0.24819323725669667,
      "learning_rate": 4.1935479211385793e-07,
      "loss": 1.5187,
      "step": 1845
    },
    {
      "epoch": 0.27841037629138077,
      "grad_norm": 0.2539600806747054,
      "learning_rate": 4.192729958375551e-07,
      "loss": 1.525,
      "step": 1846
    },
    {
      "epoch": 0.2785611944800543,
      "grad_norm": 0.2500165991563776,
      "learning_rate": 4.1919116716776306e-07,
      "loss": 1.5998,
      "step": 1847
    },
    {
      "epoch": 0.27871201266872786,
      "grad_norm": 0.25093956805624584,
      "learning_rate": 4.1910930612285463e-07,
      "loss": 1.5428,
      "step": 1848
    },
    {
      "epoch": 0.2788628308574014,
      "grad_norm": 0.24599969788501466,
      "learning_rate": 4.190274127212101e-07,
      "loss": 1.5203,
      "step": 1849
    },
    {
      "epoch": 0.27901364904607495,
      "grad_norm": 0.24701840386671767,
      "learning_rate": 4.189454869812168e-07,
      "loss": 1.5905,
      "step": 1850
    },
    {
      "epoch": 0.2791644672347485,
      "grad_norm": 0.2593434020338551,
      "learning_rate": 4.188635289212696e-07,
      "loss": 1.5726,
      "step": 1851
    },
    {
      "epoch": 0.27931528542342204,
      "grad_norm": 0.2433982875221031,
      "learning_rate": 4.1878153855977026e-07,
      "loss": 1.5422,
      "step": 1852
    },
    {
      "epoch": 0.2794661036120956,
      "grad_norm": 0.25178640775843025,
      "learning_rate": 4.1869951591512825e-07,
      "loss": 1.6041,
      "step": 1853
    },
    {
      "epoch": 0.2796169218007692,
      "grad_norm": 0.3112800150444886,
      "learning_rate": 4.186174610057599e-07,
      "loss": 1.6311,
      "step": 1854
    },
    {
      "epoch": 0.2797677399894427,
      "grad_norm": 0.34722227334108646,
      "learning_rate": 4.18535373850089e-07,
      "loss": 1.5713,
      "step": 1855
    },
    {
      "epoch": 0.2799185581781163,
      "grad_norm": 0.2556966025664008,
      "learning_rate": 4.184532544665463e-07,
      "loss": 1.6148,
      "step": 1856
    },
    {
      "epoch": 0.28006937636678986,
      "grad_norm": 0.25148191572401546,
      "learning_rate": 4.1837110287357036e-07,
      "loss": 1.5831,
      "step": 1857
    },
    {
      "epoch": 0.2802201945554634,
      "grad_norm": 0.3077566897785846,
      "learning_rate": 4.182889190896063e-07,
      "loss": 1.6374,
      "step": 1858
    },
    {
      "epoch": 0.28037101274413695,
      "grad_norm": 0.26809303377651905,
      "learning_rate": 4.182067031331068e-07,
      "loss": 1.5719,
      "step": 1859
    },
    {
      "epoch": 0.2805218309328105,
      "grad_norm": 0.2788609639999816,
      "learning_rate": 4.1812445502253187e-07,
      "loss": 1.6558,
      "step": 1860
    },
    {
      "epoch": 0.28067264912148404,
      "grad_norm": 0.2508605671959314,
      "learning_rate": 4.180421747763485e-07,
      "loss": 1.5351,
      "step": 1861
    },
    {
      "epoch": 0.2808234673101576,
      "grad_norm": 0.2575116015673467,
      "learning_rate": 4.179598624130311e-07,
      "loss": 1.5704,
      "step": 1862
    },
    {
      "epoch": 0.28097428549883113,
      "grad_norm": 0.26051246201496414,
      "learning_rate": 4.178775179510611e-07,
      "loss": 1.5248,
      "step": 1863
    },
    {
      "epoch": 0.2811251036875047,
      "grad_norm": 0.3062813945889822,
      "learning_rate": 4.177951414089273e-07,
      "loss": 1.55,
      "step": 1864
    },
    {
      "epoch": 0.2812759218761783,
      "grad_norm": 0.2452833635928116,
      "learning_rate": 4.1771273280512563e-07,
      "loss": 1.5735,
      "step": 1865
    },
    {
      "epoch": 0.2814267400648518,
      "grad_norm": 0.3323851621699343,
      "learning_rate": 4.1763029215815917e-07,
      "loss": 1.6476,
      "step": 1866
    },
    {
      "epoch": 0.2815775582535254,
      "grad_norm": 0.3002016764143178,
      "learning_rate": 4.175478194865382e-07,
      "loss": 1.5434,
      "step": 1867
    },
    {
      "epoch": 0.28172837644219895,
      "grad_norm": 0.24134420662883305,
      "learning_rate": 4.1746531480878034e-07,
      "loss": 1.5084,
      "step": 1868
    },
    {
      "epoch": 0.28187919463087246,
      "grad_norm": 0.25200348166629133,
      "learning_rate": 4.173827781434103e-07,
      "loss": 1.5788,
      "step": 1869
    },
    {
      "epoch": 0.28203001281954604,
      "grad_norm": 0.8112500246985829,
      "learning_rate": 4.1730020950895984e-07,
      "loss": 1.5803,
      "step": 1870
    },
    {
      "epoch": 0.2821808310082196,
      "grad_norm": 0.25242595654431865,
      "learning_rate": 4.172176089239682e-07,
      "loss": 1.5233,
      "step": 1871
    },
    {
      "epoch": 0.28233164919689313,
      "grad_norm": 0.2477708621373277,
      "learning_rate": 4.171349764069814e-07,
      "loss": 1.6228,
      "step": 1872
    },
    {
      "epoch": 0.2824824673855667,
      "grad_norm": 0.25166770520066656,
      "learning_rate": 4.170523119765531e-07,
      "loss": 1.5883,
      "step": 1873
    },
    {
      "epoch": 0.2826332855742403,
      "grad_norm": 0.2488566651756397,
      "learning_rate": 4.1696961565124365e-07,
      "loss": 1.6172,
      "step": 1874
    },
    {
      "epoch": 0.2827841037629138,
      "grad_norm": 0.26082171280059746,
      "learning_rate": 4.168868874496209e-07,
      "loss": 1.6535,
      "step": 1875
    },
    {
      "epoch": 0.28293492195158737,
      "grad_norm": 0.2574831982077194,
      "learning_rate": 4.1680412739025975e-07,
      "loss": 1.5699,
      "step": 1876
    },
    {
      "epoch": 0.2830857401402609,
      "grad_norm": 0.25686836612569874,
      "learning_rate": 4.1672133549174213e-07,
      "loss": 1.5909,
      "step": 1877
    },
    {
      "epoch": 0.28323655832893446,
      "grad_norm": 0.2626699964340152,
      "learning_rate": 4.166385117726574e-07,
      "loss": 1.5446,
      "step": 1878
    },
    {
      "epoch": 0.28338737651760804,
      "grad_norm": 0.2457572940259408,
      "learning_rate": 4.1655565625160175e-07,
      "loss": 1.6223,
      "step": 1879
    },
    {
      "epoch": 0.28353819470628155,
      "grad_norm": 0.2591728510568822,
      "learning_rate": 4.1647276894717877e-07,
      "loss": 1.5866,
      "step": 1880
    },
    {
      "epoch": 0.28368901289495513,
      "grad_norm": 0.2525946883907776,
      "learning_rate": 4.1638984987799896e-07,
      "loss": 1.6028,
      "step": 1881
    },
    {
      "epoch": 0.2838398310836287,
      "grad_norm": 0.26481841535374673,
      "learning_rate": 4.1630689906268015e-07,
      "loss": 1.5451,
      "step": 1882
    },
    {
      "epoch": 0.2839906492723022,
      "grad_norm": 0.26108338852677637,
      "learning_rate": 4.162239165198472e-07,
      "loss": 1.5494,
      "step": 1883
    },
    {
      "epoch": 0.2841414674609758,
      "grad_norm": 0.28379144963495145,
      "learning_rate": 4.161409022681321e-07,
      "loss": 1.5438,
      "step": 1884
    },
    {
      "epoch": 0.28429228564964937,
      "grad_norm": 0.24147918067444937,
      "learning_rate": 4.1605785632617395e-07,
      "loss": 1.5234,
      "step": 1885
    },
    {
      "epoch": 0.2844431038383229,
      "grad_norm": 0.2558554466858706,
      "learning_rate": 4.1597477871261894e-07,
      "loss": 1.5412,
      "step": 1886
    },
    {
      "epoch": 0.28459392202699646,
      "grad_norm": 0.24699696333450885,
      "learning_rate": 4.1589166944612057e-07,
      "loss": 1.5232,
      "step": 1887
    },
    {
      "epoch": 0.28474474021567003,
      "grad_norm": 0.253191918475815,
      "learning_rate": 4.1580852854533916e-07,
      "loss": 1.5336,
      "step": 1888
    },
    {
      "epoch": 0.28489555840434355,
      "grad_norm": 0.26486007911434506,
      "learning_rate": 4.157253560289423e-07,
      "loss": 1.4881,
      "step": 1889
    },
    {
      "epoch": 0.2850463765930171,
      "grad_norm": 0.24381429165526342,
      "learning_rate": 4.156421519156046e-07,
      "loss": 1.5644,
      "step": 1890
    },
    {
      "epoch": 0.2851971947816907,
      "grad_norm": 0.24278796646850606,
      "learning_rate": 4.155589162240078e-07,
      "loss": 1.5579,
      "step": 1891
    },
    {
      "epoch": 0.2853480129703642,
      "grad_norm": 0.27047678681380877,
      "learning_rate": 4.154756489728409e-07,
      "loss": 1.5339,
      "step": 1892
    },
    {
      "epoch": 0.2854988311590378,
      "grad_norm": 0.32759906068096833,
      "learning_rate": 4.153923501807996e-07,
      "loss": 1.6226,
      "step": 1893
    },
    {
      "epoch": 0.2856496493477113,
      "grad_norm": 0.850154673090816,
      "learning_rate": 4.1530901986658704e-07,
      "loss": 1.5314,
      "step": 1894
    },
    {
      "epoch": 0.2858004675363849,
      "grad_norm": 0.2445143331468664,
      "learning_rate": 4.1522565804891326e-07,
      "loss": 1.5958,
      "step": 1895
    },
    {
      "epoch": 0.28595128572505846,
      "grad_norm": 0.25907546127233794,
      "learning_rate": 4.1514226474649543e-07,
      "loss": 1.643,
      "step": 1896
    },
    {
      "epoch": 0.286102103913732,
      "grad_norm": 0.24135999646434758,
      "learning_rate": 4.150588399780578e-07,
      "loss": 1.5438,
      "step": 1897
    },
    {
      "epoch": 0.28625292210240555,
      "grad_norm": 0.247240957153842,
      "learning_rate": 4.1497538376233167e-07,
      "loss": 1.5548,
      "step": 1898
    },
    {
      "epoch": 0.2864037402910791,
      "grad_norm": 0.2702835083554822,
      "learning_rate": 4.1489189611805524e-07,
      "loss": 1.6004,
      "step": 1899
    },
    {
      "epoch": 0.28655455847975264,
      "grad_norm": 0.2459064321844834,
      "learning_rate": 4.148083770639741e-07,
      "loss": 1.6031,
      "step": 1900
    },
    {
      "epoch": 0.2867053766684262,
      "grad_norm": 0.3773518913970108,
      "learning_rate": 4.147248266188406e-07,
      "loss": 1.5511,
      "step": 1901
    },
    {
      "epoch": 0.2868561948570998,
      "grad_norm": 0.2619832946401977,
      "learning_rate": 4.146412448014143e-07,
      "loss": 1.5707,
      "step": 1902
    },
    {
      "epoch": 0.2870070130457733,
      "grad_norm": 0.25795252598404206,
      "learning_rate": 4.145576316304618e-07,
      "loss": 1.5677,
      "step": 1903
    },
    {
      "epoch": 0.2871578312344469,
      "grad_norm": 0.2598821459333334,
      "learning_rate": 4.1447398712475656e-07,
      "loss": 1.6072,
      "step": 1904
    },
    {
      "epoch": 0.28730864942312045,
      "grad_norm": 0.31528686521750887,
      "learning_rate": 4.1439031130307925e-07,
      "loss": 1.5934,
      "step": 1905
    },
    {
      "epoch": 0.287459467611794,
      "grad_norm": 0.33197145914876147,
      "learning_rate": 4.143066041842176e-07,
      "loss": 1.615,
      "step": 1906
    },
    {
      "epoch": 0.28761028580046755,
      "grad_norm": 0.2571601768464355,
      "learning_rate": 4.142228657869663e-07,
      "loss": 1.5988,
      "step": 1907
    },
    {
      "epoch": 0.28776110398914106,
      "grad_norm": 0.24326782318283038,
      "learning_rate": 4.14139096130127e-07,
      "loss": 1.5164,
      "step": 1908
    },
    {
      "epoch": 0.28791192217781464,
      "grad_norm": 0.25010534683464847,
      "learning_rate": 4.140552952325084e-07,
      "loss": 1.585,
      "step": 1909
    },
    {
      "epoch": 0.2880627403664882,
      "grad_norm": 0.2484450481343041,
      "learning_rate": 4.139714631129263e-07,
      "loss": 1.5448,
      "step": 1910
    },
    {
      "epoch": 0.28821355855516173,
      "grad_norm": 0.26683794510788705,
      "learning_rate": 4.1388759979020337e-07,
      "loss": 1.6287,
      "step": 1911
    },
    {
      "epoch": 0.2883643767438353,
      "grad_norm": 0.2549367342857066,
      "learning_rate": 4.1380370528316947e-07,
      "loss": 1.655,
      "step": 1912
    },
    {
      "epoch": 0.2885151949325089,
      "grad_norm": 0.2628214303418074,
      "learning_rate": 4.137197796106614e-07,
      "loss": 1.5859,
      "step": 1913
    },
    {
      "epoch": 0.2886660131211824,
      "grad_norm": 0.26071551769767043,
      "learning_rate": 4.1363582279152283e-07,
      "loss": 1.5769,
      "step": 1914
    },
    {
      "epoch": 0.28881683130985597,
      "grad_norm": 0.26176747168082637,
      "learning_rate": 4.1355183484460443e-07,
      "loss": 1.5321,
      "step": 1915
    },
    {
      "epoch": 0.28896764949852954,
      "grad_norm": 0.24734604568247337,
      "learning_rate": 4.134678157887641e-07,
      "loss": 1.5853,
      "step": 1916
    },
    {
      "epoch": 0.28911846768720306,
      "grad_norm": 0.259978618658322,
      "learning_rate": 4.1338376564286647e-07,
      "loss": 1.5809,
      "step": 1917
    },
    {
      "epoch": 0.28926928587587664,
      "grad_norm": 0.25786027074792967,
      "learning_rate": 4.132996844257833e-07,
      "loss": 1.5367,
      "step": 1918
    },
    {
      "epoch": 0.2894201040645502,
      "grad_norm": 0.2571241034906806,
      "learning_rate": 4.1321557215639317e-07,
      "loss": 1.5649,
      "step": 1919
    },
    {
      "epoch": 0.28957092225322373,
      "grad_norm": 0.2507150683383269,
      "learning_rate": 4.131314288535819e-07,
      "loss": 1.5159,
      "step": 1920
    },
    {
      "epoch": 0.2897217404418973,
      "grad_norm": 0.27552243362352963,
      "learning_rate": 4.1304725453624195e-07,
      "loss": 1.6037,
      "step": 1921
    },
    {
      "epoch": 0.2898725586305708,
      "grad_norm": 0.24515585991702715,
      "learning_rate": 4.12963049223273e-07,
      "loss": 1.5499,
      "step": 1922
    },
    {
      "epoch": 0.2900233768192444,
      "grad_norm": 0.26782050776616445,
      "learning_rate": 4.128788129335816e-07,
      "loss": 1.5409,
      "step": 1923
    },
    {
      "epoch": 0.29017419500791797,
      "grad_norm": 0.25978871302610407,
      "learning_rate": 4.1279454568608123e-07,
      "loss": 1.5393,
      "step": 1924
    },
    {
      "epoch": 0.2903250131965915,
      "grad_norm": 0.25351359568716264,
      "learning_rate": 4.127102474996923e-07,
      "loss": 1.5486,
      "step": 1925
    },
    {
      "epoch": 0.29047583138526506,
      "grad_norm": 0.24293422687620772,
      "learning_rate": 4.126259183933423e-07,
      "loss": 1.5937,
      "step": 1926
    },
    {
      "epoch": 0.29062664957393863,
      "grad_norm": 0.245471616869013,
      "learning_rate": 4.1254155838596544e-07,
      "loss": 1.5299,
      "step": 1927
    },
    {
      "epoch": 0.29077746776261215,
      "grad_norm": 0.25437820552331764,
      "learning_rate": 4.1245716749650305e-07,
      "loss": 1.5279,
      "step": 1928
    },
    {
      "epoch": 0.2909282859512857,
      "grad_norm": 0.2718417064207998,
      "learning_rate": 4.123727457439034e-07,
      "loss": 1.5289,
      "step": 1929
    },
    {
      "epoch": 0.2910791041399593,
      "grad_norm": 0.2415724984252487,
      "learning_rate": 4.122882931471216e-07,
      "loss": 1.563,
      "step": 1930
    },
    {
      "epoch": 0.2912299223286328,
      "grad_norm": 0.2927444262782482,
      "learning_rate": 4.122038097251197e-07,
      "loss": 1.5284,
      "step": 1931
    },
    {
      "epoch": 0.2913807405173064,
      "grad_norm": 0.27657048401816287,
      "learning_rate": 4.121192954968667e-07,
      "loss": 1.5759,
      "step": 1932
    },
    {
      "epoch": 0.29153155870597997,
      "grad_norm": 0.2630675188418442,
      "learning_rate": 4.1203475048133853e-07,
      "loss": 1.5694,
      "step": 1933
    },
    {
      "epoch": 0.2916823768946535,
      "grad_norm": 0.25207784287561097,
      "learning_rate": 4.119501746975179e-07,
      "loss": 1.5946,
      "step": 1934
    },
    {
      "epoch": 0.29183319508332706,
      "grad_norm": 0.29143907484127923,
      "learning_rate": 4.1186556816439476e-07,
      "loss": 1.5423,
      "step": 1935
    },
    {
      "epoch": 0.29198401327200063,
      "grad_norm": 0.25399531184844054,
      "learning_rate": 4.117809309009655e-07,
      "loss": 1.5576,
      "step": 1936
    },
    {
      "epoch": 0.29213483146067415,
      "grad_norm": 0.24292551587894717,
      "learning_rate": 4.116962629262338e-07,
      "loss": 1.5561,
      "step": 1937
    },
    {
      "epoch": 0.2922856496493477,
      "grad_norm": 0.26846283060694975,
      "learning_rate": 4.1161156425921e-07,
      "loss": 1.536,
      "step": 1938
    },
    {
      "epoch": 0.29243646783802124,
      "grad_norm": 0.25830246639286103,
      "learning_rate": 4.1152683491891146e-07,
      "loss": 1.5553,
      "step": 1939
    },
    {
      "epoch": 0.2925872860266948,
      "grad_norm": 0.4761048260200553,
      "learning_rate": 4.114420749243624e-07,
      "loss": 1.5477,
      "step": 1940
    },
    {
      "epoch": 0.2927381042153684,
      "grad_norm": 0.2632060076635442,
      "learning_rate": 4.1135728429459387e-07,
      "loss": 1.5913,
      "step": 1941
    },
    {
      "epoch": 0.2928889224040419,
      "grad_norm": 0.2607154117850426,
      "learning_rate": 4.112724630486439e-07,
      "loss": 1.5463,
      "step": 1942
    },
    {
      "epoch": 0.2930397405927155,
      "grad_norm": 0.24111840360689804,
      "learning_rate": 4.1118761120555725e-07,
      "loss": 1.5928,
      "step": 1943
    },
    {
      "epoch": 0.29319055878138905,
      "grad_norm": 0.2416204003895274,
      "learning_rate": 4.1110272878438567e-07,
      "loss": 1.5688,
      "step": 1944
    },
    {
      "epoch": 0.2933413769700626,
      "grad_norm": 0.27524436757956083,
      "learning_rate": 4.1101781580418764e-07,
      "loss": 1.5121,
      "step": 1945
    },
    {
      "epoch": 0.29349219515873615,
      "grad_norm": 0.24805067548995707,
      "learning_rate": 4.1093287228402886e-07,
      "loss": 1.5406,
      "step": 1946
    },
    {
      "epoch": 0.2936430133474097,
      "grad_norm": 0.2575731295687873,
      "learning_rate": 4.1084789824298125e-07,
      "loss": 1.5712,
      "step": 1947
    },
    {
      "epoch": 0.29379383153608324,
      "grad_norm": 0.6450012859290429,
      "learning_rate": 4.107628937001243e-07,
      "loss": 1.5549,
      "step": 1948
    },
    {
      "epoch": 0.2939446497247568,
      "grad_norm": 0.2611294321594528,
      "learning_rate": 4.106778586745438e-07,
      "loss": 1.6068,
      "step": 1949
    },
    {
      "epoch": 0.2940954679134304,
      "grad_norm": 0.47404066070592726,
      "learning_rate": 4.105927931853327e-07,
      "loss": 1.5262,
      "step": 1950
    },
    {
      "epoch": 0.2942462861021039,
      "grad_norm": 0.26325757681741707,
      "learning_rate": 4.1050769725159054e-07,
      "loss": 1.5558,
      "step": 1951
    },
    {
      "epoch": 0.2943971042907775,
      "grad_norm": 0.26210215015949584,
      "learning_rate": 4.10422570892424e-07,
      "loss": 1.5707,
      "step": 1952
    },
    {
      "epoch": 0.294547922479451,
      "grad_norm": 0.23889687523013875,
      "learning_rate": 4.1033741412694636e-07,
      "loss": 1.5696,
      "step": 1953
    },
    {
      "epoch": 0.29469874066812457,
      "grad_norm": 0.2864911607553577,
      "learning_rate": 4.1025222697427777e-07,
      "loss": 1.5221,
      "step": 1954
    },
    {
      "epoch": 0.29484955885679814,
      "grad_norm": 0.2507908685608037,
      "learning_rate": 4.1016700945354525e-07,
      "loss": 1.4778,
      "step": 1955
    },
    {
      "epoch": 0.29500037704547166,
      "grad_norm": 0.2679153163035175,
      "learning_rate": 4.100817615838826e-07,
      "loss": 1.5648,
      "step": 1956
    },
    {
      "epoch": 0.29515119523414524,
      "grad_norm": 0.24669025937295905,
      "learning_rate": 4.0999648338443045e-07,
      "loss": 1.5723,
      "step": 1957
    },
    {
      "epoch": 0.2953020134228188,
      "grad_norm": 0.26473189279096443,
      "learning_rate": 4.0991117487433623e-07,
      "loss": 1.5536,
      "step": 1958
    },
    {
      "epoch": 0.29545283161149233,
      "grad_norm": 0.2487310850412394,
      "learning_rate": 4.098258360727543e-07,
      "loss": 1.5793,
      "step": 1959
    },
    {
      "epoch": 0.2956036498001659,
      "grad_norm": 0.2587133676238718,
      "learning_rate": 4.097404669988455e-07,
      "loss": 1.4535,
      "step": 1960
    },
    {
      "epoch": 0.2957544679888395,
      "grad_norm": 0.2536359469005928,
      "learning_rate": 4.0965506767177784e-07,
      "loss": 1.5054,
      "step": 1961
    },
    {
      "epoch": 0.295905286177513,
      "grad_norm": 0.262082928095278,
      "learning_rate": 4.0956963811072583e-07,
      "loss": 1.598,
      "step": 1962
    },
    {
      "epoch": 0.29605610436618657,
      "grad_norm": 0.25440730089743857,
      "learning_rate": 4.0948417833487103e-07,
      "loss": 1.5339,
      "step": 1963
    },
    {
      "epoch": 0.29620692255486014,
      "grad_norm": 0.25822983100141933,
      "learning_rate": 4.093986883634016e-07,
      "loss": 1.6163,
      "step": 1964
    },
    {
      "epoch": 0.29635774074353366,
      "grad_norm": 0.24813816449687914,
      "learning_rate": 4.093131682155125e-07,
      "loss": 1.5453,
      "step": 1965
    },
    {
      "epoch": 0.29650855893220723,
      "grad_norm": 0.2548251273125417,
      "learning_rate": 4.092276179104055e-07,
      "loss": 1.5398,
      "step": 1966
    },
    {
      "epoch": 0.29665937712088075,
      "grad_norm": 0.3985691681710239,
      "learning_rate": 4.09142037467289e-07,
      "loss": 1.5861,
      "step": 1967
    },
    {
      "epoch": 0.2968101953095543,
      "grad_norm": 0.26433297032732417,
      "learning_rate": 4.090564269053787e-07,
      "loss": 1.5728,
      "step": 1968
    },
    {
      "epoch": 0.2969610134982279,
      "grad_norm": 0.23422264251292965,
      "learning_rate": 4.089707862438962e-07,
      "loss": 1.6059,
      "step": 1969
    },
    {
      "epoch": 0.2971118316869014,
      "grad_norm": 0.27084535372529556,
      "learning_rate": 4.088851155020706e-07,
      "loss": 1.6073,
      "step": 1970
    },
    {
      "epoch": 0.297262649875575,
      "grad_norm": 0.2821936345624914,
      "learning_rate": 4.087994146991374e-07,
      "loss": 1.518,
      "step": 1971
    },
    {
      "epoch": 0.29741346806424857,
      "grad_norm": 0.26212810945747056,
      "learning_rate": 4.087136838543389e-07,
      "loss": 1.5569,
      "step": 1972
    },
    {
      "epoch": 0.2975642862529221,
      "grad_norm": 0.2447975691364571,
      "learning_rate": 4.086279229869242e-07,
      "loss": 1.5593,
      "step": 1973
    },
    {
      "epoch": 0.29771510444159566,
      "grad_norm": 0.2419722057739497,
      "learning_rate": 4.0854213211614916e-07,
      "loss": 1.5567,
      "step": 1974
    },
    {
      "epoch": 0.29786592263026923,
      "grad_norm": 0.24653505228114833,
      "learning_rate": 4.084563112612763e-07,
      "loss": 1.5541,
      "step": 1975
    },
    {
      "epoch": 0.29801674081894275,
      "grad_norm": 0.2689128543871328,
      "learning_rate": 4.083704604415748e-07,
      "loss": 1.6205,
      "step": 1976
    },
    {
      "epoch": 0.2981675590076163,
      "grad_norm": 0.27997129697482137,
      "learning_rate": 4.0828457967632086e-07,
      "loss": 1.641,
      "step": 1977
    },
    {
      "epoch": 0.2983183771962899,
      "grad_norm": 0.25499025032094236,
      "learning_rate": 4.0819866898479703e-07,
      "loss": 1.5725,
      "step": 1978
    },
    {
      "epoch": 0.2984691953849634,
      "grad_norm": 0.5480214467709781,
      "learning_rate": 4.081127283862929e-07,
      "loss": 1.5784,
      "step": 1979
    },
    {
      "epoch": 0.298620013573637,
      "grad_norm": 0.2620739620572641,
      "learning_rate": 4.0802675790010456e-07,
      "loss": 1.4975,
      "step": 1980
    },
    {
      "epoch": 0.2987708317623105,
      "grad_norm": 0.25424429682397565,
      "learning_rate": 4.0794075754553483e-07,
      "loss": 1.5506,
      "step": 1981
    },
    {
      "epoch": 0.2989216499509841,
      "grad_norm": 0.5682722291288377,
      "learning_rate": 4.0785472734189345e-07,
      "loss": 1.548,
      "step": 1982
    },
    {
      "epoch": 0.29907246813965765,
      "grad_norm": 0.37582259605338,
      "learning_rate": 4.077686673084966e-07,
      "loss": 1.5605,
      "step": 1983
    },
    {
      "epoch": 0.2992232863283312,
      "grad_norm": 0.34117334933146093,
      "learning_rate": 4.076825774646672e-07,
      "loss": 1.5692,
      "step": 1984
    },
    {
      "epoch": 0.29937410451700475,
      "grad_norm": 0.24169476981631696,
      "learning_rate": 4.075964578297352e-07,
      "loss": 1.5554,
      "step": 1985
    },
    {
      "epoch": 0.2995249227056783,
      "grad_norm": 0.3749710955356763,
      "learning_rate": 4.0751030842303656e-07,
      "loss": 1.5426,
      "step": 1986
    },
    {
      "epoch": 0.29967574089435184,
      "grad_norm": 0.4484413453699465,
      "learning_rate": 4.0742412926391474e-07,
      "loss": 1.5726,
      "step": 1987
    },
    {
      "epoch": 0.2998265590830254,
      "grad_norm": 0.24642273347537877,
      "learning_rate": 4.0733792037171916e-07,
      "loss": 1.5898,
      "step": 1988
    },
    {
      "epoch": 0.299977377271699,
      "grad_norm": 0.29618837310677704,
      "learning_rate": 4.072516817658064e-07,
      "loss": 1.6294,
      "step": 1989
    },
    {
      "epoch": 0.3001281954603725,
      "grad_norm": 0.25474726896157884,
      "learning_rate": 4.071654134655395e-07,
      "loss": 1.5493,
      "step": 1990
    },
    {
      "epoch": 0.3002790136490461,
      "grad_norm": 0.7666324706571088,
      "learning_rate": 4.070791154902882e-07,
      "loss": 1.5685,
      "step": 1991
    },
    {
      "epoch": 0.30042983183771965,
      "grad_norm": 0.23909485923559612,
      "learning_rate": 4.069927878594289e-07,
      "loss": 1.5954,
      "step": 1992
    },
    {
      "epoch": 0.30058065002639317,
      "grad_norm": 0.25181643764237355,
      "learning_rate": 4.069064305923447e-07,
      "loss": 1.5721,
      "step": 1993
    },
    {
      "epoch": 0.30073146821506674,
      "grad_norm": 0.24995411296454376,
      "learning_rate": 4.068200437084252e-07,
      "loss": 1.5806,
      "step": 1994
    },
    {
      "epoch": 0.3008822864037403,
      "grad_norm": 0.2492558803326442,
      "learning_rate": 4.0673362722706695e-07,
      "loss": 1.6338,
      "step": 1995
    },
    {
      "epoch": 0.30103310459241384,
      "grad_norm": 0.24359118017630724,
      "learning_rate": 4.0664718116767287e-07,
      "loss": 1.6746,
      "step": 1996
    },
    {
      "epoch": 0.3011839227810874,
      "grad_norm": 0.25822056709052726,
      "learning_rate": 4.0656070554965257e-07,
      "loss": 1.5229,
      "step": 1997
    },
    {
      "epoch": 0.30133474096976093,
      "grad_norm": 0.25630204248834293,
      "learning_rate": 4.064742003924224e-07,
      "loss": 1.5693,
      "step": 1998
    },
    {
      "epoch": 0.3014855591584345,
      "grad_norm": 0.2815196438644947,
      "learning_rate": 4.063876657154054e-07,
      "loss": 1.5625,
      "step": 1999
    },
    {
      "epoch": 0.3016363773471081,
      "grad_norm": 0.24786544945527367,
      "learning_rate": 4.0630110153803094e-07,
      "loss": 1.5591,
      "step": 2000
    },
    {
      "epoch": 0.3017871955357816,
      "grad_norm": 0.24701583438940208,
      "learning_rate": 4.062145078797353e-07,
      "loss": 1.549,
      "step": 2001
    },
    {
      "epoch": 0.30193801372445517,
      "grad_norm": 0.25457518078584596,
      "learning_rate": 4.061278847599612e-07,
      "loss": 1.6315,
      "step": 2002
    },
    {
      "epoch": 0.30208883191312874,
      "grad_norm": 0.2360049490832213,
      "learning_rate": 4.060412321981582e-07,
      "loss": 1.5689,
      "step": 2003
    },
    {
      "epoch": 0.30223965010180226,
      "grad_norm": 0.2538901784411201,
      "learning_rate": 4.059545502137821e-07,
      "loss": 1.6141,
      "step": 2004
    },
    {
      "epoch": 0.30239046829047583,
      "grad_norm": 0.2611059042662772,
      "learning_rate": 4.0586783882629574e-07,
      "loss": 1.4949,
      "step": 2005
    },
    {
      "epoch": 0.3025412864791494,
      "grad_norm": 0.24314716429863004,
      "learning_rate": 4.0578109805516825e-07,
      "loss": 1.5134,
      "step": 2006
    },
    {
      "epoch": 0.3026921046678229,
      "grad_norm": 0.2744908505423212,
      "learning_rate": 4.0569432791987545e-07,
      "loss": 1.5875,
      "step": 2007
    },
    {
      "epoch": 0.3028429228564965,
      "grad_norm": 0.265288441657907,
      "learning_rate": 4.056075284398998e-07,
      "loss": 1.544,
      "step": 2008
    },
    {
      "epoch": 0.3029937410451701,
      "grad_norm": 0.24689543993443636,
      "learning_rate": 4.055206996347304e-07,
      "loss": 1.563,
      "step": 2009
    },
    {
      "epoch": 0.3031445592338436,
      "grad_norm": 0.2846226929353735,
      "learning_rate": 4.054338415238627e-07,
      "loss": 1.538,
      "step": 2010
    },
    {
      "epoch": 0.30329537742251717,
      "grad_norm": 0.25529054190064804,
      "learning_rate": 4.0534695412679885e-07,
      "loss": 1.5938,
      "step": 2011
    },
    {
      "epoch": 0.3034461956111907,
      "grad_norm": 0.26141624179768436,
      "learning_rate": 4.0526003746304776e-07,
      "loss": 1.6299,
      "step": 2012
    },
    {
      "epoch": 0.30359701379986426,
      "grad_norm": 0.2509275510566078,
      "learning_rate": 4.051730915521246e-07,
      "loss": 1.5485,
      "step": 2013
    },
    {
      "epoch": 0.30374783198853783,
      "grad_norm": 0.25921193285116084,
      "learning_rate": 4.0508611641355137e-07,
      "loss": 1.6706,
      "step": 2014
    },
    {
      "epoch": 0.30389865017721135,
      "grad_norm": 0.29053096237224685,
      "learning_rate": 4.049991120668565e-07,
      "loss": 1.5731,
      "step": 2015
    },
    {
      "epoch": 0.3040494683658849,
      "grad_norm": 0.2403332440983861,
      "learning_rate": 4.04912078531575e-07,
      "loss": 1.5128,
      "step": 2016
    },
    {
      "epoch": 0.3042002865545585,
      "grad_norm": 0.2429201229927698,
      "learning_rate": 4.0482501582724824e-07,
      "loss": 1.5768,
      "step": 2017
    },
    {
      "epoch": 0.304351104743232,
      "grad_norm": 0.2496444300054758,
      "learning_rate": 4.047379239734247e-07,
      "loss": 1.5181,
      "step": 2018
    },
    {
      "epoch": 0.3045019229319056,
      "grad_norm": 0.2467065446436571,
      "learning_rate": 4.046508029896588e-07,
      "loss": 1.6075,
      "step": 2019
    },
    {
      "epoch": 0.30465274112057916,
      "grad_norm": 0.24456129277039818,
      "learning_rate": 4.045636528955117e-07,
      "loss": 1.5415,
      "step": 2020
    },
    {
      "epoch": 0.3048035593092527,
      "grad_norm": 0.2641929208072259,
      "learning_rate": 4.0447647371055127e-07,
      "loss": 1.5642,
      "step": 2021
    },
    {
      "epoch": 0.30495437749792625,
      "grad_norm": 0.253542202164034,
      "learning_rate": 4.043892654543517e-07,
      "loss": 1.535,
      "step": 2022
    },
    {
      "epoch": 0.30510519568659983,
      "grad_norm": 0.24314338746301464,
      "learning_rate": 4.0430202814649386e-07,
      "loss": 1.6212,
      "step": 2023
    },
    {
      "epoch": 0.30525601387527335,
      "grad_norm": 0.25716390681239226,
      "learning_rate": 4.0421476180656495e-07,
      "loss": 1.5737,
      "step": 2024
    },
    {
      "epoch": 0.3054068320639469,
      "grad_norm": 0.2569826323216223,
      "learning_rate": 4.041274664541589e-07,
      "loss": 1.5949,
      "step": 2025
    },
    {
      "epoch": 0.30555765025262044,
      "grad_norm": 0.29526969958564175,
      "learning_rate": 4.0404014210887597e-07,
      "loss": 1.5407,
      "step": 2026
    },
    {
      "epoch": 0.305708468441294,
      "grad_norm": 0.2548618078832389,
      "learning_rate": 4.0395278879032315e-07,
      "loss": 1.5513,
      "step": 2027
    },
    {
      "epoch": 0.3058592866299676,
      "grad_norm": 0.25121971550161243,
      "learning_rate": 4.0386540651811363e-07,
      "loss": 1.5606,
      "step": 2028
    },
    {
      "epoch": 0.3060101048186411,
      "grad_norm": 0.248833321551354,
      "learning_rate": 4.037779953118674e-07,
      "loss": 1.5918,
      "step": 2029
    },
    {
      "epoch": 0.3061609230073147,
      "grad_norm": 0.2538335656892254,
      "learning_rate": 4.036905551912108e-07,
      "loss": 1.556,
      "step": 2030
    },
    {
      "epoch": 0.30631174119598825,
      "grad_norm": 0.26766284188651057,
      "learning_rate": 4.0360308617577677e-07,
      "loss": 1.5752,
      "step": 2031
    },
    {
      "epoch": 0.30646255938466177,
      "grad_norm": 0.2617566660075584,
      "learning_rate": 4.035155882852045e-07,
      "loss": 1.5394,
      "step": 2032
    },
    {
      "epoch": 0.30661337757333534,
      "grad_norm": 0.2499477755959539,
      "learning_rate": 4.034280615391398e-07,
      "loss": 1.5377,
      "step": 2033
    },
    {
      "epoch": 0.3067641957620089,
      "grad_norm": 0.26571280247917245,
      "learning_rate": 4.033405059572351e-07,
      "loss": 1.5705,
      "step": 2034
    },
    {
      "epoch": 0.30691501395068244,
      "grad_norm": 0.23425416184187495,
      "learning_rate": 4.0325292155914913e-07,
      "loss": 1.5085,
      "step": 2035
    },
    {
      "epoch": 0.307065832139356,
      "grad_norm": 0.24820658850688326,
      "learning_rate": 4.031653083645472e-07,
      "loss": 1.4914,
      "step": 2036
    },
    {
      "epoch": 0.3072166503280296,
      "grad_norm": 0.4945570448028612,
      "learning_rate": 4.0307766639310083e-07,
      "loss": 1.6127,
      "step": 2037
    },
    {
      "epoch": 0.3073674685167031,
      "grad_norm": 0.25618881711795755,
      "learning_rate": 4.0298999566448837e-07,
      "loss": 1.5884,
      "step": 2038
    },
    {
      "epoch": 0.3075182867053767,
      "grad_norm": 0.2516212800103507,
      "learning_rate": 4.029022961983944e-07,
      "loss": 1.5638,
      "step": 2039
    },
    {
      "epoch": 0.30766910489405025,
      "grad_norm": 0.26196457733833556,
      "learning_rate": 4.0281456801451004e-07,
      "loss": 1.5156,
      "step": 2040
    },
    {
      "epoch": 0.30781992308272377,
      "grad_norm": 0.2576699667694645,
      "learning_rate": 4.0272681113253274e-07,
      "loss": 1.5367,
      "step": 2041
    },
    {
      "epoch": 0.30797074127139734,
      "grad_norm": 0.2427529859890157,
      "learning_rate": 4.0263902557216654e-07,
      "loss": 1.5214,
      "step": 2042
    },
    {
      "epoch": 0.30812155946007086,
      "grad_norm": 0.27396337621628536,
      "learning_rate": 4.0255121135312186e-07,
      "loss": 1.6634,
      "step": 2043
    },
    {
      "epoch": 0.30827237764874443,
      "grad_norm": 0.24670362168152069,
      "learning_rate": 4.0246336849511554e-07,
      "loss": 1.5854,
      "step": 2044
    },
    {
      "epoch": 0.308423195837418,
      "grad_norm": 0.2509527913346677,
      "learning_rate": 4.023754970178708e-07,
      "loss": 1.5787,
      "step": 2045
    },
    {
      "epoch": 0.3085740140260915,
      "grad_norm": 0.27565341178868197,
      "learning_rate": 4.0228759694111745e-07,
      "loss": 1.574,
      "step": 2046
    },
    {
      "epoch": 0.3087248322147651,
      "grad_norm": 0.25797778584819087,
      "learning_rate": 4.0219966828459153e-07,
      "loss": 1.6037,
      "step": 2047
    },
    {
      "epoch": 0.3088756504034387,
      "grad_norm": 0.23845818730109467,
      "learning_rate": 4.0211171106803565e-07,
      "loss": 1.5155,
      "step": 2048
    },
    {
      "epoch": 0.3090264685921122,
      "grad_norm": 0.24152957942868847,
      "learning_rate": 4.0202372531119864e-07,
      "loss": 1.5317,
      "step": 2049
    },
    {
      "epoch": 0.30917728678078576,
      "grad_norm": 0.24716342526147034,
      "learning_rate": 4.0193571103383607e-07,
      "loss": 1.5569,
      "step": 2050
    },
    {
      "epoch": 0.30932810496945934,
      "grad_norm": 0.2939391755532749,
      "learning_rate": 4.0184766825570953e-07,
      "loss": 1.626,
      "step": 2051
    },
    {
      "epoch": 0.30947892315813286,
      "grad_norm": 0.23417825092694933,
      "learning_rate": 4.0175959699658724e-07,
      "loss": 1.5612,
      "step": 2052
    },
    {
      "epoch": 0.30962974134680643,
      "grad_norm": 0.2490779999885615,
      "learning_rate": 4.016714972762438e-07,
      "loss": 1.5613,
      "step": 2053
    },
    {
      "epoch": 0.30978055953548,
      "grad_norm": 0.2748134255196443,
      "learning_rate": 4.015833691144602e-07,
      "loss": 1.5536,
      "step": 2054
    },
    {
      "epoch": 0.3099313777241535,
      "grad_norm": 0.2582447556287364,
      "learning_rate": 4.0149521253102363e-07,
      "loss": 1.5974,
      "step": 2055
    },
    {
      "epoch": 0.3100821959128271,
      "grad_norm": 0.2458853398411171,
      "learning_rate": 4.01407027545728e-07,
      "loss": 1.5975,
      "step": 2056
    },
    {
      "epoch": 0.3102330141015006,
      "grad_norm": 0.243255176864826,
      "learning_rate": 4.013188141783732e-07,
      "loss": 1.598,
      "step": 2057
    },
    {
      "epoch": 0.3103838322901742,
      "grad_norm": 0.2612334752736981,
      "learning_rate": 4.0123057244876585e-07,
      "loss": 1.5072,
      "step": 2058
    },
    {
      "epoch": 0.31053465047884776,
      "grad_norm": 0.2533063131475423,
      "learning_rate": 4.0114230237671874e-07,
      "loss": 1.5773,
      "step": 2059
    },
    {
      "epoch": 0.3106854686675213,
      "grad_norm": 0.24648584229395598,
      "learning_rate": 4.010540039820511e-07,
      "loss": 1.5581,
      "step": 2060
    },
    {
      "epoch": 0.31083628685619485,
      "grad_norm": 0.24392715663996672,
      "learning_rate": 4.009656772845884e-07,
      "loss": 1.6086,
      "step": 2061
    },
    {
      "epoch": 0.31098710504486843,
      "grad_norm": 0.24558646280686036,
      "learning_rate": 4.008773223041627e-07,
      "loss": 1.6147,
      "step": 2062
    },
    {
      "epoch": 0.31113792323354195,
      "grad_norm": 0.5569634572330247,
      "learning_rate": 4.007889390606122e-07,
      "loss": 1.5862,
      "step": 2063
    },
    {
      "epoch": 0.3112887414222155,
      "grad_norm": 0.2558174149083045,
      "learning_rate": 4.007005275737814e-07,
      "loss": 1.6275,
      "step": 2064
    },
    {
      "epoch": 0.3114395596108891,
      "grad_norm": 0.2481834056158772,
      "learning_rate": 4.006120878635214e-07,
      "loss": 1.5096,
      "step": 2065
    },
    {
      "epoch": 0.3115903777995626,
      "grad_norm": 0.38793992659885385,
      "learning_rate": 4.0052361994968946e-07,
      "loss": 1.6465,
      "step": 2066
    },
    {
      "epoch": 0.3117411959882362,
      "grad_norm": 0.2628176803158427,
      "learning_rate": 4.0043512385214923e-07,
      "loss": 1.5849,
      "step": 2067
    },
    {
      "epoch": 0.31189201417690976,
      "grad_norm": 0.4098456469378603,
      "learning_rate": 4.003465995907706e-07,
      "loss": 1.549,
      "step": 2068
    },
    {
      "epoch": 0.3120428323655833,
      "grad_norm": 0.5901459740996087,
      "learning_rate": 4.0025804718542983e-07,
      "loss": 1.6553,
      "step": 2069
    },
    {
      "epoch": 0.31219365055425685,
      "grad_norm": 0.27798836334717175,
      "learning_rate": 4.001694666560096e-07,
      "loss": 1.608,
      "step": 2070
    },
    {
      "epoch": 0.31234446874293037,
      "grad_norm": 0.25224554357567286,
      "learning_rate": 4.0008085802239874e-07,
      "loss": 1.611,
      "step": 2071
    },
    {
      "epoch": 0.31249528693160394,
      "grad_norm": 0.25170188689624073,
      "learning_rate": 3.9999222130449244e-07,
      "loss": 1.6008,
      "step": 2072
    },
    {
      "epoch": 0.3126461051202775,
      "grad_norm": 0.31290105462564416,
      "learning_rate": 3.999035565221924e-07,
      "loss": 1.7323,
      "step": 2073
    },
    {
      "epoch": 0.31279692330895104,
      "grad_norm": 0.3964292109873429,
      "learning_rate": 3.9981486369540626e-07,
      "loss": 1.6061,
      "step": 2074
    },
    {
      "epoch": 0.3129477414976246,
      "grad_norm": 0.24857853557946208,
      "learning_rate": 3.9972614284404814e-07,
      "loss": 1.5927,
      "step": 2075
    },
    {
      "epoch": 0.3130985596862982,
      "grad_norm": 0.3851454200852006,
      "learning_rate": 3.9963739398803864e-07,
      "loss": 1.5901,
      "step": 2076
    },
    {
      "epoch": 0.3132493778749717,
      "grad_norm": 0.2628770261936668,
      "learning_rate": 3.9954861714730434e-07,
      "loss": 1.6159,
      "step": 2077
    },
    {
      "epoch": 0.3134001960636453,
      "grad_norm": 0.2677944478927853,
      "learning_rate": 3.994598123417781e-07,
      "loss": 1.5402,
      "step": 2078
    },
    {
      "epoch": 0.31355101425231885,
      "grad_norm": 0.27617794676009344,
      "learning_rate": 3.993709795913995e-07,
      "loss": 1.6657,
      "step": 2079
    },
    {
      "epoch": 0.31370183244099237,
      "grad_norm": 0.2456539890488146,
      "learning_rate": 3.992821189161138e-07,
      "loss": 1.6086,
      "step": 2080
    },
    {
      "epoch": 0.31385265062966594,
      "grad_norm": 0.24892899783486858,
      "learning_rate": 3.9919323033587295e-07,
      "loss": 1.5301,
      "step": 2081
    },
    {
      "epoch": 0.3140034688183395,
      "grad_norm": 0.2393655906478922,
      "learning_rate": 3.9910431387063493e-07,
      "loss": 1.5066,
      "step": 2082
    },
    {
      "epoch": 0.31415428700701303,
      "grad_norm": 0.2498382602800907,
      "learning_rate": 3.9901536954036416e-07,
      "loss": 1.5667,
      "step": 2083
    },
    {
      "epoch": 0.3143051051956866,
      "grad_norm": 0.2493389529791471,
      "learning_rate": 3.989263973650312e-07,
      "loss": 1.6019,
      "step": 2084
    },
    {
      "epoch": 0.3144559233843602,
      "grad_norm": 0.24762083529534487,
      "learning_rate": 3.988373973646129e-07,
      "loss": 1.5889,
      "step": 2085
    },
    {
      "epoch": 0.3146067415730337,
      "grad_norm": 0.6543181533090748,
      "learning_rate": 3.9874836955909236e-07,
      "loss": 1.5563,
      "step": 2086
    },
    {
      "epoch": 0.3147575597617073,
      "grad_norm": 0.2502822106752469,
      "learning_rate": 3.9865931396845887e-07,
      "loss": 1.5951,
      "step": 2087
    },
    {
      "epoch": 0.3149083779503808,
      "grad_norm": 0.27690010796350406,
      "learning_rate": 3.9857023061270805e-07,
      "loss": 1.5315,
      "step": 2088
    },
    {
      "epoch": 0.31505919613905436,
      "grad_norm": 0.24290586659146246,
      "learning_rate": 3.984811195118417e-07,
      "loss": 1.5492,
      "step": 2089
    },
    {
      "epoch": 0.31521001432772794,
      "grad_norm": 0.24725428291645074,
      "learning_rate": 3.983919806858678e-07,
      "loss": 1.5565,
      "step": 2090
    },
    {
      "epoch": 0.31536083251640146,
      "grad_norm": 0.25750934092248584,
      "learning_rate": 3.983028141548008e-07,
      "loss": 1.5927,
      "step": 2091
    },
    {
      "epoch": 0.31551165070507503,
      "grad_norm": 0.29618494596929745,
      "learning_rate": 3.982136199386609e-07,
      "loss": 1.4966,
      "step": 2092
    },
    {
      "epoch": 0.3156624688937486,
      "grad_norm": 0.2486932990700529,
      "learning_rate": 3.98124398057475e-07,
      "loss": 1.6357,
      "step": 2093
    },
    {
      "epoch": 0.3158132870824221,
      "grad_norm": 0.2533954281924328,
      "learning_rate": 3.9803514853127596e-07,
      "loss": 1.5909,
      "step": 2094
    },
    {
      "epoch": 0.3159641052710957,
      "grad_norm": 0.2589368625928245,
      "learning_rate": 3.9794587138010296e-07,
      "loss": 1.531,
      "step": 2095
    },
    {
      "epoch": 0.31611492345976927,
      "grad_norm": 0.2542643224228135,
      "learning_rate": 3.978565666240011e-07,
      "loss": 1.6488,
      "step": 2096
    },
    {
      "epoch": 0.3162657416484428,
      "grad_norm": 0.244359072570323,
      "learning_rate": 3.9776723428302216e-07,
      "loss": 1.5402,
      "step": 2097
    },
    {
      "epoch": 0.31641655983711636,
      "grad_norm": 0.25460464026757734,
      "learning_rate": 3.976778743772237e-07,
      "loss": 1.5894,
      "step": 2098
    },
    {
      "epoch": 0.31656737802578994,
      "grad_norm": 0.2868104823843204,
      "learning_rate": 3.975884869266698e-07,
      "loss": 1.5637,
      "step": 2099
    },
    {
      "epoch": 0.31671819621446345,
      "grad_norm": 0.31017877646583497,
      "learning_rate": 3.974990719514302e-07,
      "loss": 1.5839,
      "step": 2100
    },
    {
      "epoch": 0.31686901440313703,
      "grad_norm": 0.26396866795190366,
      "learning_rate": 3.974096294715816e-07,
      "loss": 1.5178,
      "step": 2101
    },
    {
      "epoch": 0.31701983259181055,
      "grad_norm": 0.30168437991242275,
      "learning_rate": 3.973201595072061e-07,
      "loss": 1.6485,
      "step": 2102
    },
    {
      "epoch": 0.3171706507804841,
      "grad_norm": 0.24033094292592433,
      "learning_rate": 3.972306620783925e-07,
      "loss": 1.5973,
      "step": 2103
    },
    {
      "epoch": 0.3173214689691577,
      "grad_norm": 0.24507913145198376,
      "learning_rate": 3.971411372052354e-07,
      "loss": 1.5712,
      "step": 2104
    },
    {
      "epoch": 0.3174722871578312,
      "grad_norm": 0.23871265245180445,
      "learning_rate": 3.9705158490783594e-07,
      "loss": 1.5227,
      "step": 2105
    },
    {
      "epoch": 0.3176231053465048,
      "grad_norm": 0.3965284184062262,
      "learning_rate": 3.969620052063011e-07,
      "loss": 1.6573,
      "step": 2106
    },
    {
      "epoch": 0.31777392353517836,
      "grad_norm": 0.25745473428479865,
      "learning_rate": 3.9687239812074417e-07,
      "loss": 1.6004,
      "step": 2107
    },
    {
      "epoch": 0.3179247417238519,
      "grad_norm": 0.2983580390523171,
      "learning_rate": 3.9678276367128457e-07,
      "loss": 1.633,
      "step": 2108
    },
    {
      "epoch": 0.31807555991252545,
      "grad_norm": 0.25214947570308643,
      "learning_rate": 3.9669310187804776e-07,
      "loss": 1.5447,
      "step": 2109
    },
    {
      "epoch": 0.318226378101199,
      "grad_norm": 0.7384214341008223,
      "learning_rate": 3.9660341276116554e-07,
      "loss": 1.4902,
      "step": 2110
    },
    {
      "epoch": 0.31837719628987254,
      "grad_norm": 0.25732836197018455,
      "learning_rate": 3.9651369634077566e-07,
      "loss": 1.6187,
      "step": 2111
    },
    {
      "epoch": 0.3185280144785461,
      "grad_norm": 0.24660030206232522,
      "learning_rate": 3.9642395263702213e-07,
      "loss": 1.5436,
      "step": 2112
    },
    {
      "epoch": 0.3186788326672197,
      "grad_norm": 0.24938974250593018,
      "learning_rate": 3.9633418167005486e-07,
      "loss": 1.5876,
      "step": 2113
    },
    {
      "epoch": 0.3188296508558932,
      "grad_norm": 0.32231112612116103,
      "learning_rate": 3.9624438346003023e-07,
      "loss": 1.5242,
      "step": 2114
    },
    {
      "epoch": 0.3189804690445668,
      "grad_norm": 0.25084366668647723,
      "learning_rate": 3.961545580271106e-07,
      "loss": 1.5895,
      "step": 2115
    },
    {
      "epoch": 0.3191312872332403,
      "grad_norm": 0.23999056686624912,
      "learning_rate": 3.960647053914641e-07,
      "loss": 1.5713,
      "step": 2116
    },
    {
      "epoch": 0.3192821054219139,
      "grad_norm": 0.23925194136729683,
      "learning_rate": 3.959748255732655e-07,
      "loss": 1.5073,
      "step": 2117
    },
    {
      "epoch": 0.31943292361058745,
      "grad_norm": 0.27256469617442297,
      "learning_rate": 3.9588491859269534e-07,
      "loss": 1.5992,
      "step": 2118
    },
    {
      "epoch": 0.31958374179926097,
      "grad_norm": 0.23824644344923226,
      "learning_rate": 3.9579498446994055e-07,
      "loss": 1.5236,
      "step": 2119
    },
    {
      "epoch": 0.31973455998793454,
      "grad_norm": 0.2652599022290534,
      "learning_rate": 3.9570502322519365e-07,
      "loss": 1.5538,
      "step": 2120
    },
    {
      "epoch": 0.3198853781766081,
      "grad_norm": 0.24330049060046072,
      "learning_rate": 3.956150348786538e-07,
      "loss": 1.6044,
      "step": 2121
    },
    {
      "epoch": 0.32003619636528163,
      "grad_norm": 0.2738743828606264,
      "learning_rate": 3.955250194505259e-07,
      "loss": 1.6004,
      "step": 2122
    },
    {
      "epoch": 0.3201870145539552,
      "grad_norm": 0.2574178597933717,
      "learning_rate": 3.9543497696102107e-07,
      "loss": 1.5936,
      "step": 2123
    },
    {
      "epoch": 0.3203378327426288,
      "grad_norm": 0.2950917159578343,
      "learning_rate": 3.9534490743035643e-07,
      "loss": 1.5641,
      "step": 2124
    },
    {
      "epoch": 0.3204886509313023,
      "grad_norm": 0.4756485614317174,
      "learning_rate": 3.952548108787552e-07,
      "loss": 1.5103,
      "step": 2125
    },
    {
      "epoch": 0.3206394691199759,
      "grad_norm": 0.2615377016755134,
      "learning_rate": 3.951646873264468e-07,
      "loss": 1.6246,
      "step": 2126
    },
    {
      "epoch": 0.32079028730864945,
      "grad_norm": 0.34577547477197323,
      "learning_rate": 3.950745367936665e-07,
      "loss": 1.5757,
      "step": 2127
    },
    {
      "epoch": 0.32094110549732296,
      "grad_norm": 0.254203670031749,
      "learning_rate": 3.9498435930065566e-07,
      "loss": 1.5617,
      "step": 2128
    },
    {
      "epoch": 0.32109192368599654,
      "grad_norm": 0.24546737378883726,
      "learning_rate": 3.94894154867662e-07,
      "loss": 1.5633,
      "step": 2129
    },
    {
      "epoch": 0.3212427418746701,
      "grad_norm": 0.2515906858162734,
      "learning_rate": 3.948039235149387e-07,
      "loss": 1.5815,
      "step": 2130
    },
    {
      "epoch": 0.32139356006334363,
      "grad_norm": 0.24451920390423948,
      "learning_rate": 3.947136652627455e-07,
      "loss": 1.6121,
      "step": 2131
    },
    {
      "epoch": 0.3215443782520172,
      "grad_norm": 0.27481738304353953,
      "learning_rate": 3.9462338013134813e-07,
      "loss": 1.5426,
      "step": 2132
    },
    {
      "epoch": 0.3216951964406907,
      "grad_norm": 0.2965333454714891,
      "learning_rate": 3.9453306814101795e-07,
      "loss": 1.6291,
      "step": 2133
    },
    {
      "epoch": 0.3218460146293643,
      "grad_norm": 0.24792020164578948,
      "learning_rate": 3.9444272931203294e-07,
      "loss": 1.499,
      "step": 2134
    },
    {
      "epoch": 0.32199683281803787,
      "grad_norm": 0.24446936022136556,
      "learning_rate": 3.943523636646765e-07,
      "loss": 1.6617,
      "step": 2135
    },
    {
      "epoch": 0.3221476510067114,
      "grad_norm": 0.24328300281267154,
      "learning_rate": 3.9426197121923853e-07,
      "loss": 1.5624,
      "step": 2136
    },
    {
      "epoch": 0.32229846919538496,
      "grad_norm": 0.26383286243955617,
      "learning_rate": 3.941715519960148e-07,
      "loss": 1.6099,
      "step": 2137
    },
    {
      "epoch": 0.32244928738405854,
      "grad_norm": 0.26621979937034657,
      "learning_rate": 3.94081106015307e-07,
      "loss": 1.5884,
      "step": 2138
    },
    {
      "epoch": 0.32260010557273205,
      "grad_norm": 0.24855711191351867,
      "learning_rate": 3.9399063329742277e-07,
      "loss": 1.4988,
      "step": 2139
    },
    {
      "epoch": 0.32275092376140563,
      "grad_norm": 0.26128115604409363,
      "learning_rate": 3.93900133862676e-07,
      "loss": 1.6908,
      "step": 2140
    },
    {
      "epoch": 0.3229017419500792,
      "grad_norm": 0.8190000147879138,
      "learning_rate": 3.9380960773138643e-07,
      "loss": 1.5914,
      "step": 2141
    },
    {
      "epoch": 0.3230525601387527,
      "grad_norm": 0.24931712688641844,
      "learning_rate": 3.9371905492387987e-07,
      "loss": 1.5765,
      "step": 2142
    },
    {
      "epoch": 0.3232033783274263,
      "grad_norm": 0.2604485651872273,
      "learning_rate": 3.9362847546048796e-07,
      "loss": 1.5407,
      "step": 2143
    },
    {
      "epoch": 0.32335419651609987,
      "grad_norm": 0.3841819530609337,
      "learning_rate": 3.935378693615484e-07,
      "loss": 1.6656,
      "step": 2144
    },
    {
      "epoch": 0.3235050147047734,
      "grad_norm": 0.25042361022528925,
      "learning_rate": 3.93447236647405e-07,
      "loss": 1.6574,
      "step": 2145
    },
    {
      "epoch": 0.32365583289344696,
      "grad_norm": 0.2479456279202682,
      "learning_rate": 3.9335657733840744e-07,
      "loss": 1.5793,
      "step": 2146
    },
    {
      "epoch": 0.3238066510821205,
      "grad_norm": 0.26260084267739775,
      "learning_rate": 3.932658914549113e-07,
      "loss": 1.6222,
      "step": 2147
    },
    {
      "epoch": 0.32395746927079405,
      "grad_norm": 0.24831072353040998,
      "learning_rate": 3.931751790172783e-07,
      "loss": 1.5327,
      "step": 2148
    },
    {
      "epoch": 0.3241082874594676,
      "grad_norm": 0.2357508409240409,
      "learning_rate": 3.930844400458759e-07,
      "loss": 1.548,
      "step": 2149
    },
    {
      "epoch": 0.32425910564814114,
      "grad_norm": 0.3313130342092389,
      "learning_rate": 3.9299367456107787e-07,
      "loss": 1.5344,
      "step": 2150
    },
    {
      "epoch": 0.3244099238368147,
      "grad_norm": 0.2655658423317557,
      "learning_rate": 3.929028825832634e-07,
      "loss": 1.5548,
      "step": 2151
    },
    {
      "epoch": 0.3245607420254883,
      "grad_norm": 0.27936045500581236,
      "learning_rate": 3.928120641328181e-07,
      "loss": 1.5474,
      "step": 2152
    },
    {
      "epoch": 0.3247115602141618,
      "grad_norm": 0.2562919828901846,
      "learning_rate": 3.9272121923013336e-07,
      "loss": 1.6042,
      "step": 2153
    },
    {
      "epoch": 0.3248623784028354,
      "grad_norm": 0.24525026196111174,
      "learning_rate": 3.9263034789560645e-07,
      "loss": 1.537,
      "step": 2154
    },
    {
      "epoch": 0.32501319659150896,
      "grad_norm": 0.2601161416687034,
      "learning_rate": 3.9253945014964066e-07,
      "loss": 1.6371,
      "step": 2155
    },
    {
      "epoch": 0.3251640147801825,
      "grad_norm": 0.2594960670160656,
      "learning_rate": 3.9244852601264525e-07,
      "loss": 1.5335,
      "step": 2156
    },
    {
      "epoch": 0.32531483296885605,
      "grad_norm": 0.2625263636856618,
      "learning_rate": 3.923575755050352e-07,
      "loss": 1.5754,
      "step": 2157
    },
    {
      "epoch": 0.3254656511575296,
      "grad_norm": 0.2463256200074851,
      "learning_rate": 3.922665986472316e-07,
      "loss": 1.5661,
      "step": 2158
    },
    {
      "epoch": 0.32561646934620314,
      "grad_norm": 0.2577915088506624,
      "learning_rate": 3.921755954596615e-07,
      "loss": 1.5505,
      "step": 2159
    },
    {
      "epoch": 0.3257672875348767,
      "grad_norm": 0.2509018610982779,
      "learning_rate": 3.920845659627577e-07,
      "loss": 1.6161,
      "step": 2160
    },
    {
      "epoch": 0.32591810572355023,
      "grad_norm": 0.25334757643146116,
      "learning_rate": 3.9199351017695887e-07,
      "loss": 1.551,
      "step": 2161
    },
    {
      "epoch": 0.3260689239122238,
      "grad_norm": 0.27512796587292415,
      "learning_rate": 3.9190242812270985e-07,
      "loss": 1.6038,
      "step": 2162
    },
    {
      "epoch": 0.3262197421008974,
      "grad_norm": 0.28985312260866014,
      "learning_rate": 3.918113198204611e-07,
      "loss": 1.5417,
      "step": 2163
    },
    {
      "epoch": 0.3263705602895709,
      "grad_norm": 0.2503779552056232,
      "learning_rate": 3.917201852906691e-07,
      "loss": 1.5842,
      "step": 2164
    },
    {
      "epoch": 0.3265213784782445,
      "grad_norm": 0.24484423110593662,
      "learning_rate": 3.9162902455379636e-07,
      "loss": 1.5275,
      "step": 2165
    },
    {
      "epoch": 0.32667219666691805,
      "grad_norm": 0.2890156902730706,
      "learning_rate": 3.9153783763031085e-07,
      "loss": 1.5492,
      "step": 2166
    },
    {
      "epoch": 0.32682301485559156,
      "grad_norm": 0.24639661617104186,
      "learning_rate": 3.914466245406869e-07,
      "loss": 1.5571,
      "step": 2167
    },
    {
      "epoch": 0.32697383304426514,
      "grad_norm": 0.2450312290995254,
      "learning_rate": 3.9135538530540447e-07,
      "loss": 1.623,
      "step": 2168
    },
    {
      "epoch": 0.3271246512329387,
      "grad_norm": 0.25377142288752186,
      "learning_rate": 3.9126411994494936e-07,
      "loss": 1.6188,
      "step": 2169
    },
    {
      "epoch": 0.32727546942161223,
      "grad_norm": 0.2547601828093652,
      "learning_rate": 3.911728284798133e-07,
      "loss": 1.5359,
      "step": 2170
    },
    {
      "epoch": 0.3274262876102858,
      "grad_norm": 0.25001982542415085,
      "learning_rate": 3.9108151093049397e-07,
      "loss": 1.5617,
      "step": 2171
    },
    {
      "epoch": 0.3275771057989594,
      "grad_norm": 0.25181077387583223,
      "learning_rate": 3.909901673174948e-07,
      "loss": 1.5743,
      "step": 2172
    },
    {
      "epoch": 0.3277279239876329,
      "grad_norm": 0.24853020045973234,
      "learning_rate": 3.90898797661325e-07,
      "loss": 1.5056,
      "step": 2173
    },
    {
      "epoch": 0.32787874217630647,
      "grad_norm": 0.24903623322804977,
      "learning_rate": 3.9080740198249976e-07,
      "loss": 1.5015,
      "step": 2174
    },
    {
      "epoch": 0.32802956036498004,
      "grad_norm": 0.2715084471234611,
      "learning_rate": 3.9071598030154015e-07,
      "loss": 1.568,
      "step": 2175
    },
    {
      "epoch": 0.32818037855365356,
      "grad_norm": 0.24719338424330595,
      "learning_rate": 3.906245326389729e-07,
      "loss": 1.4931,
      "step": 2176
    },
    {
      "epoch": 0.32833119674232714,
      "grad_norm": 0.24057406371883366,
      "learning_rate": 3.905330590153308e-07,
      "loss": 1.6049,
      "step": 2177
    },
    {
      "epoch": 0.32848201493100065,
      "grad_norm": 0.2577269643414046,
      "learning_rate": 3.904415594511521e-07,
      "loss": 1.5177,
      "step": 2178
    },
    {
      "epoch": 0.32863283311967423,
      "grad_norm": 0.2277806229734619,
      "learning_rate": 3.903500339669814e-07,
      "loss": 1.5321,
      "step": 2179
    },
    {
      "epoch": 0.3287836513083478,
      "grad_norm": 0.25168229116355645,
      "learning_rate": 3.902584825833687e-07,
      "loss": 1.573,
      "step": 2180
    },
    {
      "epoch": 0.3289344694970213,
      "grad_norm": 0.25178755697604427,
      "learning_rate": 3.9016690532086995e-07,
      "loss": 1.475,
      "step": 2181
    },
    {
      "epoch": 0.3290852876856949,
      "grad_norm": 0.23538365702618483,
      "learning_rate": 3.9007530220004686e-07,
      "loss": 1.5286,
      "step": 2182
    },
    {
      "epoch": 0.32923610587436847,
      "grad_norm": 0.2477937784114404,
      "learning_rate": 3.8998367324146716e-07,
      "loss": 1.6322,
      "step": 2183
    },
    {
      "epoch": 0.329386924063042,
      "grad_norm": 0.2545925937803815,
      "learning_rate": 3.8989201846570403e-07,
      "loss": 1.6231,
      "step": 2184
    },
    {
      "epoch": 0.32953774225171556,
      "grad_norm": 0.26975812222452605,
      "learning_rate": 3.8980033789333676e-07,
      "loss": 1.5295,
      "step": 2185
    },
    {
      "epoch": 0.32968856044038913,
      "grad_norm": 0.24686203970847262,
      "learning_rate": 3.897086315449503e-07,
      "loss": 1.4873,
      "step": 2186
    },
    {
      "epoch": 0.32983937862906265,
      "grad_norm": 0.24428165251845188,
      "learning_rate": 3.896168994411354e-07,
      "loss": 1.5535,
      "step": 2187
    },
    {
      "epoch": 0.3299901968177362,
      "grad_norm": 0.25997229380516884,
      "learning_rate": 3.895251416024885e-07,
      "loss": 1.5302,
      "step": 2188
    },
    {
      "epoch": 0.3301410150064098,
      "grad_norm": 0.24197139910273505,
      "learning_rate": 3.89433358049612e-07,
      "loss": 1.5913,
      "step": 2189
    },
    {
      "epoch": 0.3302918331950833,
      "grad_norm": 0.3740910107363862,
      "learning_rate": 3.89341548803114e-07,
      "loss": 1.5811,
      "step": 2190
    },
    {
      "epoch": 0.3304426513837569,
      "grad_norm": 0.27649441189430246,
      "learning_rate": 3.892497138836083e-07,
      "loss": 1.5538,
      "step": 2191
    },
    {
      "epoch": 0.3305934695724304,
      "grad_norm": 0.2725749848661994,
      "learning_rate": 3.8915785331171434e-07,
      "loss": 1.6107,
      "step": 2192
    },
    {
      "epoch": 0.330744287761104,
      "grad_norm": 0.25734228685195976,
      "learning_rate": 3.8906596710805786e-07,
      "loss": 1.5388,
      "step": 2193
    },
    {
      "epoch": 0.33089510594977756,
      "grad_norm": 0.2754469554795411,
      "learning_rate": 3.889740552932697e-07,
      "loss": 1.5455,
      "step": 2194
    },
    {
      "epoch": 0.3310459241384511,
      "grad_norm": 0.40458062480498685,
      "learning_rate": 3.8888211788798686e-07,
      "loss": 1.6405,
      "step": 2195
    },
    {
      "epoch": 0.33119674232712465,
      "grad_norm": 0.26011942351190315,
      "learning_rate": 3.88790154912852e-07,
      "loss": 1.5743,
      "step": 2196
    },
    {
      "epoch": 0.3313475605157982,
      "grad_norm": 0.3383973302983143,
      "learning_rate": 3.886981663885133e-07,
      "loss": 1.5769,
      "step": 2197
    },
    {
      "epoch": 0.33149837870447174,
      "grad_norm": 0.2796548395836774,
      "learning_rate": 3.8860615233562505e-07,
      "loss": 1.5225,
      "step": 2198
    },
    {
      "epoch": 0.3316491968931453,
      "grad_norm": 0.25733138375962183,
      "learning_rate": 3.8851411277484707e-07,
      "loss": 1.53,
      "step": 2199
    },
    {
      "epoch": 0.3318000150818189,
      "grad_norm": 0.25688616343477166,
      "learning_rate": 3.884220477268448e-07,
      "loss": 1.5664,
      "step": 2200
    },
    {
      "epoch": 0.3319508332704924,
      "grad_norm": 0.2503649611643573,
      "learning_rate": 3.883299572122897e-07,
      "loss": 1.5365,
      "step": 2201
    },
    {
      "epoch": 0.332101651459166,
      "grad_norm": 0.24801357715162026,
      "learning_rate": 3.882378412518586e-07,
      "loss": 1.6041,
      "step": 2202
    },
    {
      "epoch": 0.33225246964783955,
      "grad_norm": 0.25363808699274276,
      "learning_rate": 3.881456998662343e-07,
      "loss": 1.5687,
      "step": 2203
    },
    {
      "epoch": 0.3324032878365131,
      "grad_norm": 0.2784222811332939,
      "learning_rate": 3.8805353307610516e-07,
      "loss": 1.5556,
      "step": 2204
    },
    {
      "epoch": 0.33255410602518665,
      "grad_norm": 0.2468012831812265,
      "learning_rate": 3.879613409021654e-07,
      "loss": 1.5134,
      "step": 2205
    },
    {
      "epoch": 0.33270492421386016,
      "grad_norm": 0.26575926203468225,
      "learning_rate": 3.8786912336511486e-07,
      "loss": 1.6228,
      "step": 2206
    },
    {
      "epoch": 0.33285574240253374,
      "grad_norm": 0.2587565684416558,
      "learning_rate": 3.87776880485659e-07,
      "loss": 1.4733,
      "step": 2207
    },
    {
      "epoch": 0.3330065605912073,
      "grad_norm": 0.2437834927101869,
      "learning_rate": 3.87684612284509e-07,
      "loss": 1.5645,
      "step": 2208
    },
    {
      "epoch": 0.33315737877988083,
      "grad_norm": 0.26723277558649283,
      "learning_rate": 3.875923187823818e-07,
      "loss": 1.5352,
      "step": 2209
    },
    {
      "epoch": 0.3333081969685544,
      "grad_norm": 0.2504192784843705,
      "learning_rate": 3.875e-07,
      "loss": 1.5551,
      "step": 2210
    },
    {
      "epoch": 0.333459015157228,
      "grad_norm": 0.24068172574496213,
      "learning_rate": 3.8740765595809185e-07,
      "loss": 1.5108,
      "step": 2211
    },
    {
      "epoch": 0.3336098333459015,
      "grad_norm": 0.2829678841255783,
      "learning_rate": 3.8731528667739126e-07,
      "loss": 1.652,
      "step": 2212
    },
    {
      "epoch": 0.33376065153457507,
      "grad_norm": 0.24627534576714177,
      "learning_rate": 3.872228921786378e-07,
      "loss": 1.5755,
      "step": 2213
    },
    {
      "epoch": 0.33391146972324864,
      "grad_norm": 0.24708427462899366,
      "learning_rate": 3.8713047248257676e-07,
      "loss": 1.5418,
      "step": 2214
    },
    {
      "epoch": 0.33406228791192216,
      "grad_norm": 0.2679802446868686,
      "learning_rate": 3.870380276099591e-07,
      "loss": 1.5228,
      "step": 2215
    },
    {
      "epoch": 0.33421310610059574,
      "grad_norm": 0.2696195676939389,
      "learning_rate": 3.869455575815413e-07,
      "loss": 1.5744,
      "step": 2216
    },
    {
      "epoch": 0.3343639242892693,
      "grad_norm": 0.24986388598147388,
      "learning_rate": 3.868530624180856e-07,
      "loss": 1.5323,
      "step": 2217
    },
    {
      "epoch": 0.33451474247794283,
      "grad_norm": 0.27737913998484093,
      "learning_rate": 3.867605421403599e-07,
      "loss": 1.5496,
      "step": 2218
    },
    {
      "epoch": 0.3346655606666164,
      "grad_norm": 0.3270339884312583,
      "learning_rate": 3.8666799676913756e-07,
      "loss": 1.5286,
      "step": 2219
    },
    {
      "epoch": 0.33481637885529,
      "grad_norm": 0.2479114748899475,
      "learning_rate": 3.8657542632519796e-07,
      "loss": 1.5699,
      "step": 2220
    },
    {
      "epoch": 0.3349671970439635,
      "grad_norm": 0.23357546957906214,
      "learning_rate": 3.8648283082932567e-07,
      "loss": 1.5406,
      "step": 2221
    },
    {
      "epoch": 0.33511801523263707,
      "grad_norm": 0.44813768609402127,
      "learning_rate": 3.8639021030231106e-07,
      "loss": 1.6016,
      "step": 2222
    },
    {
      "epoch": 0.3352688334213106,
      "grad_norm": 0.24020122561270563,
      "learning_rate": 3.8629756476495023e-07,
      "loss": 1.5096,
      "step": 2223
    },
    {
      "epoch": 0.33541965160998416,
      "grad_norm": 0.24188367358672805,
      "learning_rate": 3.8620489423804476e-07,
      "loss": 1.5451,
      "step": 2224
    },
    {
      "epoch": 0.33557046979865773,
      "grad_norm": 0.24781994031545823,
      "learning_rate": 3.861121987424019e-07,
      "loss": 1.5975,
      "step": 2225
    },
    {
      "epoch": 0.33572128798733125,
      "grad_norm": 0.24221766652606178,
      "learning_rate": 3.860194782988345e-07,
      "loss": 1.5831,
      "step": 2226
    },
    {
      "epoch": 0.3358721061760048,
      "grad_norm": 0.24125890233354633,
      "learning_rate": 3.8592673292816094e-07,
      "loss": 1.5697,
      "step": 2227
    },
    {
      "epoch": 0.3360229243646784,
      "grad_norm": 0.2468993827799427,
      "learning_rate": 3.858339626512053e-07,
      "loss": 1.6049,
      "step": 2228
    },
    {
      "epoch": 0.3361737425533519,
      "grad_norm": 0.25379662786044455,
      "learning_rate": 3.857411674887972e-07,
      "loss": 1.5157,
      "step": 2229
    },
    {
      "epoch": 0.3363245607420255,
      "grad_norm": 1.124038381470129,
      "learning_rate": 3.856483474617718e-07,
      "loss": 1.5529,
      "step": 2230
    },
    {
      "epoch": 0.33647537893069907,
      "grad_norm": 0.26299087033602114,
      "learning_rate": 3.855555025909701e-07,
      "loss": 1.5535,
      "step": 2231
    },
    {
      "epoch": 0.3366261971193726,
      "grad_norm": 0.2474710227528756,
      "learning_rate": 3.8546263289723826e-07,
      "loss": 1.5763,
      "step": 2232
    },
    {
      "epoch": 0.33677701530804616,
      "grad_norm": 0.325380394521446,
      "learning_rate": 3.8536973840142836e-07,
      "loss": 1.544,
      "step": 2233
    },
    {
      "epoch": 0.33692783349671973,
      "grad_norm": 0.2543536772971605,
      "learning_rate": 3.852768191243977e-07,
      "loss": 1.542,
      "step": 2234
    },
    {
      "epoch": 0.33707865168539325,
      "grad_norm": 0.2777082163185314,
      "learning_rate": 3.8518387508700965e-07,
      "loss": 1.5565,
      "step": 2235
    },
    {
      "epoch": 0.3372294698740668,
      "grad_norm": 0.2348126950607934,
      "learning_rate": 3.8509090631013277e-07,
      "loss": 1.6255,
      "step": 2236
    },
    {
      "epoch": 0.33738028806274034,
      "grad_norm": 0.2588743938062479,
      "learning_rate": 3.849979128146412e-07,
      "loss": 1.5567,
      "step": 2237
    },
    {
      "epoch": 0.3375311062514139,
      "grad_norm": 0.24878932088777453,
      "learning_rate": 3.8490489462141474e-07,
      "loss": 1.6238,
      "step": 2238
    },
    {
      "epoch": 0.3376819244400875,
      "grad_norm": 0.24395580668015335,
      "learning_rate": 3.8481185175133856e-07,
      "loss": 1.5998,
      "step": 2239
    },
    {
      "epoch": 0.337832742628761,
      "grad_norm": 0.23968055186264078,
      "learning_rate": 3.847187842253036e-07,
      "loss": 1.6052,
      "step": 2240
    },
    {
      "epoch": 0.3379835608174346,
      "grad_norm": 0.2571877019236364,
      "learning_rate": 3.846256920642063e-07,
      "loss": 1.548,
      "step": 2241
    },
    {
      "epoch": 0.33813437900610815,
      "grad_norm": 0.2605293877175822,
      "learning_rate": 3.8453257528894844e-07,
      "loss": 1.639,
      "step": 2242
    },
    {
      "epoch": 0.3382851971947817,
      "grad_norm": 0.2505420542873784,
      "learning_rate": 3.8443943392043743e-07,
      "loss": 1.553,
      "step": 2243
    },
    {
      "epoch": 0.33843601538345525,
      "grad_norm": 0.27697603176255636,
      "learning_rate": 3.843462679795863e-07,
      "loss": 1.6327,
      "step": 2244
    },
    {
      "epoch": 0.3385868335721288,
      "grad_norm": 0.248589658769722,
      "learning_rate": 3.842530774873136e-07,
      "loss": 1.4545,
      "step": 2245
    },
    {
      "epoch": 0.33873765176080234,
      "grad_norm": 0.2972959581016365,
      "learning_rate": 3.841598624645431e-07,
      "loss": 1.6338,
      "step": 2246
    },
    {
      "epoch": 0.3388884699494759,
      "grad_norm": 0.25531625243830247,
      "learning_rate": 3.840666229322044e-07,
      "loss": 1.52,
      "step": 2247
    },
    {
      "epoch": 0.3390392881381495,
      "grad_norm": 0.3986180218938449,
      "learning_rate": 3.839733589112325e-07,
      "loss": 1.5638,
      "step": 2248
    },
    {
      "epoch": 0.339190106326823,
      "grad_norm": 0.2416717535849375,
      "learning_rate": 3.838800704225678e-07,
      "loss": 1.5555,
      "step": 2249
    },
    {
      "epoch": 0.3393409245154966,
      "grad_norm": 0.2445233088574405,
      "learning_rate": 3.837867574871565e-07,
      "loss": 1.5463,
      "step": 2250
    },
    {
      "epoch": 0.3394917427041701,
      "grad_norm": 0.25238831100568343,
      "learning_rate": 3.836934201259498e-07,
      "loss": 1.5774,
      "step": 2251
    },
    {
      "epoch": 0.33964256089284367,
      "grad_norm": 0.2525198477527483,
      "learning_rate": 3.836000583599048e-07,
      "loss": 1.5103,
      "step": 2252
    },
    {
      "epoch": 0.33979337908151724,
      "grad_norm": 0.2567497176908914,
      "learning_rate": 3.835066722099839e-07,
      "loss": 1.5488,
      "step": 2253
    },
    {
      "epoch": 0.33994419727019076,
      "grad_norm": 0.25762041730322705,
      "learning_rate": 3.834132616971551e-07,
      "loss": 1.5655,
      "step": 2254
    },
    {
      "epoch": 0.34009501545886434,
      "grad_norm": 0.28672621826467015,
      "learning_rate": 3.8331982684239165e-07,
      "loss": 1.5366,
      "step": 2255
    },
    {
      "epoch": 0.3402458336475379,
      "grad_norm": 0.2425603370892537,
      "learning_rate": 3.832263676666725e-07,
      "loss": 1.4937,
      "step": 2256
    },
    {
      "epoch": 0.34039665183621143,
      "grad_norm": 0.2441307058970796,
      "learning_rate": 3.8313288419098184e-07,
      "loss": 1.571,
      "step": 2257
    },
    {
      "epoch": 0.340547470024885,
      "grad_norm": 0.25371961723176456,
      "learning_rate": 3.8303937643630956e-07,
      "loss": 1.5487,
      "step": 2258
    },
    {
      "epoch": 0.3406982882135586,
      "grad_norm": 0.24999275713882987,
      "learning_rate": 3.8294584442365076e-07,
      "loss": 1.5377,
      "step": 2259
    },
    {
      "epoch": 0.3408491064022321,
      "grad_norm": 0.2901363335818867,
      "learning_rate": 3.8285228817400614e-07,
      "loss": 1.5308,
      "step": 2260
    },
    {
      "epoch": 0.34099992459090567,
      "grad_norm": 0.246290205087347,
      "learning_rate": 3.8275870770838185e-07,
      "loss": 1.5847,
      "step": 2261
    },
    {
      "epoch": 0.34115074277957924,
      "grad_norm": 0.2493151932192866,
      "learning_rate": 3.8266510304778944e-07,
      "loss": 1.6125,
      "step": 2262
    },
    {
      "epoch": 0.34130156096825276,
      "grad_norm": 0.2547615126894274,
      "learning_rate": 3.825714742132458e-07,
      "loss": 1.593,
      "step": 2263
    },
    {
      "epoch": 0.34145237915692633,
      "grad_norm": 0.25709531771387,
      "learning_rate": 3.824778212257734e-07,
      "loss": 1.5697,
      "step": 2264
    },
    {
      "epoch": 0.34160319734559985,
      "grad_norm": 0.2512506159715361,
      "learning_rate": 3.823841441064e-07,
      "loss": 1.6255,
      "step": 2265
    },
    {
      "epoch": 0.3417540155342734,
      "grad_norm": 0.2507657433009384,
      "learning_rate": 3.8229044287615895e-07,
      "loss": 1.5278,
      "step": 2266
    },
    {
      "epoch": 0.341904833722947,
      "grad_norm": 0.27216665659354644,
      "learning_rate": 3.8219671755608875e-07,
      "loss": 1.5638,
      "step": 2267
    },
    {
      "epoch": 0.3420556519116205,
      "grad_norm": 0.26557635715854694,
      "learning_rate": 3.8210296816723356e-07,
      "loss": 1.5815,
      "step": 2268
    },
    {
      "epoch": 0.3422064701002941,
      "grad_norm": 0.24469797082153816,
      "learning_rate": 3.820091947306429e-07,
      "loss": 1.4977,
      "step": 2269
    },
    {
      "epoch": 0.34235728828896766,
      "grad_norm": 0.2497279085693623,
      "learning_rate": 3.819153972673715e-07,
      "loss": 1.5477,
      "step": 2270
    },
    {
      "epoch": 0.3425081064776412,
      "grad_norm": 0.2527735661968631,
      "learning_rate": 3.8182157579847965e-07,
      "loss": 1.5595,
      "step": 2271
    },
    {
      "epoch": 0.34265892466631476,
      "grad_norm": 0.2540785759331565,
      "learning_rate": 3.8172773034503314e-07,
      "loss": 1.5908,
      "step": 2272
    },
    {
      "epoch": 0.34280974285498833,
      "grad_norm": 0.28467700544081004,
      "learning_rate": 3.816338609281028e-07,
      "loss": 1.5355,
      "step": 2273
    },
    {
      "epoch": 0.34296056104366185,
      "grad_norm": 0.2462037570531213,
      "learning_rate": 3.815399675687653e-07,
      "loss": 1.5683,
      "step": 2274
    },
    {
      "epoch": 0.3431113792323354,
      "grad_norm": 0.25120363303396226,
      "learning_rate": 3.814460502881023e-07,
      "loss": 1.5563,
      "step": 2275
    },
    {
      "epoch": 0.343262197421009,
      "grad_norm": 0.24889628742764763,
      "learning_rate": 3.8135210910720084e-07,
      "loss": 1.5814,
      "step": 2276
    },
    {
      "epoch": 0.3434130156096825,
      "grad_norm": 0.2502347607258946,
      "learning_rate": 3.8125814404715364e-07,
      "loss": 1.5306,
      "step": 2277
    },
    {
      "epoch": 0.3435638337983561,
      "grad_norm": 1.4362376842967715,
      "learning_rate": 3.8116415512905843e-07,
      "loss": 1.5799,
      "step": 2278
    },
    {
      "epoch": 0.34371465198702966,
      "grad_norm": 0.24982516403294563,
      "learning_rate": 3.8107014237401867e-07,
      "loss": 1.5381,
      "step": 2279
    },
    {
      "epoch": 0.3438654701757032,
      "grad_norm": 0.3185115617545277,
      "learning_rate": 3.8097610580314284e-07,
      "loss": 1.526,
      "step": 2280
    },
    {
      "epoch": 0.34401628836437675,
      "grad_norm": 0.33250873403709974,
      "learning_rate": 3.8088204543754483e-07,
      "loss": 1.6106,
      "step": 2281
    },
    {
      "epoch": 0.3441671065530503,
      "grad_norm": 0.24036323630812328,
      "learning_rate": 3.80787961298344e-07,
      "loss": 1.5334,
      "step": 2282
    },
    {
      "epoch": 0.34431792474172385,
      "grad_norm": 0.24417765915610368,
      "learning_rate": 3.8069385340666497e-07,
      "loss": 1.5566,
      "step": 2283
    },
    {
      "epoch": 0.3444687429303974,
      "grad_norm": 0.25273304777011013,
      "learning_rate": 3.8059972178363777e-07,
      "loss": 1.578,
      "step": 2284
    },
    {
      "epoch": 0.34461956111907094,
      "grad_norm": 0.24623320287757744,
      "learning_rate": 3.805055664503975e-07,
      "loss": 1.5188,
      "step": 2285
    },
    {
      "epoch": 0.3447703793077445,
      "grad_norm": 0.23877957030498415,
      "learning_rate": 3.8041138742808506e-07,
      "loss": 1.5832,
      "step": 2286
    },
    {
      "epoch": 0.3449211974964181,
      "grad_norm": 0.25406449718842056,
      "learning_rate": 3.803171847378461e-07,
      "loss": 1.5424,
      "step": 2287
    },
    {
      "epoch": 0.3450720156850916,
      "grad_norm": 0.23913358715230615,
      "learning_rate": 3.80222958400832e-07,
      "loss": 1.5886,
      "step": 2288
    },
    {
      "epoch": 0.3452228338737652,
      "grad_norm": 0.6101911876858062,
      "learning_rate": 3.8012870843819934e-07,
      "loss": 1.6251,
      "step": 2289
    },
    {
      "epoch": 0.34537365206243875,
      "grad_norm": 0.2509776593810166,
      "learning_rate": 3.8003443487110995e-07,
      "loss": 1.6073,
      "step": 2290
    },
    {
      "epoch": 0.34552447025111227,
      "grad_norm": 0.2422242713596001,
      "learning_rate": 3.79940137720731e-07,
      "loss": 1.5951,
      "step": 2291
    },
    {
      "epoch": 0.34567528843978584,
      "grad_norm": 0.24824865731058396,
      "learning_rate": 3.79845817008235e-07,
      "loss": 1.5508,
      "step": 2292
    },
    {
      "epoch": 0.3458261066284594,
      "grad_norm": 0.25626805095048233,
      "learning_rate": 3.7975147275479947e-07,
      "loss": 1.6062,
      "step": 2293
    },
    {
      "epoch": 0.34597692481713294,
      "grad_norm": 0.2942650248769778,
      "learning_rate": 3.7965710498160767e-07,
      "loss": 1.6115,
      "step": 2294
    },
    {
      "epoch": 0.3461277430058065,
      "grad_norm": 0.24907806981702166,
      "learning_rate": 3.795627137098479e-07,
      "loss": 1.5798,
      "step": 2295
    },
    {
      "epoch": 0.34627856119448003,
      "grad_norm": 0.2536491718380689,
      "learning_rate": 3.794682989607137e-07,
      "loss": 1.5955,
      "step": 2296
    },
    {
      "epoch": 0.3464293793831536,
      "grad_norm": 0.2536962046438531,
      "learning_rate": 3.793738607554039e-07,
      "loss": 1.5237,
      "step": 2297
    },
    {
      "epoch": 0.3465801975718272,
      "grad_norm": 0.27702294905095187,
      "learning_rate": 3.792793991151226e-07,
      "loss": 1.5761,
      "step": 2298
    },
    {
      "epoch": 0.3467310157605007,
      "grad_norm": 0.2552120558949902,
      "learning_rate": 3.791849140610794e-07,
      "loss": 1.5531,
      "step": 2299
    },
    {
      "epoch": 0.34688183394917427,
      "grad_norm": 0.2844932996298561,
      "learning_rate": 3.790904056144887e-07,
      "loss": 1.5634,
      "step": 2300
    },
    {
      "epoch": 0.34703265213784784,
      "grad_norm": 0.31350790888658125,
      "learning_rate": 3.7899587379657043e-07,
      "loss": 1.6519,
      "step": 2301
    },
    {
      "epoch": 0.34718347032652136,
      "grad_norm": 0.327166975783665,
      "learning_rate": 3.789013186285499e-07,
      "loss": 1.5463,
      "step": 2302
    },
    {
      "epoch": 0.34733428851519493,
      "grad_norm": 0.3671001662868849,
      "learning_rate": 3.7880674013165737e-07,
      "loss": 1.4951,
      "step": 2303
    },
    {
      "epoch": 0.3474851067038685,
      "grad_norm": 0.25791788975509533,
      "learning_rate": 3.7871213832712847e-07,
      "loss": 1.5153,
      "step": 2304
    },
    {
      "epoch": 0.347635924892542,
      "grad_norm": 0.25043027941632756,
      "learning_rate": 3.7861751323620405e-07,
      "loss": 1.5284,
      "step": 2305
    },
    {
      "epoch": 0.3477867430812156,
      "grad_norm": 0.27288469174574215,
      "learning_rate": 3.785228648801304e-07,
      "loss": 1.5774,
      "step": 2306
    },
    {
      "epoch": 0.3479375612698892,
      "grad_norm": 0.24291337383099834,
      "learning_rate": 3.7842819328015847e-07,
      "loss": 1.5544,
      "step": 2307
    },
    {
      "epoch": 0.3480883794585627,
      "grad_norm": 0.2462920734808384,
      "learning_rate": 3.783334984575451e-07,
      "loss": 1.4761,
      "step": 2308
    },
    {
      "epoch": 0.34823919764723626,
      "grad_norm": 0.24337486105201564,
      "learning_rate": 3.7823878043355185e-07,
      "loss": 1.5438,
      "step": 2309
    },
    {
      "epoch": 0.3483900158359098,
      "grad_norm": 0.24901269879840904,
      "learning_rate": 3.781440392294458e-07,
      "loss": 1.5417,
      "step": 2310
    },
    {
      "epoch": 0.34854083402458336,
      "grad_norm": 0.2574773833389837,
      "learning_rate": 3.78049274866499e-07,
      "loss": 1.5779,
      "step": 2311
    },
    {
      "epoch": 0.34869165221325693,
      "grad_norm": 0.27317633569887817,
      "learning_rate": 3.7795448736598895e-07,
      "loss": 1.6136,
      "step": 2312
    },
    {
      "epoch": 0.34884247040193045,
      "grad_norm": 0.24881592357265409,
      "learning_rate": 3.7785967674919806e-07,
      "loss": 1.6093,
      "step": 2313
    },
    {
      "epoch": 0.348993288590604,
      "grad_norm": 0.24956684849307417,
      "learning_rate": 3.7776484303741414e-07,
      "loss": 1.5973,
      "step": 2314
    },
    {
      "epoch": 0.3491441067792776,
      "grad_norm": 0.2464129365453703,
      "learning_rate": 3.776699862519301e-07,
      "loss": 1.5477,
      "step": 2315
    },
    {
      "epoch": 0.3492949249679511,
      "grad_norm": 0.25191976238902325,
      "learning_rate": 3.775751064140441e-07,
      "loss": 1.5582,
      "step": 2316
    },
    {
      "epoch": 0.3494457431566247,
      "grad_norm": 0.264147892478328,
      "learning_rate": 3.7748020354505936e-07,
      "loss": 1.6021,
      "step": 2317
    },
    {
      "epoch": 0.34959656134529826,
      "grad_norm": 0.2757351180919783,
      "learning_rate": 3.773852776662844e-07,
      "loss": 1.5334,
      "step": 2318
    },
    {
      "epoch": 0.3497473795339718,
      "grad_norm": 0.24862941758346846,
      "learning_rate": 3.7729032879903285e-07,
      "loss": 1.5957,
      "step": 2319
    },
    {
      "epoch": 0.34989819772264535,
      "grad_norm": 0.24897857861626696,
      "learning_rate": 3.771953569646234e-07,
      "loss": 1.5855,
      "step": 2320
    },
    {
      "epoch": 0.35004901591131893,
      "grad_norm": 0.2517463859076541,
      "learning_rate": 3.7710036218438007e-07,
      "loss": 1.5371,
      "step": 2321
    },
    {
      "epoch": 0.35019983409999245,
      "grad_norm": 0.268956424751186,
      "learning_rate": 3.7700534447963194e-07,
      "loss": 1.4844,
      "step": 2322
    },
    {
      "epoch": 0.350350652288666,
      "grad_norm": 0.2500277125108567,
      "learning_rate": 3.769103038717133e-07,
      "loss": 1.52,
      "step": 2323
    },
    {
      "epoch": 0.3505014704773396,
      "grad_norm": 0.27279449038724674,
      "learning_rate": 3.7681524038196337e-07,
      "loss": 1.5496,
      "step": 2324
    },
    {
      "epoch": 0.3506522886660131,
      "grad_norm": 0.2521931291040799,
      "learning_rate": 3.7672015403172687e-07,
      "loss": 1.5905,
      "step": 2325
    },
    {
      "epoch": 0.3508031068546867,
      "grad_norm": 0.24490012882151652,
      "learning_rate": 3.766250448423534e-07,
      "loss": 1.5401,
      "step": 2326
    },
    {
      "epoch": 0.3509539250433602,
      "grad_norm": 0.2924535818177954,
      "learning_rate": 3.765299128351976e-07,
      "loss": 1.5453,
      "step": 2327
    },
    {
      "epoch": 0.3511047432320338,
      "grad_norm": 0.29600395237439076,
      "learning_rate": 3.764347580316196e-07,
      "loss": 1.5717,
      "step": 2328
    },
    {
      "epoch": 0.35125556142070735,
      "grad_norm": 0.26532230372293647,
      "learning_rate": 3.7633958045298433e-07,
      "loss": 1.5291,
      "step": 2329
    },
    {
      "epoch": 0.35140637960938087,
      "grad_norm": 0.24916894584154997,
      "learning_rate": 3.762443801206618e-07,
      "loss": 1.5405,
      "step": 2330
    },
    {
      "epoch": 0.35155719779805444,
      "grad_norm": 0.2581134867558839,
      "learning_rate": 3.761491570560274e-07,
      "loss": 1.5962,
      "step": 2331
    },
    {
      "epoch": 0.351708015986728,
      "grad_norm": 0.2438935147855879,
      "learning_rate": 3.760539112804615e-07,
      "loss": 1.58,
      "step": 2332
    },
    {
      "epoch": 0.35185883417540154,
      "grad_norm": 0.37578295769326836,
      "learning_rate": 3.7595864281534945e-07,
      "loss": 1.5589,
      "step": 2333
    },
    {
      "epoch": 0.3520096523640751,
      "grad_norm": 0.24486877017889822,
      "learning_rate": 3.7586335168208187e-07,
      "loss": 1.5367,
      "step": 2334
    },
    {
      "epoch": 0.3521604705527487,
      "grad_norm": 0.2429857832769579,
      "learning_rate": 3.7576803790205447e-07,
      "loss": 1.4925,
      "step": 2335
    },
    {
      "epoch": 0.3523112887414222,
      "grad_norm": 0.2397245100035915,
      "learning_rate": 3.7567270149666775e-07,
      "loss": 1.5823,
      "step": 2336
    },
    {
      "epoch": 0.3524621069300958,
      "grad_norm": 0.2500161975176447,
      "learning_rate": 3.7557734248732757e-07,
      "loss": 1.6248,
      "step": 2337
    },
    {
      "epoch": 0.35261292511876935,
      "grad_norm": 0.28561743382712795,
      "learning_rate": 3.7548196089544505e-07,
      "loss": 1.5107,
      "step": 2338
    },
    {
      "epoch": 0.35276374330744287,
      "grad_norm": 0.26571887001887556,
      "learning_rate": 3.753865567424358e-07,
      "loss": 1.5116,
      "step": 2339
    },
    {
      "epoch": 0.35291456149611644,
      "grad_norm": 0.25473540068810213,
      "learning_rate": 3.7529113004972114e-07,
      "loss": 1.5812,
      "step": 2340
    },
    {
      "epoch": 0.35306537968478996,
      "grad_norm": 0.2555471401859078,
      "learning_rate": 3.751956808387268e-07,
      "loss": 1.4831,
      "step": 2341
    },
    {
      "epoch": 0.35321619787346353,
      "grad_norm": 0.27184839392756677,
      "learning_rate": 3.751002091308842e-07,
      "loss": 1.5669,
      "step": 2342
    },
    {
      "epoch": 0.3533670160621371,
      "grad_norm": 0.2515942203557161,
      "learning_rate": 3.750047149476294e-07,
      "loss": 1.6251,
      "step": 2343
    },
    {
      "epoch": 0.3535178342508106,
      "grad_norm": 0.26112307976157956,
      "learning_rate": 3.749091983104036e-07,
      "loss": 1.5943,
      "step": 2344
    },
    {
      "epoch": 0.3536686524394842,
      "grad_norm": 0.2501421913523755,
      "learning_rate": 3.748136592406531e-07,
      "loss": 1.5759,
      "step": 2345
    },
    {
      "epoch": 0.3538194706281578,
      "grad_norm": 0.24752102036976434,
      "learning_rate": 3.747180977598293e-07,
      "loss": 1.5804,
      "step": 2346
    },
    {
      "epoch": 0.3539702888168313,
      "grad_norm": 0.2455951689312032,
      "learning_rate": 3.746225138893883e-07,
      "loss": 1.5124,
      "step": 2347
    },
    {
      "epoch": 0.35412110700550486,
      "grad_norm": 0.24339549004166428,
      "learning_rate": 3.745269076507918e-07,
      "loss": 1.5778,
      "step": 2348
    },
    {
      "epoch": 0.35427192519417844,
      "grad_norm": 0.7443699191473532,
      "learning_rate": 3.744312790655058e-07,
      "loss": 1.5718,
      "step": 2349
    },
    {
      "epoch": 0.35442274338285196,
      "grad_norm": 0.24946128343868357,
      "learning_rate": 3.743356281550019e-07,
      "loss": 1.5991,
      "step": 2350
    },
    {
      "epoch": 0.35457356157152553,
      "grad_norm": 0.24247758143910678,
      "learning_rate": 3.742399549407566e-07,
      "loss": 1.4819,
      "step": 2351
    },
    {
      "epoch": 0.3547243797601991,
      "grad_norm": 0.2386100112502874,
      "learning_rate": 3.741442594442512e-07,
      "loss": 1.5642,
      "step": 2352
    },
    {
      "epoch": 0.3548751979488726,
      "grad_norm": 0.2505357416538226,
      "learning_rate": 3.7404854168697214e-07,
      "loss": 1.5825,
      "step": 2353
    },
    {
      "epoch": 0.3550260161375462,
      "grad_norm": 0.28867608528120875,
      "learning_rate": 3.7395280169041087e-07,
      "loss": 1.5876,
      "step": 2354
    },
    {
      "epoch": 0.3551768343262197,
      "grad_norm": 0.241730587623635,
      "learning_rate": 3.7385703947606374e-07,
      "loss": 1.5736,
      "step": 2355
    },
    {
      "epoch": 0.3553276525148933,
      "grad_norm": 0.24831574140721346,
      "learning_rate": 3.7376125506543227e-07,
      "loss": 1.6067,
      "step": 2356
    },
    {
      "epoch": 0.35547847070356686,
      "grad_norm": 0.2590154730802224,
      "learning_rate": 3.7366544848002277e-07,
      "loss": 1.5333,
      "step": 2357
    },
    {
      "epoch": 0.3556292888922404,
      "grad_norm": 0.31178147754240004,
      "learning_rate": 3.735696197413467e-07,
      "loss": 1.4965,
      "step": 2358
    },
    {
      "epoch": 0.35578010708091395,
      "grad_norm": 1.3552928491220932,
      "learning_rate": 3.7347376887092026e-07,
      "loss": 1.5818,
      "step": 2359
    },
    {
      "epoch": 0.35593092526958753,
      "grad_norm": 0.24500608110449468,
      "learning_rate": 3.733778958902649e-07,
      "loss": 1.5467,
      "step": 2360
    },
    {
      "epoch": 0.35608174345826105,
      "grad_norm": 0.25089900091150935,
      "learning_rate": 3.7328200082090677e-07,
      "loss": 1.5542,
      "step": 2361
    },
    {
      "epoch": 0.3562325616469346,
      "grad_norm": 0.24158656678132237,
      "learning_rate": 3.7318608368437716e-07,
      "loss": 1.5095,
      "step": 2362
    },
    {
      "epoch": 0.3563833798356082,
      "grad_norm": 0.2534149802126099,
      "learning_rate": 3.7309014450221225e-07,
      "loss": 1.5902,
      "step": 2363
    },
    {
      "epoch": 0.3565341980242817,
      "grad_norm": 0.24798687422411522,
      "learning_rate": 3.729941832959533e-07,
      "loss": 1.5484,
      "step": 2364
    },
    {
      "epoch": 0.3566850162129553,
      "grad_norm": 0.25046258736392985,
      "learning_rate": 3.728982000871462e-07,
      "loss": 1.5405,
      "step": 2365
    },
    {
      "epoch": 0.35683583440162886,
      "grad_norm": 0.2440667878039132,
      "learning_rate": 3.728021948973421e-07,
      "loss": 1.6005,
      "step": 2366
    },
    {
      "epoch": 0.3569866525903024,
      "grad_norm": 0.24610788855981183,
      "learning_rate": 3.7270616774809693e-07,
      "loss": 1.5601,
      "step": 2367
    },
    {
      "epoch": 0.35713747077897595,
      "grad_norm": 0.9454335873908799,
      "learning_rate": 3.726101186609715e-07,
      "loss": 1.5024,
      "step": 2368
    },
    {
      "epoch": 0.3572882889676495,
      "grad_norm": 0.28378028432667163,
      "learning_rate": 3.7251404765753194e-07,
      "loss": 1.557,
      "step": 2369
    },
    {
      "epoch": 0.35743910715632304,
      "grad_norm": 0.2557935616464503,
      "learning_rate": 3.7241795475934856e-07,
      "loss": 1.5365,
      "step": 2370
    },
    {
      "epoch": 0.3575899253449966,
      "grad_norm": 0.2549332722675419,
      "learning_rate": 3.7232183998799727e-07,
      "loss": 1.4674,
      "step": 2371
    },
    {
      "epoch": 0.35774074353367014,
      "grad_norm": 1.1517601933256894,
      "learning_rate": 3.722257033650586e-07,
      "loss": 1.5681,
      "step": 2372
    },
    {
      "epoch": 0.3578915617223437,
      "grad_norm": 0.272656754166095,
      "learning_rate": 3.72129544912118e-07,
      "loss": 1.5631,
      "step": 2373
    },
    {
      "epoch": 0.3580423799110173,
      "grad_norm": 0.29022175862146116,
      "learning_rate": 3.7203336465076573e-07,
      "loss": 1.5758,
      "step": 2374
    },
    {
      "epoch": 0.3581931980996908,
      "grad_norm": 0.24817613832794372,
      "learning_rate": 3.719371626025973e-07,
      "loss": 1.6106,
      "step": 2375
    },
    {
      "epoch": 0.3583440162883644,
      "grad_norm": 0.24399194599218416,
      "learning_rate": 3.7184093878921267e-07,
      "loss": 1.4977,
      "step": 2376
    },
    {
      "epoch": 0.35849483447703795,
      "grad_norm": 0.25767051635975635,
      "learning_rate": 3.7174469323221704e-07,
      "loss": 1.5812,
      "step": 2377
    },
    {
      "epoch": 0.35864565266571147,
      "grad_norm": 0.2523174711087523,
      "learning_rate": 3.7164842595322015e-07,
      "loss": 1.5518,
      "step": 2378
    },
    {
      "epoch": 0.35879647085438504,
      "grad_norm": 0.26428032115396755,
      "learning_rate": 3.71552136973837e-07,
      "loss": 1.6004,
      "step": 2379
    },
    {
      "epoch": 0.3589472890430586,
      "grad_norm": 0.2598334701344459,
      "learning_rate": 3.714558263156872e-07,
      "loss": 1.5377,
      "step": 2380
    },
    {
      "epoch": 0.35909810723173213,
      "grad_norm": 0.24231961396705243,
      "learning_rate": 3.713594940003952e-07,
      "loss": 1.5659,
      "step": 2381
    },
    {
      "epoch": 0.3592489254204057,
      "grad_norm": 0.2557782748611694,
      "learning_rate": 3.7126314004959056e-07,
      "loss": 1.4764,
      "step": 2382
    },
    {
      "epoch": 0.3593997436090793,
      "grad_norm": 0.32643135012971164,
      "learning_rate": 3.7116676448490754e-07,
      "loss": 1.5139,
      "step": 2383
    },
    {
      "epoch": 0.3595505617977528,
      "grad_norm": 0.24383151316441518,
      "learning_rate": 3.710703673279852e-07,
      "loss": 1.537,
      "step": 2384
    },
    {
      "epoch": 0.3597013799864264,
      "grad_norm": 0.2505535549620646,
      "learning_rate": 3.709739486004674e-07,
      "loss": 1.5429,
      "step": 2385
    },
    {
      "epoch": 0.3598521981750999,
      "grad_norm": 0.26628913276403715,
      "learning_rate": 3.708775083240031e-07,
      "loss": 1.6109,
      "step": 2386
    },
    {
      "epoch": 0.36000301636377346,
      "grad_norm": 0.2653073300714376,
      "learning_rate": 3.70781046520246e-07,
      "loss": 1.6284,
      "step": 2387
    },
    {
      "epoch": 0.36015383455244704,
      "grad_norm": 0.3151596090823107,
      "learning_rate": 3.706845632108545e-07,
      "loss": 1.6701,
      "step": 2388
    },
    {
      "epoch": 0.36030465274112056,
      "grad_norm": 0.2580701573700956,
      "learning_rate": 3.705880584174919e-07,
      "loss": 1.5415,
      "step": 2389
    },
    {
      "epoch": 0.36045547092979413,
      "grad_norm": 0.45731481557385795,
      "learning_rate": 3.704915321618263e-07,
      "loss": 1.5663,
      "step": 2390
    },
    {
      "epoch": 0.3606062891184677,
      "grad_norm": 0.29332627821859003,
      "learning_rate": 3.7039498446553073e-07,
      "loss": 1.541,
      "step": 2391
    },
    {
      "epoch": 0.3607571073071412,
      "grad_norm": 0.2539419562347637,
      "learning_rate": 3.7029841535028295e-07,
      "loss": 1.5933,
      "step": 2392
    },
    {
      "epoch": 0.3609079254958148,
      "grad_norm": 0.2487819911491069,
      "learning_rate": 3.702018248377656e-07,
      "loss": 1.5854,
      "step": 2393
    },
    {
      "epoch": 0.36105874368448837,
      "grad_norm": 0.2540973512020146,
      "learning_rate": 3.7010521294966583e-07,
      "loss": 1.5584,
      "step": 2394
    },
    {
      "epoch": 0.3612095618731619,
      "grad_norm": 0.276230147107862,
      "learning_rate": 3.700085797076761e-07,
      "loss": 1.5397,
      "step": 2395
    },
    {
      "epoch": 0.36136038006183546,
      "grad_norm": 0.25044667399851733,
      "learning_rate": 3.6991192513349324e-07,
      "loss": 1.5411,
      "step": 2396
    },
    {
      "epoch": 0.36151119825050904,
      "grad_norm": 0.27066315255091705,
      "learning_rate": 3.698152492488191e-07,
      "loss": 1.5477,
      "step": 2397
    },
    {
      "epoch": 0.36166201643918255,
      "grad_norm": 0.24415988508449232,
      "learning_rate": 3.697185520753601e-07,
      "loss": 1.5201,
      "step": 2398
    },
    {
      "epoch": 0.36181283462785613,
      "grad_norm": 0.24378718076324166,
      "learning_rate": 3.696218336348276e-07,
      "loss": 1.5561,
      "step": 2399
    },
    {
      "epoch": 0.36196365281652965,
      "grad_norm": 0.24117983955024197,
      "learning_rate": 3.695250939489379e-07,
      "loss": 1.5658,
      "step": 2400
    },
    {
      "epoch": 0.3621144710052032,
      "grad_norm": 0.2494661857308847,
      "learning_rate": 3.6942833303941163e-07,
      "loss": 1.596,
      "step": 2401
    },
    {
      "epoch": 0.3622652891938768,
      "grad_norm": 0.2497563413431003,
      "learning_rate": 3.6933155092797457e-07,
      "loss": 1.506,
      "step": 2402
    },
    {
      "epoch": 0.3624161073825503,
      "grad_norm": 0.23857213517373116,
      "learning_rate": 3.6923474763635716e-07,
      "loss": 1.5657,
      "step": 2403
    },
    {
      "epoch": 0.3625669255712239,
      "grad_norm": 0.30240876054872073,
      "learning_rate": 3.6913792318629433e-07,
      "loss": 1.5096,
      "step": 2404
    },
    {
      "epoch": 0.36271774375989746,
      "grad_norm": 0.25689137093919195,
      "learning_rate": 3.6904107759952623e-07,
      "loss": 1.574,
      "step": 2405
    },
    {
      "epoch": 0.362868561948571,
      "grad_norm": 0.24164303407146231,
      "learning_rate": 3.689442108977974e-07,
      "loss": 1.599,
      "step": 2406
    },
    {
      "epoch": 0.36301938013724455,
      "grad_norm": 0.2397255897218177,
      "learning_rate": 3.688473231028574e-07,
      "loss": 1.5725,
      "step": 2407
    },
    {
      "epoch": 0.3631701983259181,
      "grad_norm": 0.2946173688800962,
      "learning_rate": 3.687504142364601e-07,
      "loss": 1.5733,
      "step": 2408
    },
    {
      "epoch": 0.36332101651459164,
      "grad_norm": 0.257563218080772,
      "learning_rate": 3.6865348432036453e-07,
      "loss": 1.5593,
      "step": 2409
    },
    {
      "epoch": 0.3634718347032652,
      "grad_norm": 0.2525257965158378,
      "learning_rate": 3.6855653337633423e-07,
      "loss": 1.5775,
      "step": 2410
    },
    {
      "epoch": 0.3636226528919388,
      "grad_norm": 0.2549787899101132,
      "learning_rate": 3.684595614261375e-07,
      "loss": 1.5236,
      "step": 2411
    },
    {
      "epoch": 0.3637734710806123,
      "grad_norm": 0.25792198225445445,
      "learning_rate": 3.6836256849154735e-07,
      "loss": 1.5561,
      "step": 2412
    },
    {
      "epoch": 0.3639242892692859,
      "grad_norm": 0.24801130882772623,
      "learning_rate": 3.6826555459434165e-07,
      "loss": 1.4832,
      "step": 2413
    },
    {
      "epoch": 0.36407510745795946,
      "grad_norm": 0.2441895056224972,
      "learning_rate": 3.681685197563027e-07,
      "loss": 1.5012,
      "step": 2414
    },
    {
      "epoch": 0.364225925646633,
      "grad_norm": 0.24712875621506114,
      "learning_rate": 3.680714639992177e-07,
      "loss": 1.5336,
      "step": 2415
    },
    {
      "epoch": 0.36437674383530655,
      "grad_norm": 0.2522388109306788,
      "learning_rate": 3.679743873448785e-07,
      "loss": 1.6083,
      "step": 2416
    },
    {
      "epoch": 0.36452756202398007,
      "grad_norm": 0.5903504095147515,
      "learning_rate": 3.678772898150816e-07,
      "loss": 1.6368,
      "step": 2417
    },
    {
      "epoch": 0.36467838021265364,
      "grad_norm": 0.2597251356207736,
      "learning_rate": 3.6778017143162826e-07,
      "loss": 1.5642,
      "step": 2418
    },
    {
      "epoch": 0.3648291984013272,
      "grad_norm": 0.2531960272531877,
      "learning_rate": 3.676830322163244e-07,
      "loss": 1.5711,
      "step": 2419
    },
    {
      "epoch": 0.36498001659000073,
      "grad_norm": 0.7327336650049859,
      "learning_rate": 3.6758587219098055e-07,
      "loss": 1.58,
      "step": 2420
    },
    {
      "epoch": 0.3651308347786743,
      "grad_norm": 0.7568978409238439,
      "learning_rate": 3.674886913774121e-07,
      "loss": 1.5805,
      "step": 2421
    },
    {
      "epoch": 0.3652816529673479,
      "grad_norm": 0.2606031974117596,
      "learning_rate": 3.673914897974387e-07,
      "loss": 1.6141,
      "step": 2422
    },
    {
      "epoch": 0.3654324711560214,
      "grad_norm": 0.25452997507792435,
      "learning_rate": 3.6729426747288523e-07,
      "loss": 1.66,
      "step": 2423
    },
    {
      "epoch": 0.365583289344695,
      "grad_norm": 0.25396229477622617,
      "learning_rate": 3.671970244255807e-07,
      "loss": 1.5081,
      "step": 2424
    },
    {
      "epoch": 0.36573410753336855,
      "grad_norm": 0.25362236963149887,
      "learning_rate": 3.670997606773592e-07,
      "loss": 1.5591,
      "step": 2425
    },
    {
      "epoch": 0.36588492572204206,
      "grad_norm": 0.24657346521114726,
      "learning_rate": 3.670024762500592e-07,
      "loss": 1.565,
      "step": 2426
    },
    {
      "epoch": 0.36603574391071564,
      "grad_norm": 0.25369751050593276,
      "learning_rate": 3.6690517116552384e-07,
      "loss": 1.6045,
      "step": 2427
    },
    {
      "epoch": 0.3661865620993892,
      "grad_norm": 0.2416199710906676,
      "learning_rate": 3.6680784544560096e-07,
      "loss": 1.5997,
      "step": 2428
    },
    {
      "epoch": 0.36633738028806273,
      "grad_norm": 0.25616058190831803,
      "learning_rate": 3.66710499112143e-07,
      "loss": 1.6184,
      "step": 2429
    },
    {
      "epoch": 0.3664881984767363,
      "grad_norm": 0.2598470043980163,
      "learning_rate": 3.666131321870072e-07,
      "loss": 1.5097,
      "step": 2430
    },
    {
      "epoch": 0.3666390166654098,
      "grad_norm": 0.25397390756446714,
      "learning_rate": 3.6651574469205503e-07,
      "loss": 1.4104,
      "step": 2431
    },
    {
      "epoch": 0.3667898348540834,
      "grad_norm": 0.2532940923495218,
      "learning_rate": 3.66418336649153e-07,
      "loss": 1.5688,
      "step": 2432
    },
    {
      "epoch": 0.36694065304275697,
      "grad_norm": 0.26372885031044346,
      "learning_rate": 3.6632090808017196e-07,
      "loss": 1.5981,
      "step": 2433
    },
    {
      "epoch": 0.3670914712314305,
      "grad_norm": 0.736836117933467,
      "learning_rate": 3.662234590069876e-07,
      "loss": 1.5987,
      "step": 2434
    },
    {
      "epoch": 0.36724228942010406,
      "grad_norm": 0.25180599095166145,
      "learning_rate": 3.661259894514798e-07,
      "loss": 1.572,
      "step": 2435
    },
    {
      "epoch": 0.36739310760877764,
      "grad_norm": 0.2522775024815644,
      "learning_rate": 3.6602849943553366e-07,
      "loss": 1.5828,
      "step": 2436
    },
    {
      "epoch": 0.36754392579745115,
      "grad_norm": 0.2879773326396739,
      "learning_rate": 3.6593098898103825e-07,
      "loss": 1.5937,
      "step": 2437
    },
    {
      "epoch": 0.36769474398612473,
      "grad_norm": 0.24090414223098844,
      "learning_rate": 3.658334581098877e-07,
      "loss": 1.6163,
      "step": 2438
    },
    {
      "epoch": 0.3678455621747983,
      "grad_norm": 0.24308021934072732,
      "learning_rate": 3.6573590684398036e-07,
      "loss": 1.5329,
      "step": 2439
    },
    {
      "epoch": 0.3679963803634718,
      "grad_norm": 0.254692886434861,
      "learning_rate": 3.6563833520521944e-07,
      "loss": 1.5369,
      "step": 2440
    },
    {
      "epoch": 0.3681471985521454,
      "grad_norm": 0.2737286974799631,
      "learning_rate": 3.6554074321551263e-07,
      "loss": 1.5459,
      "step": 2441
    },
    {
      "epoch": 0.36829801674081897,
      "grad_norm": 0.23623753229922384,
      "learning_rate": 3.6544313089677204e-07,
      "loss": 1.57,
      "step": 2442
    },
    {
      "epoch": 0.3684488349294925,
      "grad_norm": 0.2555802056096531,
      "learning_rate": 3.6534549827091465e-07,
      "loss": 1.4908,
      "step": 2443
    },
    {
      "epoch": 0.36859965311816606,
      "grad_norm": 0.26079223078952796,
      "learning_rate": 3.652478453598617e-07,
      "loss": 1.5763,
      "step": 2444
    },
    {
      "epoch": 0.3687504713068396,
      "grad_norm": 0.25604931144429594,
      "learning_rate": 3.651501721855392e-07,
      "loss": 1.5196,
      "step": 2445
    },
    {
      "epoch": 0.36890128949551315,
      "grad_norm": 0.29826940759380655,
      "learning_rate": 3.6505247876987756e-07,
      "loss": 1.5709,
      "step": 2446
    },
    {
      "epoch": 0.3690521076841867,
      "grad_norm": 0.25470009222890144,
      "learning_rate": 3.649547651348118e-07,
      "loss": 1.5033,
      "step": 2447
    },
    {
      "epoch": 0.36920292587286024,
      "grad_norm": 0.24832644332181872,
      "learning_rate": 3.6485703130228147e-07,
      "loss": 1.5298,
      "step": 2448
    },
    {
      "epoch": 0.3693537440615338,
      "grad_norm": 0.23751675097100958,
      "learning_rate": 3.647592772942308e-07,
      "loss": 1.5558,
      "step": 2449
    },
    {
      "epoch": 0.3695045622502074,
      "grad_norm": 0.3061235443052442,
      "learning_rate": 3.6466150313260813e-07,
      "loss": 1.5705,
      "step": 2450
    },
    {
      "epoch": 0.3696553804388809,
      "grad_norm": 0.2420150027388054,
      "learning_rate": 3.6456370883936686e-07,
      "loss": 1.5607,
      "step": 2451
    },
    {
      "epoch": 0.3698061986275545,
      "grad_norm": 0.24859998354389287,
      "learning_rate": 3.644658944364646e-07,
      "loss": 1.5607,
      "step": 2452
    },
    {
      "epoch": 0.36995701681622806,
      "grad_norm": 0.24962571236153827,
      "learning_rate": 3.643680599458635e-07,
      "loss": 1.5867,
      "step": 2453
    },
    {
      "epoch": 0.3701078350049016,
      "grad_norm": 0.31443800935488536,
      "learning_rate": 3.6427020538953025e-07,
      "loss": 1.5697,
      "step": 2454
    },
    {
      "epoch": 0.37025865319357515,
      "grad_norm": 0.24373642702895276,
      "learning_rate": 3.64172330789436e-07,
      "loss": 1.5622,
      "step": 2455
    },
    {
      "epoch": 0.3704094713822487,
      "grad_norm": 0.25621149690462364,
      "learning_rate": 3.6407443616755657e-07,
      "loss": 1.5073,
      "step": 2456
    },
    {
      "epoch": 0.37056028957092224,
      "grad_norm": 0.25762597568130835,
      "learning_rate": 3.6397652154587206e-07,
      "loss": 1.6092,
      "step": 2457
    },
    {
      "epoch": 0.3707111077595958,
      "grad_norm": 1.1919798777471118,
      "learning_rate": 3.6387858694636715e-07,
      "loss": 1.5425,
      "step": 2458
    },
    {
      "epoch": 0.3708619259482694,
      "grad_norm": 0.251421077271863,
      "learning_rate": 3.6378063239103113e-07,
      "loss": 1.4724,
      "step": 2459
    },
    {
      "epoch": 0.3710127441369429,
      "grad_norm": 0.25366741666087783,
      "learning_rate": 3.6368265790185746e-07,
      "loss": 1.5215,
      "step": 2460
    },
    {
      "epoch": 0.3711635623256165,
      "grad_norm": 0.4207178094706561,
      "learning_rate": 3.635846635008444e-07,
      "loss": 1.5114,
      "step": 2461
    },
    {
      "epoch": 0.37131438051429,
      "grad_norm": 0.32196750384289907,
      "learning_rate": 3.6348664920999456e-07,
      "loss": 1.6209,
      "step": 2462
    },
    {
      "epoch": 0.3714651987029636,
      "grad_norm": 0.2495336315511228,
      "learning_rate": 3.633886150513149e-07,
      "loss": 1.5458,
      "step": 2463
    },
    {
      "epoch": 0.37161601689163715,
      "grad_norm": 0.7472867025315102,
      "learning_rate": 3.6329056104681704e-07,
      "loss": 1.563,
      "step": 2464
    },
    {
      "epoch": 0.37176683508031066,
      "grad_norm": 0.29741446110595743,
      "learning_rate": 3.6319248721851685e-07,
      "loss": 1.5654,
      "step": 2465
    },
    {
      "epoch": 0.37191765326898424,
      "grad_norm": 0.3236843198755218,
      "learning_rate": 3.630943935884349e-07,
      "loss": 1.4952,
      "step": 2466
    },
    {
      "epoch": 0.3720684714576578,
      "grad_norm": 0.2594476741448998,
      "learning_rate": 3.62996280178596e-07,
      "loss": 1.5008,
      "step": 2467
    },
    {
      "epoch": 0.37221928964633133,
      "grad_norm": 0.2538384281299253,
      "learning_rate": 3.6289814701102943e-07,
      "loss": 1.6065,
      "step": 2468
    },
    {
      "epoch": 0.3723701078350049,
      "grad_norm": 0.27122348792809003,
      "learning_rate": 3.62799994107769e-07,
      "loss": 1.5326,
      "step": 2469
    },
    {
      "epoch": 0.3725209260236785,
      "grad_norm": 0.25404092193633077,
      "learning_rate": 3.627018214908528e-07,
      "loss": 1.5207,
      "step": 2470
    },
    {
      "epoch": 0.372671744212352,
      "grad_norm": 0.24982996965510668,
      "learning_rate": 3.626036291823236e-07,
      "loss": 1.566,
      "step": 2471
    },
    {
      "epoch": 0.37282256240102557,
      "grad_norm": 0.24340133877324788,
      "learning_rate": 3.625054172042284e-07,
      "loss": 1.5153,
      "step": 2472
    },
    {
      "epoch": 0.37297338058969914,
      "grad_norm": 0.2449532469636368,
      "learning_rate": 3.624071855786185e-07,
      "loss": 1.5486,
      "step": 2473
    },
    {
      "epoch": 0.37312419877837266,
      "grad_norm": 0.24393436396790796,
      "learning_rate": 3.623089343275499e-07,
      "loss": 1.5293,
      "step": 2474
    },
    {
      "epoch": 0.37327501696704624,
      "grad_norm": 0.2439837149359486,
      "learning_rate": 3.6221066347308276e-07,
      "loss": 1.5383,
      "step": 2475
    },
    {
      "epoch": 0.37342583515571975,
      "grad_norm": 0.24727496115441294,
      "learning_rate": 3.621123730372819e-07,
      "loss": 1.5189,
      "step": 2476
    },
    {
      "epoch": 0.37357665334439333,
      "grad_norm": 0.2502123111049387,
      "learning_rate": 3.6201406304221625e-07,
      "loss": 1.6118,
      "step": 2477
    },
    {
      "epoch": 0.3737274715330669,
      "grad_norm": 0.2442722196888879,
      "learning_rate": 3.619157335099593e-07,
      "loss": 1.6215,
      "step": 2478
    },
    {
      "epoch": 0.3738782897217404,
      "grad_norm": 0.2798737546309143,
      "learning_rate": 3.6181738446258897e-07,
      "loss": 1.5433,
      "step": 2479
    },
    {
      "epoch": 0.374029107910414,
      "grad_norm": 0.2702349908306434,
      "learning_rate": 3.617190159221874e-07,
      "loss": 1.6105,
      "step": 2480
    },
    {
      "epoch": 0.37417992609908757,
      "grad_norm": 0.25956006639582185,
      "learning_rate": 3.6162062791084124e-07,
      "loss": 1.4806,
      "step": 2481
    },
    {
      "epoch": 0.3743307442877611,
      "grad_norm": 0.25139763886130506,
      "learning_rate": 3.6152222045064146e-07,
      "loss": 1.5881,
      "step": 2482
    },
    {
      "epoch": 0.37448156247643466,
      "grad_norm": 0.277809605479455,
      "learning_rate": 3.614237935636833e-07,
      "loss": 1.6212,
      "step": 2483
    },
    {
      "epoch": 0.37463238066510823,
      "grad_norm": 0.273163332381341,
      "learning_rate": 3.613253472720666e-07,
      "loss": 1.5924,
      "step": 2484
    },
    {
      "epoch": 0.37478319885378175,
      "grad_norm": 0.23879197574640515,
      "learning_rate": 3.6122688159789533e-07,
      "loss": 1.6423,
      "step": 2485
    },
    {
      "epoch": 0.3749340170424553,
      "grad_norm": 0.2546038431049256,
      "learning_rate": 3.61128396563278e-07,
      "loss": 1.5171,
      "step": 2486
    },
    {
      "epoch": 0.3750848352311289,
      "grad_norm": 0.23953276895403006,
      "learning_rate": 3.6102989219032733e-07,
      "loss": 1.5643,
      "step": 2487
    },
    {
      "epoch": 0.3752356534198024,
      "grad_norm": 0.2906806335181811,
      "learning_rate": 3.609313685011603e-07,
      "loss": 1.5522,
      "step": 2488
    },
    {
      "epoch": 0.375386471608476,
      "grad_norm": 0.2750965974767434,
      "learning_rate": 3.608328255178984e-07,
      "loss": 1.5947,
      "step": 2489
    },
    {
      "epoch": 0.3755372897971495,
      "grad_norm": 0.24602389200087268,
      "learning_rate": 3.6073426326266754e-07,
      "loss": 1.5711,
      "step": 2490
    },
    {
      "epoch": 0.3756881079858231,
      "grad_norm": 0.25933028795860896,
      "learning_rate": 3.6063568175759774e-07,
      "loss": 1.5236,
      "step": 2491
    },
    {
      "epoch": 0.37583892617449666,
      "grad_norm": 0.2505269949410393,
      "learning_rate": 3.6053708102482333e-07,
      "loss": 1.5529,
      "step": 2492
    },
    {
      "epoch": 0.3759897443631702,
      "grad_norm": 0.2430829129132987,
      "learning_rate": 3.6043846108648317e-07,
      "loss": 1.4762,
      "step": 2493
    },
    {
      "epoch": 0.37614056255184375,
      "grad_norm": 0.2449505059322527,
      "learning_rate": 3.6033982196472023e-07,
      "loss": 1.5179,
      "step": 2494
    },
    {
      "epoch": 0.3762913807405173,
      "grad_norm": 0.2507260037548037,
      "learning_rate": 3.602411636816819e-07,
      "loss": 1.5196,
      "step": 2495
    },
    {
      "epoch": 0.37644219892919084,
      "grad_norm": 0.37584163526536496,
      "learning_rate": 3.6014248625951984e-07,
      "loss": 1.5285,
      "step": 2496
    },
    {
      "epoch": 0.3765930171178644,
      "grad_norm": 0.25765837448196127,
      "learning_rate": 3.6004378972038995e-07,
      "loss": 1.5765,
      "step": 2497
    },
    {
      "epoch": 0.376743835306538,
      "grad_norm": 0.2524347758209021,
      "learning_rate": 3.5994507408645254e-07,
      "loss": 1.5726,
      "step": 2498
    },
    {
      "epoch": 0.3768946534952115,
      "grad_norm": 0.2464378769806035,
      "learning_rate": 3.598463393798721e-07,
      "loss": 1.5018,
      "step": 2499
    },
    {
      "epoch": 0.3770454716838851,
      "grad_norm": 0.3205722010062426,
      "learning_rate": 3.597475856228175e-07,
      "loss": 1.5475,
      "step": 2500
    },
    {
      "epoch": 0.37719628987255865,
      "grad_norm": 0.2973979291512468,
      "learning_rate": 3.596488128374618e-07,
      "loss": 1.6129,
      "step": 2501
    },
    {
      "epoch": 0.3773471080612322,
      "grad_norm": 0.25193464512230107,
      "learning_rate": 3.5955002104598233e-07,
      "loss": 1.639,
      "step": 2502
    },
    {
      "epoch": 0.37749792624990575,
      "grad_norm": 0.2564177040565832,
      "learning_rate": 3.5945121027056085e-07,
      "loss": 1.5603,
      "step": 2503
    },
    {
      "epoch": 0.3776487444385793,
      "grad_norm": 0.25736901468499984,
      "learning_rate": 3.5935238053338306e-07,
      "loss": 1.5685,
      "step": 2504
    },
    {
      "epoch": 0.37779956262725284,
      "grad_norm": 0.3265279800860292,
      "learning_rate": 3.592535318566393e-07,
      "loss": 1.5626,
      "step": 2505
    },
    {
      "epoch": 0.3779503808159264,
      "grad_norm": 0.2624954697548865,
      "learning_rate": 3.5915466426252395e-07,
      "loss": 1.4911,
      "step": 2506
    },
    {
      "epoch": 0.37810119900459993,
      "grad_norm": 0.6135301735853039,
      "learning_rate": 3.590557777732355e-07,
      "loss": 1.5582,
      "step": 2507
    },
    {
      "epoch": 0.3782520171932735,
      "grad_norm": 0.4120016764911682,
      "learning_rate": 3.589568724109771e-07,
      "loss": 1.5042,
      "step": 2508
    },
    {
      "epoch": 0.3784028353819471,
      "grad_norm": 0.2421724367048175,
      "learning_rate": 3.588579481979556e-07,
      "loss": 1.5401,
      "step": 2509
    },
    {
      "epoch": 0.3785536535706206,
      "grad_norm": 0.2530420392567595,
      "learning_rate": 3.587590051563826e-07,
      "loss": 1.5677,
      "step": 2510
    },
    {
      "epoch": 0.37870447175929417,
      "grad_norm": 0.2468715043787992,
      "learning_rate": 3.586600433084737e-07,
      "loss": 1.5502,
      "step": 2511
    },
    {
      "epoch": 0.37885528994796774,
      "grad_norm": 0.2524770444053603,
      "learning_rate": 3.585610626764485e-07,
      "loss": 1.6097,
      "step": 2512
    },
    {
      "epoch": 0.37900610813664126,
      "grad_norm": 0.2620919997505972,
      "learning_rate": 3.584620632825312e-07,
      "loss": 1.6128,
      "step": 2513
    },
    {
      "epoch": 0.37915692632531484,
      "grad_norm": 0.2658547516077316,
      "learning_rate": 3.5836304514894997e-07,
      "loss": 1.5252,
      "step": 2514
    },
    {
      "epoch": 0.3793077445139884,
      "grad_norm": 0.24404822219882544,
      "learning_rate": 3.5826400829793734e-07,
      "loss": 1.6027,
      "step": 2515
    },
    {
      "epoch": 0.37945856270266193,
      "grad_norm": 0.2468154377798453,
      "learning_rate": 3.5816495275173e-07,
      "loss": 1.5466,
      "step": 2516
    },
    {
      "epoch": 0.3796093808913355,
      "grad_norm": 0.24188988756601712,
      "learning_rate": 3.580658785325686e-07,
      "loss": 1.5999,
      "step": 2517
    },
    {
      "epoch": 0.3797601990800091,
      "grad_norm": 0.23888994835645322,
      "learning_rate": 3.5796678566269847e-07,
      "loss": 1.5828,
      "step": 2518
    },
    {
      "epoch": 0.3799110172686826,
      "grad_norm": 0.24089657083130359,
      "learning_rate": 3.578676741643686e-07,
      "loss": 1.5927,
      "step": 2519
    },
    {
      "epoch": 0.38006183545735617,
      "grad_norm": 0.2424084665110092,
      "learning_rate": 3.5776854405983247e-07,
      "loss": 1.5275,
      "step": 2520
    },
    {
      "epoch": 0.3802126536460297,
      "grad_norm": 0.2433942345662491,
      "learning_rate": 3.5766939537134785e-07,
      "loss": 1.529,
      "step": 2521
    },
    {
      "epoch": 0.38036347183470326,
      "grad_norm": 0.25291436213073637,
      "learning_rate": 3.575702281211762e-07,
      "loss": 1.5307,
      "step": 2522
    },
    {
      "epoch": 0.38051429002337683,
      "grad_norm": 0.2539334006876831,
      "learning_rate": 3.5747104233158376e-07,
      "loss": 1.5848,
      "step": 2523
    },
    {
      "epoch": 0.38066510821205035,
      "grad_norm": 0.27029092109111064,
      "learning_rate": 3.573718380248404e-07,
      "loss": 1.5669,
      "step": 2524
    },
    {
      "epoch": 0.3808159264007239,
      "grad_norm": 0.25850559989477917,
      "learning_rate": 3.5727261522322053e-07,
      "loss": 1.5996,
      "step": 2525
    },
    {
      "epoch": 0.3809667445893975,
      "grad_norm": 0.2671086272117505,
      "learning_rate": 3.571733739490025e-07,
      "loss": 1.5646,
      "step": 2526
    },
    {
      "epoch": 0.381117562778071,
      "grad_norm": 0.2408624437543867,
      "learning_rate": 3.5707411422446887e-07,
      "loss": 1.5752,
      "step": 2527
    },
    {
      "epoch": 0.3812683809667446,
      "grad_norm": 0.25504735550342,
      "learning_rate": 3.5697483607190634e-07,
      "loss": 1.5252,
      "step": 2528
    },
    {
      "epoch": 0.38141919915541816,
      "grad_norm": 0.3044244213239282,
      "learning_rate": 3.5687553951360565e-07,
      "loss": 1.5546,
      "step": 2529
    },
    {
      "epoch": 0.3815700173440917,
      "grad_norm": 0.9044492121709756,
      "learning_rate": 3.5677622457186197e-07,
      "loss": 1.5406,
      "step": 2530
    },
    {
      "epoch": 0.38172083553276526,
      "grad_norm": 0.25451969623397536,
      "learning_rate": 3.566768912689743e-07,
      "loss": 1.5538,
      "step": 2531
    },
    {
      "epoch": 0.38187165372143883,
      "grad_norm": 0.24781650874974845,
      "learning_rate": 3.5657753962724574e-07,
      "loss": 1.6394,
      "step": 2532
    },
    {
      "epoch": 0.38202247191011235,
      "grad_norm": 0.2556923716105965,
      "learning_rate": 3.5647816966898376e-07,
      "loss": 1.5072,
      "step": 2533
    },
    {
      "epoch": 0.3821732900987859,
      "grad_norm": 0.25536826723602823,
      "learning_rate": 3.5637878141649986e-07,
      "loss": 1.587,
      "step": 2534
    },
    {
      "epoch": 0.38232410828745944,
      "grad_norm": 0.35107493199094714,
      "learning_rate": 3.5627937489210946e-07,
      "loss": 1.5287,
      "step": 2535
    },
    {
      "epoch": 0.382474926476133,
      "grad_norm": 0.25482272417125473,
      "learning_rate": 3.5617995011813217e-07,
      "loss": 1.5592,
      "step": 2536
    },
    {
      "epoch": 0.3826257446648066,
      "grad_norm": 0.23779771356649057,
      "learning_rate": 3.5608050711689206e-07,
      "loss": 1.5347,
      "step": 2537
    },
    {
      "epoch": 0.3827765628534801,
      "grad_norm": 0.2536661448143563,
      "learning_rate": 3.5598104591071666e-07,
      "loss": 1.6121,
      "step": 2538
    },
    {
      "epoch": 0.3829273810421537,
      "grad_norm": 0.2743867604526638,
      "learning_rate": 3.55881566521938e-07,
      "loss": 1.6331,
      "step": 2539
    },
    {
      "epoch": 0.38307819923082725,
      "grad_norm": 0.28125217492724036,
      "learning_rate": 3.557820689728922e-07,
      "loss": 1.5647,
      "step": 2540
    },
    {
      "epoch": 0.3832290174195008,
      "grad_norm": 0.29985129969621616,
      "learning_rate": 3.5568255328591927e-07,
      "loss": 1.5693,
      "step": 2541
    },
    {
      "epoch": 0.38337983560817435,
      "grad_norm": 0.2575088246208123,
      "learning_rate": 3.555830194833634e-07,
      "loss": 1.5679,
      "step": 2542
    },
    {
      "epoch": 0.3835306537968479,
      "grad_norm": 0.24227748590037862,
      "learning_rate": 3.5548346758757276e-07,
      "loss": 1.5884,
      "step": 2543
    },
    {
      "epoch": 0.38368147198552144,
      "grad_norm": 0.25730342177076415,
      "learning_rate": 3.553838976208998e-07,
      "loss": 1.5064,
      "step": 2544
    },
    {
      "epoch": 0.383832290174195,
      "grad_norm": 0.2518233947961992,
      "learning_rate": 3.552843096057007e-07,
      "loss": 1.5662,
      "step": 2545
    },
    {
      "epoch": 0.3839831083628686,
      "grad_norm": 0.24662947641756058,
      "learning_rate": 3.5518470356433607e-07,
      "loss": 1.533,
      "step": 2546
    },
    {
      "epoch": 0.3841339265515421,
      "grad_norm": 0.2707362685244374,
      "learning_rate": 3.550850795191701e-07,
      "loss": 1.5642,
      "step": 2547
    },
    {
      "epoch": 0.3842847447402157,
      "grad_norm": 0.24828838160982408,
      "learning_rate": 3.549854374925715e-07,
      "loss": 1.6159,
      "step": 2548
    },
    {
      "epoch": 0.38443556292888925,
      "grad_norm": 0.32263784919365934,
      "learning_rate": 3.548857775069128e-07,
      "loss": 1.5652,
      "step": 2549
    },
    {
      "epoch": 0.38458638111756277,
      "grad_norm": 0.24798883657905776,
      "learning_rate": 3.547860995845705e-07,
      "loss": 1.5941,
      "step": 2550
    },
    {
      "epoch": 0.38473719930623634,
      "grad_norm": 0.24788484749743359,
      "learning_rate": 3.546864037479252e-07,
      "loss": 1.575,
      "step": 2551
    },
    {
      "epoch": 0.38488801749490986,
      "grad_norm": 0.2679950197180121,
      "learning_rate": 3.5458669001936154e-07,
      "loss": 1.5955,
      "step": 2552
    },
    {
      "epoch": 0.38503883568358344,
      "grad_norm": 0.27920578096628507,
      "learning_rate": 3.5448695842126815e-07,
      "loss": 1.5511,
      "step": 2553
    },
    {
      "epoch": 0.385189653872257,
      "grad_norm": 0.26415238272720953,
      "learning_rate": 3.543872089760376e-07,
      "loss": 1.5579,
      "step": 2554
    },
    {
      "epoch": 0.38534047206093053,
      "grad_norm": 0.27149305935859813,
      "learning_rate": 3.542874417060667e-07,
      "loss": 1.4529,
      "step": 2555
    },
    {
      "epoch": 0.3854912902496041,
      "grad_norm": 0.250792164223544,
      "learning_rate": 3.5418765663375613e-07,
      "loss": 1.4984,
      "step": 2556
    },
    {
      "epoch": 0.3856421084382777,
      "grad_norm": 0.2616412145153655,
      "learning_rate": 3.540878537815103e-07,
      "loss": 1.5504,
      "step": 2557
    },
    {
      "epoch": 0.3857929266269512,
      "grad_norm": 0.2538882269923289,
      "learning_rate": 3.53988033171738e-07,
      "loss": 1.6398,
      "step": 2558
    },
    {
      "epoch": 0.38594374481562477,
      "grad_norm": 0.2474357832914362,
      "learning_rate": 3.53888194826852e-07,
      "loss": 1.6093,
      "step": 2559
    },
    {
      "epoch": 0.38609456300429834,
      "grad_norm": 0.2960300497422995,
      "learning_rate": 3.5378833876926865e-07,
      "loss": 1.5457,
      "step": 2560
    },
    {
      "epoch": 0.38624538119297186,
      "grad_norm": 0.25897439161674723,
      "learning_rate": 3.536884650214087e-07,
      "loss": 1.5842,
      "step": 2561
    },
    {
      "epoch": 0.38639619938164543,
      "grad_norm": 0.25717842800991464,
      "learning_rate": 3.535885736056968e-07,
      "loss": 1.6571,
      "step": 2562
    },
    {
      "epoch": 0.386547017570319,
      "grad_norm": 0.24206278661173006,
      "learning_rate": 3.534886645445613e-07,
      "loss": 1.5238,
      "step": 2563
    },
    {
      "epoch": 0.3866978357589925,
      "grad_norm": 0.25262257528774823,
      "learning_rate": 3.5338873786043464e-07,
      "loss": 1.5816,
      "step": 2564
    },
    {
      "epoch": 0.3868486539476661,
      "grad_norm": 0.26635351438110183,
      "learning_rate": 3.5328879357575357e-07,
      "loss": 1.5669,
      "step": 2565
    },
    {
      "epoch": 0.3869994721363396,
      "grad_norm": 0.23778541687692611,
      "learning_rate": 3.5318883171295827e-07,
      "loss": 1.6074,
      "step": 2566
    },
    {
      "epoch": 0.3871502903250132,
      "grad_norm": 0.3082502795672136,
      "learning_rate": 3.5308885229449314e-07,
      "loss": 1.5698,
      "step": 2567
    },
    {
      "epoch": 0.38730110851368676,
      "grad_norm": 0.2716911724218832,
      "learning_rate": 3.529888553428065e-07,
      "loss": 1.6068,
      "step": 2568
    },
    {
      "epoch": 0.3874519267023603,
      "grad_norm": 0.24057573264881182,
      "learning_rate": 3.5288884088035047e-07,
      "loss": 1.5755,
      "step": 2569
    },
    {
      "epoch": 0.38760274489103386,
      "grad_norm": 0.3001263217639516,
      "learning_rate": 3.527888089295813e-07,
      "loss": 1.5804,
      "step": 2570
    },
    {
      "epoch": 0.38775356307970743,
      "grad_norm": 0.2849484147258995,
      "learning_rate": 3.526887595129592e-07,
      "loss": 1.6406,
      "step": 2571
    },
    {
      "epoch": 0.38790438126838095,
      "grad_norm": 0.45158680187400113,
      "learning_rate": 3.525886926529479e-07,
      "loss": 1.5028,
      "step": 2572
    },
    {
      "epoch": 0.3880551994570545,
      "grad_norm": 0.23841489018388096,
      "learning_rate": 3.524886083720155e-07,
      "loss": 1.5066,
      "step": 2573
    },
    {
      "epoch": 0.3882060176457281,
      "grad_norm": 0.2505767757173705,
      "learning_rate": 3.523885066926338e-07,
      "loss": 1.5941,
      "step": 2574
    },
    {
      "epoch": 0.3883568358344016,
      "grad_norm": 0.24869451937061052,
      "learning_rate": 3.522883876372786e-07,
      "loss": 1.5655,
      "step": 2575
    },
    {
      "epoch": 0.3885076540230752,
      "grad_norm": 0.2506663800167466,
      "learning_rate": 3.521882512284294e-07,
      "loss": 1.5076,
      "step": 2576
    },
    {
      "epoch": 0.38865847221174876,
      "grad_norm": 0.27321653566711673,
      "learning_rate": 3.5208809748856993e-07,
      "loss": 1.5595,
      "step": 2577
    },
    {
      "epoch": 0.3888092904004223,
      "grad_norm": 0.27974757704273684,
      "learning_rate": 3.519879264401874e-07,
      "loss": 1.5089,
      "step": 2578
    },
    {
      "epoch": 0.38896010858909585,
      "grad_norm": 0.46818369455980297,
      "learning_rate": 3.518877381057733e-07,
      "loss": 1.6114,
      "step": 2579
    },
    {
      "epoch": 0.3891109267777694,
      "grad_norm": 0.25803071382785336,
      "learning_rate": 3.517875325078229e-07,
      "loss": 1.5559,
      "step": 2580
    },
    {
      "epoch": 0.38926174496644295,
      "grad_norm": 0.2583288630929102,
      "learning_rate": 3.51687309668835e-07,
      "loss": 1.5375,
      "step": 2581
    },
    {
      "epoch": 0.3894125631551165,
      "grad_norm": 0.2640446446214051,
      "learning_rate": 3.515870696113127e-07,
      "loss": 1.6234,
      "step": 2582
    },
    {
      "epoch": 0.38956338134379004,
      "grad_norm": 0.27696063835751633,
      "learning_rate": 3.5148681235776285e-07,
      "loss": 1.5855,
      "step": 2583
    },
    {
      "epoch": 0.3897141995324636,
      "grad_norm": 0.2573817113265274,
      "learning_rate": 3.513865379306961e-07,
      "loss": 1.5196,
      "step": 2584
    },
    {
      "epoch": 0.3898650177211372,
      "grad_norm": 0.2423886129051474,
      "learning_rate": 3.512862463526269e-07,
      "loss": 1.5065,
      "step": 2585
    },
    {
      "epoch": 0.3900158359098107,
      "grad_norm": 4.06785605153624,
      "learning_rate": 3.511859376460738e-07,
      "loss": 1.5648,
      "step": 2586
    },
    {
      "epoch": 0.3901666540984843,
      "grad_norm": 0.2635253748519167,
      "learning_rate": 3.5108561183355884e-07,
      "loss": 1.6157,
      "step": 2587
    },
    {
      "epoch": 0.39031747228715785,
      "grad_norm": 0.2505502056943804,
      "learning_rate": 3.5098526893760823e-07,
      "loss": 1.6346,
      "step": 2588
    },
    {
      "epoch": 0.39046829047583137,
      "grad_norm": 0.2530678378552461,
      "learning_rate": 3.508849089807517e-07,
      "loss": 1.5542,
      "step": 2589
    },
    {
      "epoch": 0.39061910866450494,
      "grad_norm": 0.26292279162317655,
      "learning_rate": 3.507845319855232e-07,
      "loss": 1.56,
      "step": 2590
    },
    {
      "epoch": 0.3907699268531785,
      "grad_norm": 0.29078243973984197,
      "learning_rate": 3.506841379744601e-07,
      "loss": 1.5713,
      "step": 2591
    },
    {
      "epoch": 0.39092074504185204,
      "grad_norm": 0.32868406059323346,
      "learning_rate": 3.5058372697010384e-07,
      "loss": 1.6039,
      "step": 2592
    },
    {
      "epoch": 0.3910715632305256,
      "grad_norm": 0.2525036536065519,
      "learning_rate": 3.5048329899499975e-07,
      "loss": 1.5555,
      "step": 2593
    },
    {
      "epoch": 0.39122238141919913,
      "grad_norm": 0.2630087537592004,
      "learning_rate": 3.5038285407169665e-07,
      "loss": 1.5709,
      "step": 2594
    },
    {
      "epoch": 0.3913731996078727,
      "grad_norm": 0.28143126812196245,
      "learning_rate": 3.502823922227474e-07,
      "loss": 1.5727,
      "step": 2595
    },
    {
      "epoch": 0.3915240177965463,
      "grad_norm": 0.24490217881819218,
      "learning_rate": 3.501819134707087e-07,
      "loss": 1.6159,
      "step": 2596
    },
    {
      "epoch": 0.3916748359852198,
      "grad_norm": 0.7937926014143071,
      "learning_rate": 3.5008141783814077e-07,
      "loss": 1.5408,
      "step": 2597
    },
    {
      "epoch": 0.39182565417389337,
      "grad_norm": 0.24505278919691315,
      "learning_rate": 3.4998090534760804e-07,
      "loss": 1.6168,
      "step": 2598
    },
    {
      "epoch": 0.39197647236256694,
      "grad_norm": 0.2536043806196611,
      "learning_rate": 3.498803760216783e-07,
      "loss": 1.6077,
      "step": 2599
    },
    {
      "epoch": 0.39212729055124046,
      "grad_norm": 0.40477068068604105,
      "learning_rate": 3.497798298829234e-07,
      "loss": 1.5067,
      "step": 2600
    },
    {
      "epoch": 0.39227810873991403,
      "grad_norm": 0.24658759789845558,
      "learning_rate": 3.4967926695391885e-07,
      "loss": 1.5888,
      "step": 2601
    },
    {
      "epoch": 0.3924289269285876,
      "grad_norm": 0.250593063485952,
      "learning_rate": 3.49578687257244e-07,
      "loss": 1.6084,
      "step": 2602
    },
    {
      "epoch": 0.3925797451172611,
      "grad_norm": 0.24253691878675834,
      "learning_rate": 3.494780908154819e-07,
      "loss": 1.6294,
      "step": 2603
    },
    {
      "epoch": 0.3927305633059347,
      "grad_norm": 0.24103654342297792,
      "learning_rate": 3.4937747765121934e-07,
      "loss": 1.5471,
      "step": 2604
    },
    {
      "epoch": 0.3928813814946083,
      "grad_norm": 0.2728927050656147,
      "learning_rate": 3.4927684778704693e-07,
      "loss": 1.6348,
      "step": 2605
    },
    {
      "epoch": 0.3930321996832818,
      "grad_norm": 0.24506085358646462,
      "learning_rate": 3.491762012455591e-07,
      "loss": 1.4937,
      "step": 2606
    },
    {
      "epoch": 0.39318301787195536,
      "grad_norm": 0.24014219757507166,
      "learning_rate": 3.4907553804935376e-07,
      "loss": 1.538,
      "step": 2607
    },
    {
      "epoch": 0.39333383606062894,
      "grad_norm": 0.24429115753577182,
      "learning_rate": 3.489748582210328e-07,
      "loss": 1.5113,
      "step": 2608
    },
    {
      "epoch": 0.39348465424930246,
      "grad_norm": 0.28537403136894296,
      "learning_rate": 3.4887416178320185e-07,
      "loss": 1.568,
      "step": 2609
    },
    {
      "epoch": 0.39363547243797603,
      "grad_norm": 0.23825991470195018,
      "learning_rate": 3.4877344875847e-07,
      "loss": 1.5273,
      "step": 2610
    },
    {
      "epoch": 0.39378629062664955,
      "grad_norm": 0.2576386407608002,
      "learning_rate": 3.486727191694505e-07,
      "loss": 1.5482,
      "step": 2611
    },
    {
      "epoch": 0.3939371088153231,
      "grad_norm": 0.23839533111212974,
      "learning_rate": 3.485719730387599e-07,
      "loss": 1.5535,
      "step": 2612
    },
    {
      "epoch": 0.3940879270039967,
      "grad_norm": 0.2576875948496164,
      "learning_rate": 3.484712103890187e-07,
      "loss": 1.5805,
      "step": 2613
    },
    {
      "epoch": 0.3942387451926702,
      "grad_norm": 0.2578129139016205,
      "learning_rate": 3.483704312428511e-07,
      "loss": 1.5268,
      "step": 2614
    },
    {
      "epoch": 0.3943895633813438,
      "grad_norm": 0.25788059541015274,
      "learning_rate": 3.4826963562288493e-07,
      "loss": 1.5542,
      "step": 2615
    },
    {
      "epoch": 0.39454038157001736,
      "grad_norm": 0.2755588250827172,
      "learning_rate": 3.4816882355175164e-07,
      "loss": 1.5969,
      "step": 2616
    },
    {
      "epoch": 0.3946911997586909,
      "grad_norm": 0.25099898704009094,
      "learning_rate": 3.4806799505208654e-07,
      "loss": 1.5318,
      "step": 2617
    },
    {
      "epoch": 0.39484201794736445,
      "grad_norm": 0.24912268915215166,
      "learning_rate": 3.479671501465286e-07,
      "loss": 1.6251,
      "step": 2618
    },
    {
      "epoch": 0.39499283613603803,
      "grad_norm": 0.33127514454847884,
      "learning_rate": 3.478662888577205e-07,
      "loss": 1.5788,
      "step": 2619
    },
    {
      "epoch": 0.39514365432471155,
      "grad_norm": 0.25157259363870266,
      "learning_rate": 3.4776541120830824e-07,
      "loss": 1.5696,
      "step": 2620
    },
    {
      "epoch": 0.3952944725133851,
      "grad_norm": 0.2738464876476835,
      "learning_rate": 3.476645172209422e-07,
      "loss": 1.5516,
      "step": 2621
    },
    {
      "epoch": 0.3954452907020587,
      "grad_norm": 0.23811298747874665,
      "learning_rate": 3.475636069182758e-07,
      "loss": 1.6193,
      "step": 2622
    },
    {
      "epoch": 0.3955961088907322,
      "grad_norm": 0.25875497952173504,
      "learning_rate": 3.474626803229662e-07,
      "loss": 1.5521,
      "step": 2623
    },
    {
      "epoch": 0.3957469270794058,
      "grad_norm": 0.28847182070562355,
      "learning_rate": 3.473617374576747e-07,
      "loss": 1.5747,
      "step": 2624
    },
    {
      "epoch": 0.3958977452680793,
      "grad_norm": 0.23824845500467842,
      "learning_rate": 3.472607783450657e-07,
      "loss": 1.5624,
      "step": 2625
    },
    {
      "epoch": 0.3960485634567529,
      "grad_norm": 0.2542487631269686,
      "learning_rate": 3.471598030078074e-07,
      "loss": 1.5358,
      "step": 2626
    },
    {
      "epoch": 0.39619938164542645,
      "grad_norm": 0.2554835540513806,
      "learning_rate": 3.4705881146857186e-07,
      "loss": 1.6046,
      "step": 2627
    },
    {
      "epoch": 0.39635019983409997,
      "grad_norm": 0.23936580356092402,
      "learning_rate": 3.469578037500346e-07,
      "loss": 1.5496,
      "step": 2628
    },
    {
      "epoch": 0.39650101802277354,
      "grad_norm": 0.24235177378643663,
      "learning_rate": 3.4685677987487464e-07,
      "loss": 1.6188,
      "step": 2629
    },
    {
      "epoch": 0.3966518362114471,
      "grad_norm": 0.24893632276790695,
      "learning_rate": 3.4675573986577505e-07,
      "loss": 1.5966,
      "step": 2630
    },
    {
      "epoch": 0.39680265440012064,
      "grad_norm": 0.2629841664263705,
      "learning_rate": 3.4665468374542197e-07,
      "loss": 1.5352,
      "step": 2631
    },
    {
      "epoch": 0.3969534725887942,
      "grad_norm": 0.25793233524671794,
      "learning_rate": 3.4655361153650565e-07,
      "loss": 1.5239,
      "step": 2632
    },
    {
      "epoch": 0.3971042907774678,
      "grad_norm": 0.24429982363546654,
      "learning_rate": 3.4645252326171966e-07,
      "loss": 1.6371,
      "step": 2633
    },
    {
      "epoch": 0.3972551089661413,
      "grad_norm": 0.2573467243824734,
      "learning_rate": 3.463514189437613e-07,
      "loss": 1.5225,
      "step": 2634
    },
    {
      "epoch": 0.3974059271548149,
      "grad_norm": 0.679063985170996,
      "learning_rate": 3.4625029860533134e-07,
      "loss": 1.5862,
      "step": 2635
    },
    {
      "epoch": 0.39755674534348845,
      "grad_norm": 0.25171052854332737,
      "learning_rate": 3.4614916226913433e-07,
      "loss": 1.5616,
      "step": 2636
    },
    {
      "epoch": 0.39770756353216197,
      "grad_norm": 0.24633977010308533,
      "learning_rate": 3.4604800995787826e-07,
      "loss": 1.5499,
      "step": 2637
    },
    {
      "epoch": 0.39785838172083554,
      "grad_norm": 0.2951019560113865,
      "learning_rate": 3.4594684169427487e-07,
      "loss": 1.6254,
      "step": 2638
    },
    {
      "epoch": 0.39800919990950906,
      "grad_norm": 0.23811915492407157,
      "learning_rate": 3.4584565750103927e-07,
      "loss": 1.5325,
      "step": 2639
    },
    {
      "epoch": 0.39816001809818263,
      "grad_norm": 0.24558452212160967,
      "learning_rate": 3.4574445740089036e-07,
      "loss": 1.5681,
      "step": 2640
    },
    {
      "epoch": 0.3983108362868562,
      "grad_norm": 0.25258021327735125,
      "learning_rate": 3.456432414165504e-07,
      "loss": 1.5854,
      "step": 2641
    },
    {
      "epoch": 0.3984616544755297,
      "grad_norm": 0.24705997248524708,
      "learning_rate": 3.4554200957074535e-07,
      "loss": 1.4883,
      "step": 2642
    },
    {
      "epoch": 0.3986124726642033,
      "grad_norm": 0.9393781438635932,
      "learning_rate": 3.4544076188620484e-07,
      "loss": 1.5811,
      "step": 2643
    },
    {
      "epoch": 0.3987632908528769,
      "grad_norm": 0.24081186996868634,
      "learning_rate": 3.453394983856618e-07,
      "loss": 1.5615,
      "step": 2644
    },
    {
      "epoch": 0.3989141090415504,
      "grad_norm": 0.3548482655547463,
      "learning_rate": 3.4523821909185287e-07,
      "loss": 1.5285,
      "step": 2645
    },
    {
      "epoch": 0.39906492723022396,
      "grad_norm": 0.25244220556229424,
      "learning_rate": 3.4513692402751813e-07,
      "loss": 1.5206,
      "step": 2646
    },
    {
      "epoch": 0.39921574541889754,
      "grad_norm": 0.3010429865855147,
      "learning_rate": 3.4503561321540137e-07,
      "loss": 1.5322,
      "step": 2647
    },
    {
      "epoch": 0.39936656360757106,
      "grad_norm": 0.24745554269634792,
      "learning_rate": 3.4493428667824976e-07,
      "loss": 1.5899,
      "step": 2648
    },
    {
      "epoch": 0.39951738179624463,
      "grad_norm": 0.24692220967298187,
      "learning_rate": 3.4483294443881414e-07,
      "loss": 1.5691,
      "step": 2649
    },
    {
      "epoch": 0.3996681999849182,
      "grad_norm": 0.24957805000851577,
      "learning_rate": 3.4473158651984866e-07,
      "loss": 1.5357,
      "step": 2650
    },
    {
      "epoch": 0.3998190181735917,
      "grad_norm": 0.26422320009653594,
      "learning_rate": 3.4463021294411125e-07,
      "loss": 1.5988,
      "step": 2651
    },
    {
      "epoch": 0.3999698363622653,
      "grad_norm": 0.2620999742863255,
      "learning_rate": 3.4452882373436316e-07,
      "loss": 1.5631,
      "step": 2652
    },
    {
      "epoch": 0.40012065455093887,
      "grad_norm": 0.261503235413909,
      "learning_rate": 3.4442741891336926e-07,
      "loss": 1.6121,
      "step": 2653
    },
    {
      "epoch": 0.4002714727396124,
      "grad_norm": 0.2491475970261047,
      "learning_rate": 3.443259985038978e-07,
      "loss": 1.5968,
      "step": 2654
    },
    {
      "epoch": 0.40042229092828596,
      "grad_norm": 0.25473496113658356,
      "learning_rate": 3.4422456252872074e-07,
      "loss": 1.6159,
      "step": 2655
    },
    {
      "epoch": 0.4005731091169595,
      "grad_norm": 0.2796363857378898,
      "learning_rate": 3.441231110106133e-07,
      "loss": 1.5169,
      "step": 2656
    },
    {
      "epoch": 0.40072392730563305,
      "grad_norm": 0.2580527347727552,
      "learning_rate": 3.440216439723544e-07,
      "loss": 1.5556,
      "step": 2657
    },
    {
      "epoch": 0.40087474549430663,
      "grad_norm": 0.2961651530051946,
      "learning_rate": 3.439201614367263e-07,
      "loss": 1.5758,
      "step": 2658
    },
    {
      "epoch": 0.40102556368298015,
      "grad_norm": 0.24727990480200315,
      "learning_rate": 3.4381866342651474e-07,
      "loss": 1.559,
      "step": 2659
    },
    {
      "epoch": 0.4011763818716537,
      "grad_norm": 1.8441125513901913,
      "learning_rate": 3.43717149964509e-07,
      "loss": 1.5534,
      "step": 2660
    },
    {
      "epoch": 0.4013272000603273,
      "grad_norm": 0.4130681269520023,
      "learning_rate": 3.436156210735019e-07,
      "loss": 1.5958,
      "step": 2661
    },
    {
      "epoch": 0.4014780182490008,
      "grad_norm": 0.24928135649559705,
      "learning_rate": 3.4351407677628955e-07,
      "loss": 1.6431,
      "step": 2662
    },
    {
      "epoch": 0.4016288364376744,
      "grad_norm": 0.2557444447850399,
      "learning_rate": 3.434125170956716e-07,
      "loss": 1.555,
      "step": 2663
    },
    {
      "epoch": 0.40177965462634796,
      "grad_norm": 0.2569607174542637,
      "learning_rate": 3.433109420544511e-07,
      "loss": 1.5394,
      "step": 2664
    },
    {
      "epoch": 0.4019304728150215,
      "grad_norm": 0.2550208704852457,
      "learning_rate": 3.4320935167543476e-07,
      "loss": 1.5759,
      "step": 2665
    },
    {
      "epoch": 0.40208129100369505,
      "grad_norm": 0.2462457579459392,
      "learning_rate": 3.4310774598143244e-07,
      "loss": 1.5612,
      "step": 2666
    },
    {
      "epoch": 0.4022321091923686,
      "grad_norm": 0.2539463489198951,
      "learning_rate": 3.430061249952577e-07,
      "loss": 1.5093,
      "step": 2667
    },
    {
      "epoch": 0.40238292738104214,
      "grad_norm": 0.25143123583424176,
      "learning_rate": 3.429044887397273e-07,
      "loss": 1.5635,
      "step": 2668
    },
    {
      "epoch": 0.4025337455697157,
      "grad_norm": 0.2546528852562432,
      "learning_rate": 3.4280283723766156e-07,
      "loss": 1.6254,
      "step": 2669
    },
    {
      "epoch": 0.40268456375838924,
      "grad_norm": 0.2577657012079826,
      "learning_rate": 3.4270117051188427e-07,
      "loss": 1.4889,
      "step": 2670
    },
    {
      "epoch": 0.4028353819470628,
      "grad_norm": 0.2614460935676659,
      "learning_rate": 3.425994885852225e-07,
      "loss": 1.6185,
      "step": 2671
    },
    {
      "epoch": 0.4029862001357364,
      "grad_norm": 0.3979414514693881,
      "learning_rate": 3.424977914805068e-07,
      "loss": 1.558,
      "step": 2672
    },
    {
      "epoch": 0.4031370183244099,
      "grad_norm": 0.24555757646921053,
      "learning_rate": 3.4239607922057126e-07,
      "loss": 1.5621,
      "step": 2673
    },
    {
      "epoch": 0.4032878365130835,
      "grad_norm": 0.26714140749175647,
      "learning_rate": 3.42294351828253e-07,
      "loss": 1.507,
      "step": 2674
    },
    {
      "epoch": 0.40343865470175705,
      "grad_norm": 0.25886542993727496,
      "learning_rate": 3.42192609326393e-07,
      "loss": 1.6575,
      "step": 2675
    },
    {
      "epoch": 0.40358947289043057,
      "grad_norm": 0.2431401880097489,
      "learning_rate": 3.420908517378353e-07,
      "loss": 1.588,
      "step": 2676
    },
    {
      "epoch": 0.40374029107910414,
      "grad_norm": 0.23557654628041694,
      "learning_rate": 3.4198907908542756e-07,
      "loss": 1.5192,
      "step": 2677
    },
    {
      "epoch": 0.4038911092677777,
      "grad_norm": 0.28842546678351816,
      "learning_rate": 3.418872913920206e-07,
      "loss": 1.6055,
      "step": 2678
    },
    {
      "epoch": 0.40404192745645123,
      "grad_norm": 0.24739791616041887,
      "learning_rate": 3.4178548868046877e-07,
      "loss": 1.5644,
      "step": 2679
    },
    {
      "epoch": 0.4041927456451248,
      "grad_norm": 0.24882884949485204,
      "learning_rate": 3.4168367097362964e-07,
      "loss": 1.5794,
      "step": 2680
    },
    {
      "epoch": 0.4043435638337984,
      "grad_norm": 0.2861327829733997,
      "learning_rate": 3.415818382943644e-07,
      "loss": 1.5698,
      "step": 2681
    },
    {
      "epoch": 0.4044943820224719,
      "grad_norm": 0.3916787715235066,
      "learning_rate": 3.414799906655373e-07,
      "loss": 1.5366,
      "step": 2682
    },
    {
      "epoch": 0.4046452002111455,
      "grad_norm": 0.5960627175521959,
      "learning_rate": 3.4137812811001633e-07,
      "loss": 1.535,
      "step": 2683
    },
    {
      "epoch": 0.404796018399819,
      "grad_norm": 0.2562472635187039,
      "learning_rate": 3.4127625065067234e-07,
      "loss": 1.5125,
      "step": 2684
    },
    {
      "epoch": 0.40494683658849256,
      "grad_norm": 0.2513351595066356,
      "learning_rate": 3.4117435831037993e-07,
      "loss": 1.5257,
      "step": 2685
    },
    {
      "epoch": 0.40509765477716614,
      "grad_norm": 0.24845408181487894,
      "learning_rate": 3.4107245111201695e-07,
      "loss": 1.6515,
      "step": 2686
    },
    {
      "epoch": 0.40524847296583966,
      "grad_norm": 0.25030534842547747,
      "learning_rate": 3.4097052907846434e-07,
      "loss": 1.5621,
      "step": 2687
    },
    {
      "epoch": 0.40539929115451323,
      "grad_norm": 0.2512963303725947,
      "learning_rate": 3.408685922326067e-07,
      "loss": 1.5798,
      "step": 2688
    },
    {
      "epoch": 0.4055501093431868,
      "grad_norm": 0.2542188616465837,
      "learning_rate": 3.4076664059733186e-07,
      "loss": 1.5431,
      "step": 2689
    },
    {
      "epoch": 0.4057009275318603,
      "grad_norm": 0.2401795250475313,
      "learning_rate": 3.406646741955307e-07,
      "loss": 1.5495,
      "step": 2690
    },
    {
      "epoch": 0.4058517457205339,
      "grad_norm": 0.2544121534001211,
      "learning_rate": 3.4056269305009803e-07,
      "loss": 1.5652,
      "step": 2691
    },
    {
      "epoch": 0.40600256390920747,
      "grad_norm": 0.26262048054862935,
      "learning_rate": 3.4046069718393125e-07,
      "loss": 1.5826,
      "step": 2692
    },
    {
      "epoch": 0.406153382097881,
      "grad_norm": 0.27831235971342577,
      "learning_rate": 3.403586866199316e-07,
      "loss": 1.555,
      "step": 2693
    },
    {
      "epoch": 0.40630420028655456,
      "grad_norm": 0.2453047521498626,
      "learning_rate": 3.4025666138100327e-07,
      "loss": 1.547,
      "step": 2694
    },
    {
      "epoch": 0.40645501847522814,
      "grad_norm": 0.2547803857460511,
      "learning_rate": 3.4015462149005405e-07,
      "loss": 1.5881,
      "step": 2695
    },
    {
      "epoch": 0.40660583666390165,
      "grad_norm": 0.25239583782409786,
      "learning_rate": 3.400525669699948e-07,
      "loss": 1.5354,
      "step": 2696
    },
    {
      "epoch": 0.40675665485257523,
      "grad_norm": 0.2463906252270824,
      "learning_rate": 3.3995049784373966e-07,
      "loss": 1.5675,
      "step": 2697
    },
    {
      "epoch": 0.4069074730412488,
      "grad_norm": 0.2668245641763361,
      "learning_rate": 3.3984841413420616e-07,
      "loss": 1.5503,
      "step": 2698
    },
    {
      "epoch": 0.4070582912299223,
      "grad_norm": 0.40821576169654655,
      "learning_rate": 3.397463158643152e-07,
      "loss": 1.52,
      "step": 2699
    },
    {
      "epoch": 0.4072091094185959,
      "grad_norm": 0.2733214381159171,
      "learning_rate": 3.3964420305699064e-07,
      "loss": 1.6039,
      "step": 2700
    },
    {
      "epoch": 0.4073599276072694,
      "grad_norm": 0.24540087175714917,
      "learning_rate": 3.395420757351599e-07,
      "loss": 1.5331,
      "step": 2701
    },
    {
      "epoch": 0.407510745795943,
      "grad_norm": 0.303911791630233,
      "learning_rate": 3.394399339217534e-07,
      "loss": 1.5215,
      "step": 2702
    },
    {
      "epoch": 0.40766156398461656,
      "grad_norm": 0.27149720289102464,
      "learning_rate": 3.3933777763970507e-07,
      "loss": 1.7047,
      "step": 2703
    },
    {
      "epoch": 0.4078123821732901,
      "grad_norm": 0.2532605667347167,
      "learning_rate": 3.3923560691195194e-07,
      "loss": 1.5458,
      "step": 2704
    },
    {
      "epoch": 0.40796320036196365,
      "grad_norm": 1.1476192380139676,
      "learning_rate": 3.3913342176143435e-07,
      "loss": 1.5811,
      "step": 2705
    },
    {
      "epoch": 0.4081140185506372,
      "grad_norm": 0.2650873428391726,
      "learning_rate": 3.390312222110958e-07,
      "loss": 1.5058,
      "step": 2706
    },
    {
      "epoch": 0.40826483673931074,
      "grad_norm": 0.2507828693312355,
      "learning_rate": 3.3892900828388297e-07,
      "loss": 1.625,
      "step": 2707
    },
    {
      "epoch": 0.4084156549279843,
      "grad_norm": 0.2495042684876567,
      "learning_rate": 3.3882678000274607e-07,
      "loss": 1.6235,
      "step": 2708
    },
    {
      "epoch": 0.4085664731166579,
      "grad_norm": 0.26317188043167067,
      "learning_rate": 3.3872453739063815e-07,
      "loss": 1.6039,
      "step": 2709
    },
    {
      "epoch": 0.4087172913053314,
      "grad_norm": 0.2596342542294235,
      "learning_rate": 3.386222804705157e-07,
      "loss": 1.6036,
      "step": 2710
    },
    {
      "epoch": 0.408868109494005,
      "grad_norm": 0.25026437509997485,
      "learning_rate": 3.385200092653385e-07,
      "loss": 1.5756,
      "step": 2711
    },
    {
      "epoch": 0.40901892768267856,
      "grad_norm": 0.26226689742056464,
      "learning_rate": 3.384177237980693e-07,
      "loss": 1.5487,
      "step": 2712
    },
    {
      "epoch": 0.4091697458713521,
      "grad_norm": 0.25161489452840613,
      "learning_rate": 3.3831542409167406e-07,
      "loss": 1.5503,
      "step": 2713
    },
    {
      "epoch": 0.40932056406002565,
      "grad_norm": 0.2434400936495514,
      "learning_rate": 3.382131101691223e-07,
      "loss": 1.5178,
      "step": 2714
    },
    {
      "epoch": 0.40947138224869917,
      "grad_norm": 0.3428898089175938,
      "learning_rate": 3.381107820533863e-07,
      "loss": 1.5736,
      "step": 2715
    },
    {
      "epoch": 0.40962220043737274,
      "grad_norm": 0.24419776034880394,
      "learning_rate": 3.3800843976744165e-07,
      "loss": 1.5423,
      "step": 2716
    },
    {
      "epoch": 0.4097730186260463,
      "grad_norm": 0.2407398273910957,
      "learning_rate": 3.3790608333426733e-07,
      "loss": 1.5681,
      "step": 2717
    },
    {
      "epoch": 0.40992383681471983,
      "grad_norm": 0.23939499826381833,
      "learning_rate": 3.378037127768453e-07,
      "loss": 1.5486,
      "step": 2718
    },
    {
      "epoch": 0.4100746550033934,
      "grad_norm": 0.24805477172017687,
      "learning_rate": 3.377013281181607e-07,
      "loss": 1.5111,
      "step": 2719
    },
    {
      "epoch": 0.410225473192067,
      "grad_norm": 0.25290876011385005,
      "learning_rate": 3.375989293812018e-07,
      "loss": 1.5495,
      "step": 2720
    },
    {
      "epoch": 0.4103762913807405,
      "grad_norm": 0.24257394562603146,
      "learning_rate": 3.3749651658896023e-07,
      "loss": 1.6402,
      "step": 2721
    },
    {
      "epoch": 0.4105271095694141,
      "grad_norm": 0.2610073953080338,
      "learning_rate": 3.3739408976443065e-07,
      "loss": 1.6429,
      "step": 2722
    },
    {
      "epoch": 0.41067792775808765,
      "grad_norm": 0.24386260729457818,
      "learning_rate": 3.3729164893061066e-07,
      "loss": 1.6125,
      "step": 2723
    },
    {
      "epoch": 0.41082874594676116,
      "grad_norm": 0.3210020860293678,
      "learning_rate": 3.3718919411050144e-07,
      "loss": 1.5688,
      "step": 2724
    },
    {
      "epoch": 0.41097956413543474,
      "grad_norm": 0.2492960591940099,
      "learning_rate": 3.37086725327107e-07,
      "loss": 1.6307,
      "step": 2725
    },
    {
      "epoch": 0.4111303823241083,
      "grad_norm": 0.27238769894642695,
      "learning_rate": 3.369842426034345e-07,
      "loss": 1.5561,
      "step": 2726
    },
    {
      "epoch": 0.41128120051278183,
      "grad_norm": 0.2504464887219244,
      "learning_rate": 3.3688174596249446e-07,
      "loss": 1.5891,
      "step": 2727
    },
    {
      "epoch": 0.4114320187014554,
      "grad_norm": 0.29778856341510407,
      "learning_rate": 3.367792354273002e-07,
      "loss": 1.584,
      "step": 2728
    },
    {
      "epoch": 0.4115828368901289,
      "grad_norm": 0.27437651029692267,
      "learning_rate": 3.3667671102086835e-07,
      "loss": 1.5621,
      "step": 2729
    },
    {
      "epoch": 0.4117336550788025,
      "grad_norm": 0.32706072566989813,
      "learning_rate": 3.365741727662187e-07,
      "loss": 1.5984,
      "step": 2730
    },
    {
      "epoch": 0.41188447326747607,
      "grad_norm": 0.247227902267857,
      "learning_rate": 3.3647162068637393e-07,
      "loss": 1.6167,
      "step": 2731
    },
    {
      "epoch": 0.4120352914561496,
      "grad_norm": 0.2662345941585568,
      "learning_rate": 3.363690548043601e-07,
      "loss": 1.6178,
      "step": 2732
    },
    {
      "epoch": 0.41218610964482316,
      "grad_norm": 0.24907157756058096,
      "learning_rate": 3.362664751432063e-07,
      "loss": 1.5623,
      "step": 2733
    },
    {
      "epoch": 0.41233692783349674,
      "grad_norm": 0.24995333698556652,
      "learning_rate": 3.3616388172594446e-07,
      "loss": 1.5508,
      "step": 2734
    },
    {
      "epoch": 0.41248774602217025,
      "grad_norm": 0.26098406349777675,
      "learning_rate": 3.360612745756099e-07,
      "loss": 1.5691,
      "step": 2735
    },
    {
      "epoch": 0.41263856421084383,
      "grad_norm": 0.2492959279976695,
      "learning_rate": 3.359586537152409e-07,
      "loss": 1.5863,
      "step": 2736
    },
    {
      "epoch": 0.4127893823995174,
      "grad_norm": 0.26751413781066885,
      "learning_rate": 3.3585601916787886e-07,
      "loss": 1.5894,
      "step": 2737
    },
    {
      "epoch": 0.4129402005881909,
      "grad_norm": 0.2637667909873854,
      "learning_rate": 3.3575337095656815e-07,
      "loss": 1.5735,
      "step": 2738
    },
    {
      "epoch": 0.4130910187768645,
      "grad_norm": 0.25564659236591,
      "learning_rate": 3.356507091043563e-07,
      "loss": 1.6042,
      "step": 2739
    },
    {
      "epoch": 0.41324183696553807,
      "grad_norm": 0.2796082549809646,
      "learning_rate": 3.3554803363429403e-07,
      "loss": 1.5887,
      "step": 2740
    },
    {
      "epoch": 0.4133926551542116,
      "grad_norm": 0.272236777731565,
      "learning_rate": 3.3544534456943473e-07,
      "loss": 1.6221,
      "step": 2741
    },
    {
      "epoch": 0.41354347334288516,
      "grad_norm": 0.2543721016184445,
      "learning_rate": 3.353426419328352e-07,
      "loss": 1.5614,
      "step": 2742
    },
    {
      "epoch": 0.41369429153155873,
      "grad_norm": 0.25385962280711966,
      "learning_rate": 3.3523992574755525e-07,
      "loss": 1.5136,
      "step": 2743
    },
    {
      "epoch": 0.41384510972023225,
      "grad_norm": 0.32122229217521414,
      "learning_rate": 3.3513719603665764e-07,
      "loss": 1.5633,
      "step": 2744
    },
    {
      "epoch": 0.4139959279089058,
      "grad_norm": 0.246448545767713,
      "learning_rate": 3.350344528232081e-07,
      "loss": 1.537,
      "step": 2745
    },
    {
      "epoch": 0.41414674609757934,
      "grad_norm": 0.24637374083252297,
      "learning_rate": 3.3493169613027537e-07,
      "loss": 1.551,
      "step": 2746
    },
    {
      "epoch": 0.4142975642862529,
      "grad_norm": 0.3146802244636798,
      "learning_rate": 3.348289259809316e-07,
      "loss": 1.5986,
      "step": 2747
    },
    {
      "epoch": 0.4144483824749265,
      "grad_norm": 0.2462132778706303,
      "learning_rate": 3.347261423982515e-07,
      "loss": 1.5687,
      "step": 2748
    },
    {
      "epoch": 0.4145992006636,
      "grad_norm": 0.3066452786536837,
      "learning_rate": 3.34623345405313e-07,
      "loss": 1.6079,
      "step": 2749
    },
    {
      "epoch": 0.4147500188522736,
      "grad_norm": 0.2911543804614631,
      "learning_rate": 3.3452053502519705e-07,
      "loss": 1.5894,
      "step": 2750
    },
    {
      "epoch": 0.41490083704094716,
      "grad_norm": 0.3482933065137948,
      "learning_rate": 3.3441771128098757e-07,
      "loss": 1.6232,
      "step": 2751
    },
    {
      "epoch": 0.4150516552296207,
      "grad_norm": 0.23898678439185386,
      "learning_rate": 3.343148741957714e-07,
      "loss": 1.5574,
      "step": 2752
    },
    {
      "epoch": 0.41520247341829425,
      "grad_norm": 0.27797561961030265,
      "learning_rate": 3.3421202379263867e-07,
      "loss": 1.5671,
      "step": 2753
    },
    {
      "epoch": 0.4153532916069678,
      "grad_norm": 0.25358337703610595,
      "learning_rate": 3.34109160094682e-07,
      "loss": 1.5522,
      "step": 2754
    },
    {
      "epoch": 0.41550410979564134,
      "grad_norm": 0.2550694845202222,
      "learning_rate": 3.340062831249976e-07,
      "loss": 1.5342,
      "step": 2755
    },
    {
      "epoch": 0.4156549279843149,
      "grad_norm": 0.2597958803137637,
      "learning_rate": 3.339033929066841e-07,
      "loss": 1.5963,
      "step": 2756
    },
    {
      "epoch": 0.4158057461729885,
      "grad_norm": 0.2496316763184769,
      "learning_rate": 3.338004894628433e-07,
      "loss": 1.5589,
      "step": 2757
    },
    {
      "epoch": 0.415956564361662,
      "grad_norm": 0.25095372701614893,
      "learning_rate": 3.336975728165804e-07,
      "loss": 1.5429,
      "step": 2758
    },
    {
      "epoch": 0.4161073825503356,
      "grad_norm": 0.2631901524092377,
      "learning_rate": 3.3359464299100283e-07,
      "loss": 1.5541,
      "step": 2759
    },
    {
      "epoch": 0.4162582007390091,
      "grad_norm": 0.26130848981171917,
      "learning_rate": 3.3349170000922147e-07,
      "loss": 1.5923,
      "step": 2760
    },
    {
      "epoch": 0.4164090189276827,
      "grad_norm": 0.2865319671503223,
      "learning_rate": 3.333887438943499e-07,
      "loss": 1.6258,
      "step": 2761
    },
    {
      "epoch": 0.41655983711635625,
      "grad_norm": 0.25909940994325115,
      "learning_rate": 3.332857746695049e-07,
      "loss": 1.5499,
      "step": 2762
    },
    {
      "epoch": 0.41671065530502976,
      "grad_norm": 0.33967485157438465,
      "learning_rate": 3.3318279235780613e-07,
      "loss": 1.5917,
      "step": 2763
    },
    {
      "epoch": 0.41686147349370334,
      "grad_norm": 0.24378290445820855,
      "learning_rate": 3.3307979698237586e-07,
      "loss": 1.5501,
      "step": 2764
    },
    {
      "epoch": 0.4170122916823769,
      "grad_norm": 0.24693579990180703,
      "learning_rate": 3.329767885663398e-07,
      "loss": 1.5273,
      "step": 2765
    },
    {
      "epoch": 0.41716310987105043,
      "grad_norm": 0.2501417658905253,
      "learning_rate": 3.328737671328261e-07,
      "loss": 1.5227,
      "step": 2766
    },
    {
      "epoch": 0.417313928059724,
      "grad_norm": 0.2419502820018382,
      "learning_rate": 3.3277073270496635e-07,
      "loss": 1.5894,
      "step": 2767
    },
    {
      "epoch": 0.4174647462483976,
      "grad_norm": 0.2612808760168227,
      "learning_rate": 3.326676853058946e-07,
      "loss": 1.5314,
      "step": 2768
    },
    {
      "epoch": 0.4176155644370711,
      "grad_norm": 0.2540232188196327,
      "learning_rate": 3.3256462495874796e-07,
      "loss": 1.5732,
      "step": 2769
    },
    {
      "epoch": 0.41776638262574467,
      "grad_norm": 0.24457269312722849,
      "learning_rate": 3.324615516866666e-07,
      "loss": 1.5317,
      "step": 2770
    },
    {
      "epoch": 0.41791720081441824,
      "grad_norm": 0.24145564489233273,
      "learning_rate": 3.3235846551279346e-07,
      "loss": 1.5454,
      "step": 2771
    },
    {
      "epoch": 0.41806801900309176,
      "grad_norm": 0.31892115881563216,
      "learning_rate": 3.322553664602743e-07,
      "loss": 1.5604,
      "step": 2772
    },
    {
      "epoch": 0.41821883719176534,
      "grad_norm": 0.2586506860694086,
      "learning_rate": 3.32152254552258e-07,
      "loss": 1.5664,
      "step": 2773
    },
    {
      "epoch": 0.41836965538043885,
      "grad_norm": 0.3439546269177732,
      "learning_rate": 3.32049129811896e-07,
      "loss": 1.5469,
      "step": 2774
    },
    {
      "epoch": 0.41852047356911243,
      "grad_norm": 0.24418138452939914,
      "learning_rate": 3.3194599226234286e-07,
      "loss": 1.5578,
      "step": 2775
    },
    {
      "epoch": 0.418671291757786,
      "grad_norm": 0.2590844173586612,
      "learning_rate": 3.318428419267561e-07,
      "loss": 1.5369,
      "step": 2776
    },
    {
      "epoch": 0.4188221099464595,
      "grad_norm": 0.25552119759810754,
      "learning_rate": 3.3173967882829587e-07,
      "loss": 1.5683,
      "step": 2777
    },
    {
      "epoch": 0.4189729281351331,
      "grad_norm": 0.2669611498457822,
      "learning_rate": 3.3163650299012523e-07,
      "loss": 1.5806,
      "step": 2778
    },
    {
      "epoch": 0.41912374632380667,
      "grad_norm": 0.2611379003511046,
      "learning_rate": 3.315333144354103e-07,
      "loss": 1.6681,
      "step": 2779
    },
    {
      "epoch": 0.4192745645124802,
      "grad_norm": 0.352230203131426,
      "learning_rate": 3.3143011318731984e-07,
      "loss": 1.5714,
      "step": 2780
    },
    {
      "epoch": 0.41942538270115376,
      "grad_norm": 0.23926876122500249,
      "learning_rate": 3.313268992690255e-07,
      "loss": 1.62,
      "step": 2781
    },
    {
      "epoch": 0.41957620088982733,
      "grad_norm": 0.26209112358816156,
      "learning_rate": 3.312236727037019e-07,
      "loss": 1.6144,
      "step": 2782
    },
    {
      "epoch": 0.41972701907850085,
      "grad_norm": 0.3244389317643446,
      "learning_rate": 3.3112043351452635e-07,
      "loss": 1.6534,
      "step": 2783
    },
    {
      "epoch": 0.4198778372671744,
      "grad_norm": 0.24825052232823228,
      "learning_rate": 3.3101718172467914e-07,
      "loss": 1.522,
      "step": 2784
    },
    {
      "epoch": 0.420028655455848,
      "grad_norm": 0.24975060345118297,
      "learning_rate": 3.309139173573431e-07,
      "loss": 1.5199,
      "step": 2785
    },
    {
      "epoch": 0.4201794736445215,
      "grad_norm": 0.24013439238796144,
      "learning_rate": 3.3081064043570435e-07,
      "loss": 1.6011,
      "step": 2786
    },
    {
      "epoch": 0.4203302918331951,
      "grad_norm": 0.3894513948675421,
      "learning_rate": 3.3070735098295135e-07,
      "loss": 1.5331,
      "step": 2787
    },
    {
      "epoch": 0.42048111002186866,
      "grad_norm": 0.24346605286743947,
      "learning_rate": 3.306040490222756e-07,
      "loss": 1.5844,
      "step": 2788
    },
    {
      "epoch": 0.4206319282105422,
      "grad_norm": 0.2536043145275316,
      "learning_rate": 3.3050073457687165e-07,
      "loss": 1.6263,
      "step": 2789
    },
    {
      "epoch": 0.42078274639921576,
      "grad_norm": 0.2622097772722774,
      "learning_rate": 3.303974076699363e-07,
      "loss": 1.5541,
      "step": 2790
    },
    {
      "epoch": 0.4209335645878893,
      "grad_norm": 0.2597367981247587,
      "learning_rate": 3.3029406832466965e-07,
      "loss": 1.5695,
      "step": 2791
    },
    {
      "epoch": 0.42108438277656285,
      "grad_norm": 0.260190379204413,
      "learning_rate": 3.3019071656427426e-07,
      "loss": 1.6289,
      "step": 2792
    },
    {
      "epoch": 0.4212352009652364,
      "grad_norm": 0.25219201790085294,
      "learning_rate": 3.3008735241195573e-07,
      "loss": 1.5846,
      "step": 2793
    },
    {
      "epoch": 0.42138601915390994,
      "grad_norm": 0.2517974939357513,
      "learning_rate": 3.2998397589092224e-07,
      "loss": 1.5589,
      "step": 2794
    },
    {
      "epoch": 0.4215368373425835,
      "grad_norm": 0.27580484795282745,
      "learning_rate": 3.2988058702438486e-07,
      "loss": 1.5779,
      "step": 2795
    },
    {
      "epoch": 0.4216876555312571,
      "grad_norm": 0.2735825859157333,
      "learning_rate": 3.2977718583555744e-07,
      "loss": 1.5275,
      "step": 2796
    },
    {
      "epoch": 0.4218384737199306,
      "grad_norm": 0.27188477004475226,
      "learning_rate": 3.2967377234765635e-07,
      "loss": 1.6078,
      "step": 2797
    },
    {
      "epoch": 0.4219892919086042,
      "grad_norm": 0.2601800017524752,
      "learning_rate": 3.2957034658390125e-07,
      "loss": 1.6309,
      "step": 2798
    },
    {
      "epoch": 0.42214011009727775,
      "grad_norm": 0.24986428985487438,
      "learning_rate": 3.2946690856751404e-07,
      "loss": 1.5696,
      "step": 2799
    },
    {
      "epoch": 0.4222909282859513,
      "grad_norm": 0.2586438246567823,
      "learning_rate": 3.293634583217195e-07,
      "loss": 1.5653,
      "step": 2800
    },
    {
      "epoch": 0.42244174647462485,
      "grad_norm": 0.26875412070114774,
      "learning_rate": 3.2925999586974545e-07,
      "loss": 1.5808,
      "step": 2801
    },
    {
      "epoch": 0.4225925646632984,
      "grad_norm": 0.2459104246750409,
      "learning_rate": 3.2915652123482215e-07,
      "loss": 1.5677,
      "step": 2802
    },
    {
      "epoch": 0.42274338285197194,
      "grad_norm": 0.2839961716031081,
      "learning_rate": 3.2905303444018253e-07,
      "loss": 1.563,
      "step": 2803
    },
    {
      "epoch": 0.4228942010406455,
      "grad_norm": 0.24736716745677786,
      "learning_rate": 3.2894953550906246e-07,
      "loss": 1.523,
      "step": 2804
    },
    {
      "epoch": 0.42304501922931903,
      "grad_norm": 0.27418143350308405,
      "learning_rate": 3.288460244647005e-07,
      "loss": 1.5728,
      "step": 2805
    },
    {
      "epoch": 0.4231958374179926,
      "grad_norm": 0.3195441770865284,
      "learning_rate": 3.2874250133033786e-07,
      "loss": 1.6053,
      "step": 2806
    },
    {
      "epoch": 0.4233466556066662,
      "grad_norm": 0.24631535702514804,
      "learning_rate": 3.286389661292186e-07,
      "loss": 1.5078,
      "step": 2807
    },
    {
      "epoch": 0.4234974737953397,
      "grad_norm": 0.25668273674967906,
      "learning_rate": 3.285354188845892e-07,
      "loss": 1.6132,
      "step": 2808
    },
    {
      "epoch": 0.42364829198401327,
      "grad_norm": 0.2507520791825448,
      "learning_rate": 3.284318596196992e-07,
      "loss": 1.5598,
      "step": 2809
    },
    {
      "epoch": 0.42379911017268684,
      "grad_norm": 0.26571287124523163,
      "learning_rate": 3.2832828835780065e-07,
      "loss": 1.5631,
      "step": 2810
    },
    {
      "epoch": 0.42394992836136036,
      "grad_norm": 0.2439539718460567,
      "learning_rate": 3.2822470512214816e-07,
      "loss": 1.5626,
      "step": 2811
    },
    {
      "epoch": 0.42410074655003394,
      "grad_norm": 0.2674550525834154,
      "learning_rate": 3.281211099359994e-07,
      "loss": 1.5791,
      "step": 2812
    },
    {
      "epoch": 0.4242515647387075,
      "grad_norm": 0.24557218187885993,
      "learning_rate": 3.280175028226143e-07,
      "loss": 1.554,
      "step": 2813
    },
    {
      "epoch": 0.42440238292738103,
      "grad_norm": 0.2564709265749323,
      "learning_rate": 3.2791388380525583e-07,
      "loss": 1.5749,
      "step": 2814
    },
    {
      "epoch": 0.4245532011160546,
      "grad_norm": 0.2471382396594902,
      "learning_rate": 3.278102529071894e-07,
      "loss": 1.6417,
      "step": 2815
    },
    {
      "epoch": 0.4247040193047282,
      "grad_norm": 0.28361301842175707,
      "learning_rate": 3.2770661015168327e-07,
      "loss": 1.5692,
      "step": 2816
    },
    {
      "epoch": 0.4248548374934017,
      "grad_norm": 0.24649131145090075,
      "learning_rate": 3.2760295556200813e-07,
      "loss": 1.5649,
      "step": 2817
    },
    {
      "epoch": 0.42500565568207527,
      "grad_norm": 0.25011966391967305,
      "learning_rate": 3.274992891614375e-07,
      "loss": 1.4936,
      "step": 2818
    },
    {
      "epoch": 0.4251564738707488,
      "grad_norm": 0.26769277225623017,
      "learning_rate": 3.2739561097324746e-07,
      "loss": 1.5286,
      "step": 2819
    },
    {
      "epoch": 0.42530729205942236,
      "grad_norm": 0.25553634024998556,
      "learning_rate": 3.2729192102071683e-07,
      "loss": 1.5756,
      "step": 2820
    },
    {
      "epoch": 0.42545811024809593,
      "grad_norm": 0.26612513346681166,
      "learning_rate": 3.2718821932712704e-07,
      "loss": 1.5705,
      "step": 2821
    },
    {
      "epoch": 0.42560892843676945,
      "grad_norm": 0.270197771738574,
      "learning_rate": 3.270845059157622e-07,
      "loss": 1.6422,
      "step": 2822
    },
    {
      "epoch": 0.425759746625443,
      "grad_norm": 0.30734421602780476,
      "learning_rate": 3.269807808099088e-07,
      "loss": 1.6486,
      "step": 2823
    },
    {
      "epoch": 0.4259105648141166,
      "grad_norm": 0.2514541593491989,
      "learning_rate": 3.268770440328563e-07,
      "loss": 1.5418,
      "step": 2824
    },
    {
      "epoch": 0.4260613830027901,
      "grad_norm": 0.2614494218667816,
      "learning_rate": 3.2677329560789664e-07,
      "loss": 1.5701,
      "step": 2825
    },
    {
      "epoch": 0.4262122011914637,
      "grad_norm": 0.2511602678497061,
      "learning_rate": 3.266695355583242e-07,
      "loss": 1.5158,
      "step": 2826
    },
    {
      "epoch": 0.42636301938013726,
      "grad_norm": 0.26400297098564646,
      "learning_rate": 3.265657639074363e-07,
      "loss": 1.6608,
      "step": 2827
    },
    {
      "epoch": 0.4265138375688108,
      "grad_norm": 0.2564658434956295,
      "learning_rate": 3.264619806785327e-07,
      "loss": 1.5434,
      "step": 2828
    },
    {
      "epoch": 0.42666465575748436,
      "grad_norm": 0.38207198468778597,
      "learning_rate": 3.2635818589491567e-07,
      "loss": 1.5657,
      "step": 2829
    },
    {
      "epoch": 0.42681547394615793,
      "grad_norm": 0.27509380119835014,
      "learning_rate": 3.262543795798902e-07,
      "loss": 1.5504,
      "step": 2830
    },
    {
      "epoch": 0.42696629213483145,
      "grad_norm": 0.25133984016735206,
      "learning_rate": 3.2615056175676376e-07,
      "loss": 1.5923,
      "step": 2831
    },
    {
      "epoch": 0.427117110323505,
      "grad_norm": 0.244357207006092,
      "learning_rate": 3.2604673244884656e-07,
      "loss": 1.5887,
      "step": 2832
    },
    {
      "epoch": 0.4272679285121786,
      "grad_norm": 0.29225697760141345,
      "learning_rate": 3.259428916794513e-07,
      "loss": 1.561,
      "step": 2833
    },
    {
      "epoch": 0.4274187467008521,
      "grad_norm": 0.24786007772500737,
      "learning_rate": 3.2583903947189324e-07,
      "loss": 1.5648,
      "step": 2834
    },
    {
      "epoch": 0.4275695648895257,
      "grad_norm": 0.27825183380842167,
      "learning_rate": 3.2573517584949016e-07,
      "loss": 1.5252,
      "step": 2835
    },
    {
      "epoch": 0.4277203830781992,
      "grad_norm": 0.24809497916679388,
      "learning_rate": 3.2563130083556263e-07,
      "loss": 1.5027,
      "step": 2836
    },
    {
      "epoch": 0.4278712012668728,
      "grad_norm": 0.2597344556034982,
      "learning_rate": 3.2552741445343343e-07,
      "loss": 1.5662,
      "step": 2837
    },
    {
      "epoch": 0.42802201945554635,
      "grad_norm": 0.2567179160633774,
      "learning_rate": 3.254235167264282e-07,
      "loss": 1.5976,
      "step": 2838
    },
    {
      "epoch": 0.4281728376442199,
      "grad_norm": 0.24920655921464463,
      "learning_rate": 3.25319607677875e-07,
      "loss": 1.571,
      "step": 2839
    },
    {
      "epoch": 0.42832365583289345,
      "grad_norm": 0.27293874543904306,
      "learning_rate": 3.2521568733110435e-07,
      "loss": 1.6008,
      "step": 2840
    },
    {
      "epoch": 0.428474474021567,
      "grad_norm": 0.2878965239814466,
      "learning_rate": 3.251117557094495e-07,
      "loss": 1.5456,
      "step": 2841
    },
    {
      "epoch": 0.42862529221024054,
      "grad_norm": 0.24775200312433204,
      "learning_rate": 3.2500781283624606e-07,
      "loss": 1.5894,
      "step": 2842
    },
    {
      "epoch": 0.4287761103989141,
      "grad_norm": 0.2615997014805728,
      "learning_rate": 3.249038587348323e-07,
      "loss": 1.5442,
      "step": 2843
    },
    {
      "epoch": 0.4289269285875877,
      "grad_norm": 0.2812310283534168,
      "learning_rate": 3.247998934285488e-07,
      "loss": 1.6119,
      "step": 2844
    },
    {
      "epoch": 0.4290777467762612,
      "grad_norm": 0.2640209190131911,
      "learning_rate": 3.246959169407389e-07,
      "loss": 1.5895,
      "step": 2845
    },
    {
      "epoch": 0.4292285649649348,
      "grad_norm": 0.2894966145146055,
      "learning_rate": 3.245919292947484e-07,
      "loss": 1.5367,
      "step": 2846
    },
    {
      "epoch": 0.42937938315360835,
      "grad_norm": 0.2514511542793688,
      "learning_rate": 3.2448793051392546e-07,
      "loss": 1.5186,
      "step": 2847
    },
    {
      "epoch": 0.42953020134228187,
      "grad_norm": 0.2515606117685996,
      "learning_rate": 3.243839206216209e-07,
      "loss": 1.5835,
      "step": 2848
    },
    {
      "epoch": 0.42968101953095544,
      "grad_norm": 0.2393427989651705,
      "learning_rate": 3.2427989964118784e-07,
      "loss": 1.5502,
      "step": 2849
    },
    {
      "epoch": 0.42983183771962896,
      "grad_norm": 0.2793399959806875,
      "learning_rate": 3.2417586759598225e-07,
      "loss": 1.5552,
      "step": 2850
    },
    {
      "epoch": 0.42998265590830254,
      "grad_norm": 0.25010490471432933,
      "learning_rate": 3.2407182450936217e-07,
      "loss": 1.5586,
      "step": 2851
    },
    {
      "epoch": 0.4301334740969761,
      "grad_norm": 0.251943279627523,
      "learning_rate": 3.2396777040468833e-07,
      "loss": 1.5344,
      "step": 2852
    },
    {
      "epoch": 0.4302842922856496,
      "grad_norm": 0.2605855624176244,
      "learning_rate": 3.2386370530532397e-07,
      "loss": 1.5077,
      "step": 2853
    },
    {
      "epoch": 0.4304351104743232,
      "grad_norm": 0.2534032198280918,
      "learning_rate": 3.2375962923463476e-07,
      "loss": 1.5534,
      "step": 2854
    },
    {
      "epoch": 0.4305859286629968,
      "grad_norm": 0.26276413466059373,
      "learning_rate": 3.236555422159887e-07,
      "loss": 1.5436,
      "step": 2855
    },
    {
      "epoch": 0.4307367468516703,
      "grad_norm": 0.24882989004331033,
      "learning_rate": 3.2355144427275637e-07,
      "loss": 1.5596,
      "step": 2856
    },
    {
      "epoch": 0.43088756504034387,
      "grad_norm": 0.2842043516544378,
      "learning_rate": 3.2344733542831086e-07,
      "loss": 1.5613,
      "step": 2857
    },
    {
      "epoch": 0.43103838322901744,
      "grad_norm": 0.23323082429628905,
      "learning_rate": 3.233432157060276e-07,
      "loss": 1.6159,
      "step": 2858
    },
    {
      "epoch": 0.43118920141769096,
      "grad_norm": 0.36941519147313756,
      "learning_rate": 3.232390851292845e-07,
      "loss": 1.5453,
      "step": 2859
    },
    {
      "epoch": 0.43134001960636453,
      "grad_norm": 0.23799456270118677,
      "learning_rate": 3.231349437214619e-07,
      "loss": 1.5456,
      "step": 2860
    },
    {
      "epoch": 0.4314908377950381,
      "grad_norm": 0.2550429890826443,
      "learning_rate": 3.2303079150594264e-07,
      "loss": 1.5765,
      "step": 2861
    },
    {
      "epoch": 0.4316416559837116,
      "grad_norm": 0.25440445508837595,
      "learning_rate": 3.229266285061118e-07,
      "loss": 1.6726,
      "step": 2862
    },
    {
      "epoch": 0.4317924741723852,
      "grad_norm": 0.2540189384768522,
      "learning_rate": 3.228224547453571e-07,
      "loss": 1.5742,
      "step": 2863
    },
    {
      "epoch": 0.4319432923610587,
      "grad_norm": 0.25509213382412893,
      "learning_rate": 3.227182702470686e-07,
      "loss": 1.638,
      "step": 2864
    },
    {
      "epoch": 0.4320941105497323,
      "grad_norm": 0.25794465385284837,
      "learning_rate": 3.2261407503463874e-07,
      "loss": 1.609,
      "step": 2865
    },
    {
      "epoch": 0.43224492873840586,
      "grad_norm": 0.26403881152382164,
      "learning_rate": 3.225098691314623e-07,
      "loss": 1.6337,
      "step": 2866
    },
    {
      "epoch": 0.4323957469270794,
      "grad_norm": 0.24986562370671758,
      "learning_rate": 3.224056525609366e-07,
      "loss": 1.5586,
      "step": 2867
    },
    {
      "epoch": 0.43254656511575296,
      "grad_norm": 0.24697766775196395,
      "learning_rate": 3.223014253464614e-07,
      "loss": 1.5504,
      "step": 2868
    },
    {
      "epoch": 0.43269738330442653,
      "grad_norm": 0.2550985722985474,
      "learning_rate": 3.221971875114385e-07,
      "loss": 1.5356,
      "step": 2869
    },
    {
      "epoch": 0.43284820149310005,
      "grad_norm": 0.2471461523400474,
      "learning_rate": 3.2209293907927256e-07,
      "loss": 1.5304,
      "step": 2870
    },
    {
      "epoch": 0.4329990196817736,
      "grad_norm": 0.30486015466742766,
      "learning_rate": 3.219886800733703e-07,
      "loss": 1.5771,
      "step": 2871
    },
    {
      "epoch": 0.4331498378704472,
      "grad_norm": 0.24892059657585422,
      "learning_rate": 3.2188441051714086e-07,
      "loss": 1.6376,
      "step": 2872
    },
    {
      "epoch": 0.4333006560591207,
      "grad_norm": 0.25228093749319536,
      "learning_rate": 3.217801304339958e-07,
      "loss": 1.5088,
      "step": 2873
    },
    {
      "epoch": 0.4334514742477943,
      "grad_norm": 0.24124651139886405,
      "learning_rate": 3.216758398473492e-07,
      "loss": 1.5354,
      "step": 2874
    },
    {
      "epoch": 0.43360229243646786,
      "grad_norm": 0.24938492951014538,
      "learning_rate": 3.2157153878061705e-07,
      "loss": 1.5817,
      "step": 2875
    },
    {
      "epoch": 0.4337531106251414,
      "grad_norm": 0.25757749869874946,
      "learning_rate": 3.2146722725721826e-07,
      "loss": 1.5755,
      "step": 2876
    },
    {
      "epoch": 0.43390392881381495,
      "grad_norm": 0.25676529276672144,
      "learning_rate": 3.213629053005736e-07,
      "loss": 1.6694,
      "step": 2877
    },
    {
      "epoch": 0.4340547470024885,
      "grad_norm": 0.25196048770398494,
      "learning_rate": 3.212585729341064e-07,
      "loss": 1.6174,
      "step": 2878
    },
    {
      "epoch": 0.43420556519116205,
      "grad_norm": 0.2498727193813626,
      "learning_rate": 3.211542301812425e-07,
      "loss": 1.5537,
      "step": 2879
    },
    {
      "epoch": 0.4343563833798356,
      "grad_norm": 0.2477314087337444,
      "learning_rate": 3.2104987706540975e-07,
      "loss": 1.5551,
      "step": 2880
    },
    {
      "epoch": 0.43450720156850914,
      "grad_norm": 0.2767502201857406,
      "learning_rate": 3.209455136100385e-07,
      "loss": 1.5546,
      "step": 2881
    },
    {
      "epoch": 0.4346580197571827,
      "grad_norm": 0.2548227677714052,
      "learning_rate": 3.2084113983856125e-07,
      "loss": 1.5466,
      "step": 2882
    },
    {
      "epoch": 0.4348088379458563,
      "grad_norm": 0.24014031121639437,
      "learning_rate": 3.2073675577441314e-07,
      "loss": 1.5498,
      "step": 2883
    },
    {
      "epoch": 0.4349596561345298,
      "grad_norm": 0.24294953710051626,
      "learning_rate": 3.206323614410314e-07,
      "loss": 1.548,
      "step": 2884
    },
    {
      "epoch": 0.4351104743232034,
      "grad_norm": 0.26743355263189356,
      "learning_rate": 3.205279568618554e-07,
      "loss": 1.5803,
      "step": 2885
    },
    {
      "epoch": 0.43526129251187695,
      "grad_norm": 0.2663085951111445,
      "learning_rate": 3.2042354206032727e-07,
      "loss": 1.5586,
      "step": 2886
    },
    {
      "epoch": 0.43541211070055047,
      "grad_norm": 0.2718118661493261,
      "learning_rate": 3.2031911705989113e-07,
      "loss": 1.5262,
      "step": 2887
    },
    {
      "epoch": 0.43556292888922404,
      "grad_norm": 0.2765415943315982,
      "learning_rate": 3.2021468188399325e-07,
      "loss": 1.5542,
      "step": 2888
    },
    {
      "epoch": 0.4357137470778976,
      "grad_norm": 0.25151574286126827,
      "learning_rate": 3.2011023655608256e-07,
      "loss": 1.5123,
      "step": 2889
    },
    {
      "epoch": 0.43586456526657114,
      "grad_norm": 0.2569228721930573,
      "learning_rate": 3.2000578109961003e-07,
      "loss": 1.5425,
      "step": 2890
    },
    {
      "epoch": 0.4360153834552447,
      "grad_norm": 0.2581041939483696,
      "learning_rate": 3.199013155380289e-07,
      "loss": 1.5974,
      "step": 2891
    },
    {
      "epoch": 0.4361662016439183,
      "grad_norm": 0.24858529905108517,
      "learning_rate": 3.197968398947947e-07,
      "loss": 1.5472,
      "step": 2892
    },
    {
      "epoch": 0.4363170198325918,
      "grad_norm": 0.2655668765193959,
      "learning_rate": 3.1969235419336533e-07,
      "loss": 1.6225,
      "step": 2893
    },
    {
      "epoch": 0.4364678380212654,
      "grad_norm": 0.25538155411435326,
      "learning_rate": 3.1958785845720094e-07,
      "loss": 1.4932,
      "step": 2894
    },
    {
      "epoch": 0.4366186562099389,
      "grad_norm": 0.2513728933246105,
      "learning_rate": 3.194833527097637e-07,
      "loss": 1.6067,
      "step": 2895
    },
    {
      "epoch": 0.43676947439861247,
      "grad_norm": 0.2688528228720648,
      "learning_rate": 3.1937883697451835e-07,
      "loss": 1.5867,
      "step": 2896
    },
    {
      "epoch": 0.43692029258728604,
      "grad_norm": 0.27820117449902454,
      "learning_rate": 3.1927431127493164e-07,
      "loss": 1.5504,
      "step": 2897
    },
    {
      "epoch": 0.43707111077595956,
      "grad_norm": 0.2587225093120209,
      "learning_rate": 3.1916977563447257e-07,
      "loss": 1.5992,
      "step": 2898
    },
    {
      "epoch": 0.43722192896463313,
      "grad_norm": 0.25957784875946716,
      "learning_rate": 3.1906523007661256e-07,
      "loss": 1.4921,
      "step": 2899
    },
    {
      "epoch": 0.4373727471533067,
      "grad_norm": 0.25242465566484146,
      "learning_rate": 3.189606746248251e-07,
      "loss": 1.5765,
      "step": 2900
    },
    {
      "epoch": 0.4375235653419802,
      "grad_norm": 0.2936725023072022,
      "learning_rate": 3.1885610930258585e-07,
      "loss": 1.5895,
      "step": 2901
    },
    {
      "epoch": 0.4376743835306538,
      "grad_norm": 0.24296660652410004,
      "learning_rate": 3.187515341333729e-07,
      "loss": 1.6201,
      "step": 2902
    },
    {
      "epoch": 0.4378252017193274,
      "grad_norm": 0.2660340565961371,
      "learning_rate": 3.1864694914066624e-07,
      "loss": 1.4993,
      "step": 2903
    },
    {
      "epoch": 0.4379760199080009,
      "grad_norm": 0.2678230768413757,
      "learning_rate": 3.185423543479484e-07,
      "loss": 1.6239,
      "step": 2904
    },
    {
      "epoch": 0.43812683809667446,
      "grad_norm": 0.23864790824148918,
      "learning_rate": 3.18437749778704e-07,
      "loss": 1.5226,
      "step": 2905
    },
    {
      "epoch": 0.43827765628534804,
      "grad_norm": 0.24579629971463482,
      "learning_rate": 3.183331354564196e-07,
      "loss": 1.5225,
      "step": 2906
    },
    {
      "epoch": 0.43842847447402156,
      "grad_norm": 0.3521358692586838,
      "learning_rate": 3.182285114045844e-07,
      "loss": 1.576,
      "step": 2907
    },
    {
      "epoch": 0.43857929266269513,
      "grad_norm": 0.2590866917875334,
      "learning_rate": 3.181238776466894e-07,
      "loss": 1.5539,
      "step": 2908
    },
    {
      "epoch": 0.43873011085136865,
      "grad_norm": 0.25550321049136315,
      "learning_rate": 3.180192342062279e-07,
      "loss": 1.5528,
      "step": 2909
    },
    {
      "epoch": 0.4388809290400422,
      "grad_norm": 0.24770792136612937,
      "learning_rate": 3.1791458110669557e-07,
      "loss": 1.5544,
      "step": 2910
    },
    {
      "epoch": 0.4390317472287158,
      "grad_norm": 0.27145834138714725,
      "learning_rate": 3.178099183715899e-07,
      "loss": 1.6244,
      "step": 2911
    },
    {
      "epoch": 0.4391825654173893,
      "grad_norm": 0.7306377787337551,
      "learning_rate": 3.177052460244108e-07,
      "loss": 1.5126,
      "step": 2912
    },
    {
      "epoch": 0.4393333836060629,
      "grad_norm": 0.2423739274479909,
      "learning_rate": 3.176005640886603e-07,
      "loss": 1.5814,
      "step": 2913
    },
    {
      "epoch": 0.43948420179473646,
      "grad_norm": 0.27344887574578947,
      "learning_rate": 3.174958725878424e-07,
      "loss": 1.5617,
      "step": 2914
    },
    {
      "epoch": 0.43963501998341,
      "grad_norm": 0.2719054718336235,
      "learning_rate": 3.1739117154546357e-07,
      "loss": 1.6183,
      "step": 2915
    },
    {
      "epoch": 0.43978583817208355,
      "grad_norm": 0.2519172488059112,
      "learning_rate": 3.1728646098503216e-07,
      "loss": 1.5871,
      "step": 2916
    },
    {
      "epoch": 0.43993665636075713,
      "grad_norm": 0.25133162404568815,
      "learning_rate": 3.171817409300587e-07,
      "loss": 1.5503,
      "step": 2917
    },
    {
      "epoch": 0.44008747454943065,
      "grad_norm": 0.2792435671678952,
      "learning_rate": 3.17077011404056e-07,
      "loss": 1.5728,
      "step": 2918
    },
    {
      "epoch": 0.4402382927381042,
      "grad_norm": 0.24427337476009683,
      "learning_rate": 3.169722724305387e-07,
      "loss": 1.6415,
      "step": 2919
    },
    {
      "epoch": 0.4403891109267778,
      "grad_norm": 0.3465354844042566,
      "learning_rate": 3.16867524033024e-07,
      "loss": 1.5484,
      "step": 2920
    },
    {
      "epoch": 0.4405399291154513,
      "grad_norm": 0.2586043215420097,
      "learning_rate": 3.1676276623503075e-07,
      "loss": 1.603,
      "step": 2921
    },
    {
      "epoch": 0.4406907473041249,
      "grad_norm": 0.24859283143903943,
      "learning_rate": 3.1665799906008023e-07,
      "loss": 1.5771,
      "step": 2922
    },
    {
      "epoch": 0.4408415654927984,
      "grad_norm": 0.24570719218752252,
      "learning_rate": 3.1655322253169576e-07,
      "loss": 1.4971,
      "step": 2923
    },
    {
      "epoch": 0.440992383681472,
      "grad_norm": 0.31073154586652807,
      "learning_rate": 3.1644843667340257e-07,
      "loss": 1.545,
      "step": 2924
    },
    {
      "epoch": 0.44114320187014555,
      "grad_norm": 0.25075904681549044,
      "learning_rate": 3.163436415087283e-07,
      "loss": 1.5834,
      "step": 2925
    },
    {
      "epoch": 0.44129402005881907,
      "grad_norm": 0.25314906080943284,
      "learning_rate": 3.1623883706120247e-07,
      "loss": 1.5579,
      "step": 2926
    },
    {
      "epoch": 0.44144483824749264,
      "grad_norm": 0.2360378844836213,
      "learning_rate": 3.161340233543567e-07,
      "loss": 1.5135,
      "step": 2927
    },
    {
      "epoch": 0.4415956564361662,
      "grad_norm": 0.24292165414084746,
      "learning_rate": 3.1602920041172473e-07,
      "loss": 1.5137,
      "step": 2928
    },
    {
      "epoch": 0.44174647462483974,
      "grad_norm": 0.2997251514094543,
      "learning_rate": 3.159243682568423e-07,
      "loss": 1.6018,
      "step": 2929
    },
    {
      "epoch": 0.4418972928135133,
      "grad_norm": 0.3497763748048155,
      "learning_rate": 3.158195269132474e-07,
      "loss": 1.5502,
      "step": 2930
    },
    {
      "epoch": 0.4420481110021869,
      "grad_norm": 0.27338608798627173,
      "learning_rate": 3.1571467640447997e-07,
      "loss": 1.6476,
      "step": 2931
    },
    {
      "epoch": 0.4421989291908604,
      "grad_norm": 0.26887073330294453,
      "learning_rate": 3.156098167540818e-07,
      "loss": 1.549,
      "step": 2932
    },
    {
      "epoch": 0.442349747379534,
      "grad_norm": 0.2559339493115962,
      "learning_rate": 3.155049479855973e-07,
      "loss": 1.5785,
      "step": 2933
    },
    {
      "epoch": 0.44250056556820755,
      "grad_norm": 0.24419500081836334,
      "learning_rate": 3.1540007012257215e-07,
      "loss": 1.5312,
      "step": 2934
    },
    {
      "epoch": 0.44265138375688107,
      "grad_norm": 0.3446463400081573,
      "learning_rate": 3.1529518318855474e-07,
      "loss": 1.5488,
      "step": 2935
    },
    {
      "epoch": 0.44280220194555464,
      "grad_norm": 0.250792786442541,
      "learning_rate": 3.1519028720709515e-07,
      "loss": 1.5238,
      "step": 2936
    },
    {
      "epoch": 0.4429530201342282,
      "grad_norm": 0.25277066988345265,
      "learning_rate": 3.1508538220174565e-07,
      "loss": 1.5322,
      "step": 2937
    },
    {
      "epoch": 0.44310383832290173,
      "grad_norm": 0.2574346793763021,
      "learning_rate": 3.149804681960605e-07,
      "loss": 1.5623,
      "step": 2938
    },
    {
      "epoch": 0.4432546565115753,
      "grad_norm": 0.3390501313428078,
      "learning_rate": 3.148755452135957e-07,
      "loss": 1.5602,
      "step": 2939
    },
    {
      "epoch": 0.4434054747002488,
      "grad_norm": 0.25222635073802524,
      "learning_rate": 3.147706132779098e-07,
      "loss": 1.5604,
      "step": 2940
    },
    {
      "epoch": 0.4435562928889224,
      "grad_norm": 0.2376778230089622,
      "learning_rate": 3.1466567241256304e-07,
      "loss": 1.4962,
      "step": 2941
    },
    {
      "epoch": 0.443707111077596,
      "grad_norm": 0.2703135489064138,
      "learning_rate": 3.145607226411175e-07,
      "loss": 1.6,
      "step": 2942
    },
    {
      "epoch": 0.4438579292662695,
      "grad_norm": 0.24834680304175888,
      "learning_rate": 3.144557639871377e-07,
      "loss": 1.5691,
      "step": 2943
    },
    {
      "epoch": 0.44400874745494306,
      "grad_norm": 0.24414712441004643,
      "learning_rate": 3.143507964741897e-07,
      "loss": 1.5583,
      "step": 2944
    },
    {
      "epoch": 0.44415956564361664,
      "grad_norm": 0.2634739776520894,
      "learning_rate": 3.1424582012584197e-07,
      "loss": 1.5459,
      "step": 2945
    },
    {
      "epoch": 0.44431038383229016,
      "grad_norm": 0.2486655684747634,
      "learning_rate": 3.141408349656646e-07,
      "loss": 1.5303,
      "step": 2946
    },
    {
      "epoch": 0.44446120202096373,
      "grad_norm": 0.2552836521878257,
      "learning_rate": 3.1403584101722984e-07,
      "loss": 1.5216,
      "step": 2947
    },
    {
      "epoch": 0.4446120202096373,
      "grad_norm": 0.24430062763377616,
      "learning_rate": 3.13930838304112e-07,
      "loss": 1.6221,
      "step": 2948
    },
    {
      "epoch": 0.4447628383983108,
      "grad_norm": 0.2547464676737134,
      "learning_rate": 3.1382582684988713e-07,
      "loss": 1.5393,
      "step": 2949
    },
    {
      "epoch": 0.4449136565869844,
      "grad_norm": 0.2582525317132371,
      "learning_rate": 3.1372080667813346e-07,
      "loss": 1.5336,
      "step": 2950
    },
    {
      "epoch": 0.44506447477565797,
      "grad_norm": 0.25259837083207576,
      "learning_rate": 3.136157778124309e-07,
      "loss": 1.5129,
      "step": 2951
    },
    {
      "epoch": 0.4452152929643315,
      "grad_norm": 0.24497532992454443,
      "learning_rate": 3.135107402763617e-07,
      "loss": 1.5764,
      "step": 2952
    },
    {
      "epoch": 0.44536611115300506,
      "grad_norm": 0.2684931409910681,
      "learning_rate": 3.134056940935097e-07,
      "loss": 1.6112,
      "step": 2953
    },
    {
      "epoch": 0.4455169293416786,
      "grad_norm": 0.2772107291494322,
      "learning_rate": 3.133006392874609e-07,
      "loss": 1.4843,
      "step": 2954
    },
    {
      "epoch": 0.44566774753035215,
      "grad_norm": 0.2571460569029784,
      "learning_rate": 3.1319557588180313e-07,
      "loss": 1.558,
      "step": 2955
    },
    {
      "epoch": 0.44581856571902573,
      "grad_norm": 0.2560514367861158,
      "learning_rate": 3.130905039001262e-07,
      "loss": 1.5363,
      "step": 2956
    },
    {
      "epoch": 0.44596938390769925,
      "grad_norm": 0.24476015297538142,
      "learning_rate": 3.129854233660218e-07,
      "loss": 1.579,
      "step": 2957
    },
    {
      "epoch": 0.4461202020963728,
      "grad_norm": 0.31510599110357745,
      "learning_rate": 3.1288033430308356e-07,
      "loss": 1.5831,
      "step": 2958
    },
    {
      "epoch": 0.4462710202850464,
      "grad_norm": 0.3357344820178751,
      "learning_rate": 3.1277523673490723e-07,
      "loss": 1.5095,
      "step": 2959
    },
    {
      "epoch": 0.4464218384737199,
      "grad_norm": 0.2502751163810743,
      "learning_rate": 3.1267013068509e-07,
      "loss": 1.6416,
      "step": 2960
    },
    {
      "epoch": 0.4465726566623935,
      "grad_norm": 0.23393203221738712,
      "learning_rate": 3.125650161772314e-07,
      "loss": 1.5145,
      "step": 2961
    },
    {
      "epoch": 0.44672347485106706,
      "grad_norm": 0.24746422788563224,
      "learning_rate": 3.1245989323493265e-07,
      "loss": 1.5863,
      "step": 2962
    },
    {
      "epoch": 0.4468742930397406,
      "grad_norm": 0.2525927927400594,
      "learning_rate": 3.123547618817969e-07,
      "loss": 1.5796,
      "step": 2963
    },
    {
      "epoch": 0.44702511122841415,
      "grad_norm": 0.2774687244669163,
      "learning_rate": 3.1224962214142926e-07,
      "loss": 1.5681,
      "step": 2964
    },
    {
      "epoch": 0.4471759294170877,
      "grad_norm": 0.2826685061590997,
      "learning_rate": 3.1214447403743654e-07,
      "loss": 1.5822,
      "step": 2965
    },
    {
      "epoch": 0.44732674760576124,
      "grad_norm": 0.48549240991872344,
      "learning_rate": 3.120393175934277e-07,
      "loss": 1.5265,
      "step": 2966
    },
    {
      "epoch": 0.4474775657944348,
      "grad_norm": 0.24606427073477874,
      "learning_rate": 3.1193415283301327e-07,
      "loss": 1.6047,
      "step": 2967
    },
    {
      "epoch": 0.44762838398310834,
      "grad_norm": 0.31989339634341063,
      "learning_rate": 3.11828979779806e-07,
      "loss": 1.4801,
      "step": 2968
    },
    {
      "epoch": 0.4477792021717819,
      "grad_norm": 0.28372722722218535,
      "learning_rate": 3.1172379845742013e-07,
      "loss": 1.5434,
      "step": 2969
    },
    {
      "epoch": 0.4479300203604555,
      "grad_norm": 0.23650434231940212,
      "learning_rate": 3.1161860888947196e-07,
      "loss": 1.5475,
      "step": 2970
    },
    {
      "epoch": 0.448080838549129,
      "grad_norm": 0.2614404249119146,
      "learning_rate": 3.115134110995797e-07,
      "loss": 1.5261,
      "step": 2971
    },
    {
      "epoch": 0.4482316567378026,
      "grad_norm": 0.2582017725860931,
      "learning_rate": 3.1140820511136324e-07,
      "loss": 1.5125,
      "step": 2972
    },
    {
      "epoch": 0.44838247492647615,
      "grad_norm": 0.26566919697793095,
      "learning_rate": 3.113029909484444e-07,
      "loss": 1.5394,
      "step": 2973
    },
    {
      "epoch": 0.44853329311514967,
      "grad_norm": 0.25140893194319885,
      "learning_rate": 3.111977686344469e-07,
      "loss": 1.5757,
      "step": 2974
    },
    {
      "epoch": 0.44868411130382324,
      "grad_norm": 0.24657104009489716,
      "learning_rate": 3.1109253819299604e-07,
      "loss": 1.587,
      "step": 2975
    },
    {
      "epoch": 0.4488349294924968,
      "grad_norm": 0.2583748689263846,
      "learning_rate": 3.1098729964771927e-07,
      "loss": 1.4778,
      "step": 2976
    },
    {
      "epoch": 0.44898574768117033,
      "grad_norm": 0.27974542845709593,
      "learning_rate": 3.108820530222457e-07,
      "loss": 1.5788,
      "step": 2977
    },
    {
      "epoch": 0.4491365658698439,
      "grad_norm": 0.2472039824429126,
      "learning_rate": 3.1077679834020614e-07,
      "loss": 1.5599,
      "step": 2978
    },
    {
      "epoch": 0.4492873840585175,
      "grad_norm": 0.25024405958197155,
      "learning_rate": 3.1067153562523353e-07,
      "loss": 1.5422,
      "step": 2979
    },
    {
      "epoch": 0.449438202247191,
      "grad_norm": 0.2963575226276379,
      "learning_rate": 3.1056626490096217e-07,
      "loss": 1.5686,
      "step": 2980
    },
    {
      "epoch": 0.4495890204358646,
      "grad_norm": 0.3800371321944895,
      "learning_rate": 3.104609861910286e-07,
      "loss": 1.4612,
      "step": 2981
    },
    {
      "epoch": 0.44973983862453815,
      "grad_norm": 0.29875841844483303,
      "learning_rate": 3.10355699519071e-07,
      "loss": 1.5737,
      "step": 2982
    },
    {
      "epoch": 0.44989065681321166,
      "grad_norm": 0.3116187392098609,
      "learning_rate": 3.10250404908729e-07,
      "loss": 1.6017,
      "step": 2983
    },
    {
      "epoch": 0.45004147500188524,
      "grad_norm": 0.2684681459704842,
      "learning_rate": 3.101451023836446e-07,
      "loss": 1.5569,
      "step": 2984
    },
    {
      "epoch": 0.45019229319055876,
      "grad_norm": 0.24529043085332278,
      "learning_rate": 3.100397919674611e-07,
      "loss": 1.4803,
      "step": 2985
    },
    {
      "epoch": 0.45034311137923233,
      "grad_norm": 0.24358539839894225,
      "learning_rate": 3.0993447368382394e-07,
      "loss": 1.5215,
      "step": 2986
    },
    {
      "epoch": 0.4504939295679059,
      "grad_norm": 0.2502172714852228,
      "learning_rate": 3.0982914755637994e-07,
      "loss": 1.5563,
      "step": 2987
    },
    {
      "epoch": 0.4506447477565794,
      "grad_norm": 0.2582609803453465,
      "learning_rate": 3.0972381360877805e-07,
      "loss": 1.6238,
      "step": 2988
    },
    {
      "epoch": 0.450795565945253,
      "grad_norm": 0.23778611944596112,
      "learning_rate": 3.0961847186466865e-07,
      "loss": 1.625,
      "step": 2989
    },
    {
      "epoch": 0.45094638413392657,
      "grad_norm": 0.24203211370235533,
      "learning_rate": 3.0951312234770424e-07,
      "loss": 1.5784,
      "step": 2990
    },
    {
      "epoch": 0.4510972023226001,
      "grad_norm": 0.2624287141724599,
      "learning_rate": 3.094077650815386e-07,
      "loss": 1.5745,
      "step": 2991
    },
    {
      "epoch": 0.45124802051127366,
      "grad_norm": 0.2515527215618168,
      "learning_rate": 3.093024000898277e-07,
      "loss": 1.5258,
      "step": 2992
    },
    {
      "epoch": 0.45139883869994724,
      "grad_norm": 0.2455295872157118,
      "learning_rate": 3.09197027396229e-07,
      "loss": 1.5576,
      "step": 2993
    },
    {
      "epoch": 0.45154965688862075,
      "grad_norm": 0.2531843365496076,
      "learning_rate": 3.0909164702440166e-07,
      "loss": 1.5989,
      "step": 2994
    },
    {
      "epoch": 0.45170047507729433,
      "grad_norm": 0.6742080276924932,
      "learning_rate": 3.089862589980067e-07,
      "loss": 1.6633,
      "step": 2995
    },
    {
      "epoch": 0.4518512932659679,
      "grad_norm": 0.25751381611676893,
      "learning_rate": 3.088808633407069e-07,
      "loss": 1.5642,
      "step": 2996
    },
    {
      "epoch": 0.4520021114546414,
      "grad_norm": 0.2805235006759765,
      "learning_rate": 3.0877546007616656e-07,
      "loss": 1.5874,
      "step": 2997
    },
    {
      "epoch": 0.452152929643315,
      "grad_norm": 0.2811107254257494,
      "learning_rate": 3.086700492280517e-07,
      "loss": 1.4869,
      "step": 2998
    },
    {
      "epoch": 0.4523037478319885,
      "grad_norm": 0.25027636099509565,
      "learning_rate": 3.085646308200302e-07,
      "loss": 1.5712,
      "step": 2999
    },
    {
      "epoch": 0.4524545660206621,
      "grad_norm": 0.2576381839230806,
      "learning_rate": 3.084592048757716e-07,
      "loss": 1.5812,
      "step": 3000
    },
    {
      "epoch": 0.45260538420933566,
      "grad_norm": 0.24988067035681372,
      "learning_rate": 3.083537714189471e-07,
      "loss": 1.5602,
      "step": 3001
    },
    {
      "epoch": 0.4527562023980092,
      "grad_norm": 0.35235848555933336,
      "learning_rate": 3.082483304732295e-07,
      "loss": 1.5458,
      "step": 3002
    },
    {
      "epoch": 0.45290702058668275,
      "grad_norm": 0.3335011281009601,
      "learning_rate": 3.0814288206229343e-07,
      "loss": 1.5411,
      "step": 3003
    },
    {
      "epoch": 0.4530578387753563,
      "grad_norm": 0.2521306698791328,
      "learning_rate": 3.080374262098152e-07,
      "loss": 1.551,
      "step": 3004
    },
    {
      "epoch": 0.45320865696402984,
      "grad_norm": 0.24140037223562982,
      "learning_rate": 3.079319629394724e-07,
      "loss": 1.5487,
      "step": 3005
    },
    {
      "epoch": 0.4533594751527034,
      "grad_norm": 0.254843874309324,
      "learning_rate": 3.07826492274945e-07,
      "loss": 1.5515,
      "step": 3006
    },
    {
      "epoch": 0.453510293341377,
      "grad_norm": 0.255684472071238,
      "learning_rate": 3.0772101423991407e-07,
      "loss": 1.5745,
      "step": 3007
    },
    {
      "epoch": 0.4536611115300505,
      "grad_norm": 0.29836640169693746,
      "learning_rate": 3.076155288580625e-07,
      "loss": 1.5514,
      "step": 3008
    },
    {
      "epoch": 0.4538119297187241,
      "grad_norm": 0.2518153306090227,
      "learning_rate": 3.0751003615307474e-07,
      "loss": 1.5746,
      "step": 3009
    },
    {
      "epoch": 0.45396274790739766,
      "grad_norm": 0.23894055808240028,
      "learning_rate": 3.074045361486372e-07,
      "loss": 1.5579,
      "step": 3010
    },
    {
      "epoch": 0.4541135660960712,
      "grad_norm": 0.24187245513753344,
      "learning_rate": 3.072990288684375e-07,
      "loss": 1.5418,
      "step": 3011
    },
    {
      "epoch": 0.45426438428474475,
      "grad_norm": 0.252062380284074,
      "learning_rate": 3.071935143361651e-07,
      "loss": 1.5777,
      "step": 3012
    },
    {
      "epoch": 0.45441520247341827,
      "grad_norm": 0.2505843481528514,
      "learning_rate": 3.0708799257551124e-07,
      "loss": 1.6039,
      "step": 3013
    },
    {
      "epoch": 0.45456602066209184,
      "grad_norm": 0.5721456375520992,
      "learning_rate": 3.0698246361016855e-07,
      "loss": 1.4929,
      "step": 3014
    },
    {
      "epoch": 0.4547168388507654,
      "grad_norm": 0.27545721685806984,
      "learning_rate": 3.068769274638314e-07,
      "loss": 1.5755,
      "step": 3015
    },
    {
      "epoch": 0.45486765703943893,
      "grad_norm": 0.2782343763490815,
      "learning_rate": 3.0677138416019556e-07,
      "loss": 1.4811,
      "step": 3016
    },
    {
      "epoch": 0.4550184752281125,
      "grad_norm": 0.261269937796045,
      "learning_rate": 3.066658337229587e-07,
      "loss": 1.5506,
      "step": 3017
    },
    {
      "epoch": 0.4551692934167861,
      "grad_norm": 0.2710553683918107,
      "learning_rate": 3.0656027617582e-07,
      "loss": 1.634,
      "step": 3018
    },
    {
      "epoch": 0.4553201116054596,
      "grad_norm": 0.28984734875281387,
      "learning_rate": 3.064547115424802e-07,
      "loss": 1.5897,
      "step": 3019
    },
    {
      "epoch": 0.4554709297941332,
      "grad_norm": 0.2570452835035077,
      "learning_rate": 3.063491398466415e-07,
      "loss": 1.608,
      "step": 3020
    },
    {
      "epoch": 0.45562174798280675,
      "grad_norm": 0.27443754458266795,
      "learning_rate": 3.06243561112008e-07,
      "loss": 1.5134,
      "step": 3021
    },
    {
      "epoch": 0.45577256617148026,
      "grad_norm": 0.24472397015478006,
      "learning_rate": 3.0613797536228507e-07,
      "loss": 1.5881,
      "step": 3022
    },
    {
      "epoch": 0.45592338436015384,
      "grad_norm": 0.2797386856040963,
      "learning_rate": 3.060323826211799e-07,
      "loss": 1.5273,
      "step": 3023
    },
    {
      "epoch": 0.4560742025488274,
      "grad_norm": 0.24202137903029028,
      "learning_rate": 3.05926782912401e-07,
      "loss": 1.6152,
      "step": 3024
    },
    {
      "epoch": 0.45622502073750093,
      "grad_norm": 0.2589108746963981,
      "learning_rate": 3.0582117625965864e-07,
      "loss": 1.5963,
      "step": 3025
    },
    {
      "epoch": 0.4563758389261745,
      "grad_norm": 0.25427670811518555,
      "learning_rate": 3.0571556268666463e-07,
      "loss": 1.6224,
      "step": 3026
    },
    {
      "epoch": 0.4565266571148481,
      "grad_norm": 0.2587703030443672,
      "learning_rate": 3.0560994221713223e-07,
      "loss": 1.5612,
      "step": 3027
    },
    {
      "epoch": 0.4566774753035216,
      "grad_norm": 0.2886694938081297,
      "learning_rate": 3.055043148747764e-07,
      "loss": 1.5324,
      "step": 3028
    },
    {
      "epoch": 0.45682829349219517,
      "grad_norm": 0.31831612429832107,
      "learning_rate": 3.0539868068331344e-07,
      "loss": 1.518,
      "step": 3029
    },
    {
      "epoch": 0.4569791116808687,
      "grad_norm": 0.24782382733151562,
      "learning_rate": 3.0529303966646137e-07,
      "loss": 1.5928,
      "step": 3030
    },
    {
      "epoch": 0.45712992986954226,
      "grad_norm": 0.2599559228051775,
      "learning_rate": 3.051873918479397e-07,
      "loss": 1.5563,
      "step": 3031
    },
    {
      "epoch": 0.45728074805821584,
      "grad_norm": 0.2444135393221858,
      "learning_rate": 3.0508173725146934e-07,
      "loss": 1.5639,
      "step": 3032
    },
    {
      "epoch": 0.45743156624688935,
      "grad_norm": 0.25805125896054365,
      "learning_rate": 3.049760759007729e-07,
      "loss": 1.5088,
      "step": 3033
    },
    {
      "epoch": 0.4575823844355629,
      "grad_norm": 0.2738356446282475,
      "learning_rate": 3.0487040781957447e-07,
      "loss": 1.5391,
      "step": 3034
    },
    {
      "epoch": 0.4577332026242365,
      "grad_norm": 0.2570037548605811,
      "learning_rate": 3.0476473303159957e-07,
      "loss": 1.4543,
      "step": 3035
    },
    {
      "epoch": 0.45788402081291,
      "grad_norm": 0.271751224706055,
      "learning_rate": 3.0465905156057527e-07,
      "loss": 1.5705,
      "step": 3036
    },
    {
      "epoch": 0.4580348390015836,
      "grad_norm": 0.24170106683820172,
      "learning_rate": 3.045533634302301e-07,
      "loss": 1.506,
      "step": 3037
    },
    {
      "epoch": 0.45818565719025717,
      "grad_norm": 0.2590299736767002,
      "learning_rate": 3.044476686642941e-07,
      "loss": 1.5761,
      "step": 3038
    },
    {
      "epoch": 0.4583364753789307,
      "grad_norm": 0.23943188904155357,
      "learning_rate": 3.043419672864989e-07,
      "loss": 1.5635,
      "step": 3039
    },
    {
      "epoch": 0.45848729356760426,
      "grad_norm": 0.3128732040011657,
      "learning_rate": 3.042362593205775e-07,
      "loss": 1.5952,
      "step": 3040
    },
    {
      "epoch": 0.45863811175627783,
      "grad_norm": 0.25503919824492927,
      "learning_rate": 3.0413054479026445e-07,
      "loss": 1.575,
      "step": 3041
    },
    {
      "epoch": 0.45878892994495135,
      "grad_norm": 0.24210971507913032,
      "learning_rate": 3.040248237192958e-07,
      "loss": 1.5675,
      "step": 3042
    },
    {
      "epoch": 0.4589397481336249,
      "grad_norm": 0.26816280778517526,
      "learning_rate": 3.039190961314088e-07,
      "loss": 1.559,
      "step": 3043
    },
    {
      "epoch": 0.45909056632229844,
      "grad_norm": 0.3617187003581852,
      "learning_rate": 3.0381336205034254e-07,
      "loss": 1.5871,
      "step": 3044
    },
    {
      "epoch": 0.459241384510972,
      "grad_norm": 0.25964865678477417,
      "learning_rate": 3.0370762149983743e-07,
      "loss": 1.5954,
      "step": 3045
    },
    {
      "epoch": 0.4593922026996456,
      "grad_norm": 0.306725325338039,
      "learning_rate": 3.036018745036351e-07,
      "loss": 1.5613,
      "step": 3046
    },
    {
      "epoch": 0.4595430208883191,
      "grad_norm": 0.32936632251261777,
      "learning_rate": 3.034961210854791e-07,
      "loss": 1.5149,
      "step": 3047
    },
    {
      "epoch": 0.4596938390769927,
      "grad_norm": 0.25780996824913954,
      "learning_rate": 3.033903612691139e-07,
      "loss": 1.6034,
      "step": 3048
    },
    {
      "epoch": 0.45984465726566626,
      "grad_norm": 0.24749906096744248,
      "learning_rate": 3.032845950782859e-07,
      "loss": 1.5395,
      "step": 3049
    },
    {
      "epoch": 0.4599954754543398,
      "grad_norm": 0.27887263046014993,
      "learning_rate": 3.031788225367425e-07,
      "loss": 1.5389,
      "step": 3050
    },
    {
      "epoch": 0.46014629364301335,
      "grad_norm": 0.2564103298744664,
      "learning_rate": 3.0307304366823276e-07,
      "loss": 1.5118,
      "step": 3051
    },
    {
      "epoch": 0.4602971118316869,
      "grad_norm": 0.24527121197302515,
      "learning_rate": 3.029672584965072e-07,
      "loss": 1.6176,
      "step": 3052
    },
    {
      "epoch": 0.46044793002036044,
      "grad_norm": 0.25248485727309306,
      "learning_rate": 3.0286146704531756e-07,
      "loss": 1.5881,
      "step": 3053
    },
    {
      "epoch": 0.460598748209034,
      "grad_norm": 0.24647820855713623,
      "learning_rate": 3.027556693384172e-07,
      "loss": 1.6539,
      "step": 3054
    },
    {
      "epoch": 0.4607495663977076,
      "grad_norm": 0.24630504819560922,
      "learning_rate": 3.0264986539956065e-07,
      "loss": 1.5517,
      "step": 3055
    },
    {
      "epoch": 0.4609003845863811,
      "grad_norm": 0.2670965324389809,
      "learning_rate": 3.025440552525042e-07,
      "loss": 1.5974,
      "step": 3056
    },
    {
      "epoch": 0.4610512027750547,
      "grad_norm": 0.2439765244039028,
      "learning_rate": 3.02438238921005e-07,
      "loss": 1.5821,
      "step": 3057
    },
    {
      "epoch": 0.4612020209637282,
      "grad_norm": 0.24861411178427464,
      "learning_rate": 3.0233241642882225e-07,
      "loss": 1.5246,
      "step": 3058
    },
    {
      "epoch": 0.4613528391524018,
      "grad_norm": 0.3855492466112723,
      "learning_rate": 3.0222658779971587e-07,
      "loss": 1.5778,
      "step": 3059
    },
    {
      "epoch": 0.46150365734107535,
      "grad_norm": 0.3141524633237537,
      "learning_rate": 3.0212075305744767e-07,
      "loss": 1.521,
      "step": 3060
    },
    {
      "epoch": 0.46165447552974886,
      "grad_norm": 0.26132407224167226,
      "learning_rate": 3.020149122257805e-07,
      "loss": 1.578,
      "step": 3061
    },
    {
      "epoch": 0.46180529371842244,
      "grad_norm": 0.3654418283420951,
      "learning_rate": 3.019090653284789e-07,
      "loss": 1.562,
      "step": 3062
    },
    {
      "epoch": 0.461956111907096,
      "grad_norm": 0.6281917699855586,
      "learning_rate": 3.0180321238930833e-07,
      "loss": 1.6092,
      "step": 3063
    },
    {
      "epoch": 0.46210693009576953,
      "grad_norm": 0.4258721548197494,
      "learning_rate": 3.016973534320361e-07,
      "loss": 1.5411,
      "step": 3064
    },
    {
      "epoch": 0.4622577482844431,
      "grad_norm": 0.2658585703340917,
      "learning_rate": 3.015914884804304e-07,
      "loss": 1.5973,
      "step": 3065
    },
    {
      "epoch": 0.4624085664731167,
      "grad_norm": 0.25974042461853114,
      "learning_rate": 3.0148561755826117e-07,
      "loss": 1.6426,
      "step": 3066
    },
    {
      "epoch": 0.4625593846617902,
      "grad_norm": 0.36689913431669297,
      "learning_rate": 3.0137974068929945e-07,
      "loss": 1.593,
      "step": 3067
    },
    {
      "epoch": 0.46271020285046377,
      "grad_norm": 0.25700433697530406,
      "learning_rate": 3.0127385789731765e-07,
      "loss": 1.5169,
      "step": 3068
    },
    {
      "epoch": 0.46286102103913734,
      "grad_norm": 0.5325222957041995,
      "learning_rate": 3.011679692060896e-07,
      "loss": 1.5924,
      "step": 3069
    },
    {
      "epoch": 0.46301183922781086,
      "grad_norm": 0.28376571813597284,
      "learning_rate": 3.010620746393903e-07,
      "loss": 1.5968,
      "step": 3070
    },
    {
      "epoch": 0.46316265741648444,
      "grad_norm": 0.2497627878737484,
      "learning_rate": 3.0095617422099636e-07,
      "loss": 1.6907,
      "step": 3071
    },
    {
      "epoch": 0.463313475605158,
      "grad_norm": 0.258362993634684,
      "learning_rate": 3.008502679746853e-07,
      "loss": 1.5409,
      "step": 3072
    },
    {
      "epoch": 0.4634642937938315,
      "grad_norm": 0.27915505042843364,
      "learning_rate": 3.0074435592423624e-07,
      "loss": 1.569,
      "step": 3073
    },
    {
      "epoch": 0.4636151119825051,
      "grad_norm": 0.24501981442207682,
      "learning_rate": 3.0063843809342956e-07,
      "loss": 1.4864,
      "step": 3074
    },
    {
      "epoch": 0.4637659301711786,
      "grad_norm": 0.2510754371313347,
      "learning_rate": 3.005325145060468e-07,
      "loss": 1.5793,
      "step": 3075
    },
    {
      "epoch": 0.4639167483598522,
      "grad_norm": 0.2918637363545153,
      "learning_rate": 3.0042658518587095e-07,
      "loss": 1.5442,
      "step": 3076
    },
    {
      "epoch": 0.46406756654852577,
      "grad_norm": 0.41762300505875893,
      "learning_rate": 3.003206501566862e-07,
      "loss": 1.5794,
      "step": 3077
    },
    {
      "epoch": 0.4642183847371993,
      "grad_norm": 0.2499665774551008,
      "learning_rate": 3.002147094422781e-07,
      "loss": 1.4934,
      "step": 3078
    },
    {
      "epoch": 0.46436920292587286,
      "grad_norm": 0.26302306616238386,
      "learning_rate": 3.001087630664334e-07,
      "loss": 1.5621,
      "step": 3079
    },
    {
      "epoch": 0.46452002111454643,
      "grad_norm": 0.2897997184428593,
      "learning_rate": 3.0000281105294016e-07,
      "loss": 1.5685,
      "step": 3080
    },
    {
      "epoch": 0.46467083930321995,
      "grad_norm": 0.2920128436423677,
      "learning_rate": 2.998968534255877e-07,
      "loss": 1.4975,
      "step": 3081
    },
    {
      "epoch": 0.4648216574918935,
      "grad_norm": 0.2430946934216263,
      "learning_rate": 2.9979089020816656e-07,
      "loss": 1.5453,
      "step": 3082
    },
    {
      "epoch": 0.4649724756805671,
      "grad_norm": 0.29485762214691674,
      "learning_rate": 2.996849214244685e-07,
      "loss": 1.5913,
      "step": 3083
    },
    {
      "epoch": 0.4651232938692406,
      "grad_norm": 0.27298216278564874,
      "learning_rate": 2.995789470982867e-07,
      "loss": 1.604,
      "step": 3084
    },
    {
      "epoch": 0.4652741120579142,
      "grad_norm": 0.24041170018947816,
      "learning_rate": 2.994729672534155e-07,
      "loss": 1.637,
      "step": 3085
    },
    {
      "epoch": 0.46542493024658776,
      "grad_norm": 0.2443504997293059,
      "learning_rate": 2.9936698191365047e-07,
      "loss": 1.5101,
      "step": 3086
    },
    {
      "epoch": 0.4655757484352613,
      "grad_norm": 0.24946065590667868,
      "learning_rate": 2.9926099110278823e-07,
      "loss": 1.6411,
      "step": 3087
    },
    {
      "epoch": 0.46572656662393486,
      "grad_norm": 0.2517409955245919,
      "learning_rate": 2.99154994844627e-07,
      "loss": 1.5731,
      "step": 3088
    },
    {
      "epoch": 0.4658773848126084,
      "grad_norm": 0.4712226365325641,
      "learning_rate": 2.9904899316296593e-07,
      "loss": 1.5635,
      "step": 3089
    },
    {
      "epoch": 0.46602820300128195,
      "grad_norm": 0.26267189317506817,
      "learning_rate": 2.9894298608160557e-07,
      "loss": 1.5482,
      "step": 3090
    },
    {
      "epoch": 0.4661790211899555,
      "grad_norm": 0.25096318846535004,
      "learning_rate": 2.9883697362434743e-07,
      "loss": 1.5473,
      "step": 3091
    },
    {
      "epoch": 0.46632983937862904,
      "grad_norm": 0.26490772731617646,
      "learning_rate": 2.9873095581499456e-07,
      "loss": 1.5459,
      "step": 3092
    },
    {
      "epoch": 0.4664806575673026,
      "grad_norm": 0.25622179136950635,
      "learning_rate": 2.986249326773511e-07,
      "loss": 1.5778,
      "step": 3093
    },
    {
      "epoch": 0.4666314757559762,
      "grad_norm": 0.4181208343695358,
      "learning_rate": 2.98518904235222e-07,
      "loss": 1.5525,
      "step": 3094
    },
    {
      "epoch": 0.4667822939446497,
      "grad_norm": 0.24402009173677996,
      "learning_rate": 2.984128705124141e-07,
      "loss": 1.4885,
      "step": 3095
    },
    {
      "epoch": 0.4669331121333233,
      "grad_norm": 0.27012935338310423,
      "learning_rate": 2.983068315327349e-07,
      "loss": 1.5589,
      "step": 3096
    },
    {
      "epoch": 0.46708393032199685,
      "grad_norm": 0.2822387723132698,
      "learning_rate": 2.982007873199932e-07,
      "loss": 1.583,
      "step": 3097
    },
    {
      "epoch": 0.4672347485106704,
      "grad_norm": 1.384971284858126,
      "learning_rate": 2.980947378979991e-07,
      "loss": 1.5548,
      "step": 3098
    },
    {
      "epoch": 0.46738556669934395,
      "grad_norm": 0.24793988515751572,
      "learning_rate": 2.9798868329056377e-07,
      "loss": 1.5959,
      "step": 3099
    },
    {
      "epoch": 0.4675363848880175,
      "grad_norm": 0.2562310729223106,
      "learning_rate": 2.978826235214995e-07,
      "loss": 1.5912,
      "step": 3100
    },
    {
      "epoch": 0.46768720307669104,
      "grad_norm": 0.26118119823269886,
      "learning_rate": 2.9777655861461983e-07,
      "loss": 1.5677,
      "step": 3101
    },
    {
      "epoch": 0.4678380212653646,
      "grad_norm": 0.25834739995520306,
      "learning_rate": 2.976704885937394e-07,
      "loss": 1.5678,
      "step": 3102
    },
    {
      "epoch": 0.46798883945403813,
      "grad_norm": 0.34608384913561635,
      "learning_rate": 2.9756441348267415e-07,
      "loss": 1.5201,
      "step": 3103
    },
    {
      "epoch": 0.4681396576427117,
      "grad_norm": 0.3488184526720022,
      "learning_rate": 2.9745833330524077e-07,
      "loss": 1.5947,
      "step": 3104
    },
    {
      "epoch": 0.4682904758313853,
      "grad_norm": 0.24643312832198783,
      "learning_rate": 2.973522480852576e-07,
      "loss": 1.5429,
      "step": 3105
    },
    {
      "epoch": 0.4684412940200588,
      "grad_norm": 0.24367773790462405,
      "learning_rate": 2.9724615784654376e-07,
      "loss": 1.577,
      "step": 3106
    },
    {
      "epoch": 0.46859211220873237,
      "grad_norm": 0.25019008015061117,
      "learning_rate": 2.9714006261291964e-07,
      "loss": 1.5464,
      "step": 3107
    },
    {
      "epoch": 0.46874293039740594,
      "grad_norm": 1.4631440444528507,
      "learning_rate": 2.970339624082066e-07,
      "loss": 1.53,
      "step": 3108
    },
    {
      "epoch": 0.46889374858607946,
      "grad_norm": 0.2553858145910921,
      "learning_rate": 2.9692785725622737e-07,
      "loss": 1.602,
      "step": 3109
    },
    {
      "epoch": 0.46904456677475304,
      "grad_norm": 0.28008841638244675,
      "learning_rate": 2.968217471808055e-07,
      "loss": 1.5214,
      "step": 3110
    },
    {
      "epoch": 0.4691953849634266,
      "grad_norm": 0.465635413118692,
      "learning_rate": 2.967156322057659e-07,
      "loss": 1.6019,
      "step": 3111
    },
    {
      "epoch": 0.4693462031521001,
      "grad_norm": 0.26496000760437305,
      "learning_rate": 2.966095123549345e-07,
      "loss": 1.5575,
      "step": 3112
    },
    {
      "epoch": 0.4694970213407737,
      "grad_norm": 0.2574373126488699,
      "learning_rate": 2.965033876521382e-07,
      "loss": 1.588,
      "step": 3113
    },
    {
      "epoch": 0.4696478395294473,
      "grad_norm": 0.2541011934834178,
      "learning_rate": 2.9639725812120505e-07,
      "loss": 1.6224,
      "step": 3114
    },
    {
      "epoch": 0.4697986577181208,
      "grad_norm": 0.25609855203229226,
      "learning_rate": 2.962911237859643e-07,
      "loss": 1.6339,
      "step": 3115
    },
    {
      "epoch": 0.46994947590679437,
      "grad_norm": 0.2457549898166761,
      "learning_rate": 2.9618498467024634e-07,
      "loss": 1.6087,
      "step": 3116
    },
    {
      "epoch": 0.47010029409546794,
      "grad_norm": 0.26116033707396874,
      "learning_rate": 2.960788407978822e-07,
      "loss": 1.472,
      "step": 3117
    },
    {
      "epoch": 0.47025111228414146,
      "grad_norm": 0.2494061211539528,
      "learning_rate": 2.9597269219270444e-07,
      "loss": 1.5734,
      "step": 3118
    },
    {
      "epoch": 0.47040193047281503,
      "grad_norm": 0.24559955866207117,
      "learning_rate": 2.9586653887854655e-07,
      "loss": 1.5252,
      "step": 3119
    },
    {
      "epoch": 0.47055274866148855,
      "grad_norm": 0.24695024440276261,
      "learning_rate": 2.957603808792429e-07,
      "loss": 1.5879,
      "step": 3120
    },
    {
      "epoch": 0.4707035668501621,
      "grad_norm": 0.7056298660755264,
      "learning_rate": 2.956542182186293e-07,
      "loss": 1.5673,
      "step": 3121
    },
    {
      "epoch": 0.4708543850388357,
      "grad_norm": 6.141345332581017,
      "learning_rate": 2.95548050920542e-07,
      "loss": 1.613,
      "step": 3122
    },
    {
      "epoch": 0.4710052032275092,
      "grad_norm": 0.26760459105842394,
      "learning_rate": 2.9544187900881893e-07,
      "loss": 1.5802,
      "step": 3123
    },
    {
      "epoch": 0.4711560214161828,
      "grad_norm": 0.27852768061774646,
      "learning_rate": 2.9533570250729875e-07,
      "loss": 1.5725,
      "step": 3124
    },
    {
      "epoch": 0.47130683960485636,
      "grad_norm": 0.2553576747982788,
      "learning_rate": 2.95229521439821e-07,
      "loss": 1.4959,
      "step": 3125
    },
    {
      "epoch": 0.4714576577935299,
      "grad_norm": 1.228355826218024,
      "learning_rate": 2.951233358302266e-07,
      "loss": 1.5456,
      "step": 3126
    },
    {
      "epoch": 0.47160847598220346,
      "grad_norm": 0.26371091643543787,
      "learning_rate": 2.950171457023572e-07,
      "loss": 1.5563,
      "step": 3127
    },
    {
      "epoch": 0.47175929417087703,
      "grad_norm": 0.2970776800812065,
      "learning_rate": 2.9491095108005567e-07,
      "loss": 1.576,
      "step": 3128
    },
    {
      "epoch": 0.47191011235955055,
      "grad_norm": 0.25871172786052893,
      "learning_rate": 2.9480475198716576e-07,
      "loss": 1.4758,
      "step": 3129
    },
    {
      "epoch": 0.4720609305482241,
      "grad_norm": 0.25080898052773576,
      "learning_rate": 2.9469854844753217e-07,
      "loss": 1.6025,
      "step": 3130
    },
    {
      "epoch": 0.4722117487368977,
      "grad_norm": 0.264607073476825,
      "learning_rate": 2.9459234048500086e-07,
      "loss": 1.5558,
      "step": 3131
    },
    {
      "epoch": 0.4723625669255712,
      "grad_norm": 0.25309447185710987,
      "learning_rate": 2.9448612812341855e-07,
      "loss": 1.5803,
      "step": 3132
    },
    {
      "epoch": 0.4725133851142448,
      "grad_norm": 0.2651919462998949,
      "learning_rate": 2.9437991138663286e-07,
      "loss": 1.5595,
      "step": 3133
    },
    {
      "epoch": 0.4726642033029183,
      "grad_norm": 0.25671274255063314,
      "learning_rate": 2.9427369029849276e-07,
      "loss": 1.6497,
      "step": 3134
    },
    {
      "epoch": 0.4728150214915919,
      "grad_norm": 0.2523704972360593,
      "learning_rate": 2.941674648828478e-07,
      "loss": 1.5396,
      "step": 3135
    },
    {
      "epoch": 0.47296583968026545,
      "grad_norm": 0.24582104701925128,
      "learning_rate": 2.9406123516354873e-07,
      "loss": 1.56,
      "step": 3136
    },
    {
      "epoch": 0.473116657868939,
      "grad_norm": 0.2414495282978198,
      "learning_rate": 2.9395500116444736e-07,
      "loss": 1.4739,
      "step": 3137
    },
    {
      "epoch": 0.47326747605761255,
      "grad_norm": 0.2700220556284739,
      "learning_rate": 2.9384876290939613e-07,
      "loss": 1.5267,
      "step": 3138
    },
    {
      "epoch": 0.4734182942462861,
      "grad_norm": 0.2502324744654963,
      "learning_rate": 2.9374252042224875e-07,
      "loss": 1.5575,
      "step": 3139
    },
    {
      "epoch": 0.47356911243495964,
      "grad_norm": 0.24877617133567306,
      "learning_rate": 2.9363627372685957e-07,
      "loss": 1.5306,
      "step": 3140
    },
    {
      "epoch": 0.4737199306236332,
      "grad_norm": 0.2763882887099,
      "learning_rate": 2.935300228470843e-07,
      "loss": 1.5368,
      "step": 3141
    },
    {
      "epoch": 0.4738707488123068,
      "grad_norm": 0.23525758933340374,
      "learning_rate": 2.934237678067792e-07,
      "loss": 1.5106,
      "step": 3142
    },
    {
      "epoch": 0.4740215670009803,
      "grad_norm": 0.2940263563569856,
      "learning_rate": 2.9331750862980166e-07,
      "loss": 1.5898,
      "step": 3143
    },
    {
      "epoch": 0.4741723851896539,
      "grad_norm": 0.2596657573389365,
      "learning_rate": 2.9321124534001e-07,
      "loss": 1.5476,
      "step": 3144
    },
    {
      "epoch": 0.47432320337832745,
      "grad_norm": 0.2529751981624563,
      "learning_rate": 2.931049779612634e-07,
      "loss": 1.5242,
      "step": 3145
    },
    {
      "epoch": 0.47447402156700097,
      "grad_norm": 0.2500990983748969,
      "learning_rate": 2.929987065174219e-07,
      "loss": 1.6329,
      "step": 3146
    },
    {
      "epoch": 0.47462483975567454,
      "grad_norm": 0.24968977008231993,
      "learning_rate": 2.9289243103234665e-07,
      "loss": 1.5681,
      "step": 3147
    },
    {
      "epoch": 0.47477565794434806,
      "grad_norm": 0.24654663425811366,
      "learning_rate": 2.9278615152989955e-07,
      "loss": 1.5237,
      "step": 3148
    },
    {
      "epoch": 0.47492647613302164,
      "grad_norm": 0.24200923324226278,
      "learning_rate": 2.9267986803394346e-07,
      "loss": 1.482,
      "step": 3149
    },
    {
      "epoch": 0.4750772943216952,
      "grad_norm": 0.26612235405655793,
      "learning_rate": 2.9257358056834206e-07,
      "loss": 1.604,
      "step": 3150
    },
    {
      "epoch": 0.4752281125103687,
      "grad_norm": 0.24988595389853024,
      "learning_rate": 2.9246728915696006e-07,
      "loss": 1.5565,
      "step": 3151
    },
    {
      "epoch": 0.4753789306990423,
      "grad_norm": 0.2651130099799271,
      "learning_rate": 2.923609938236629e-07,
      "loss": 1.4943,
      "step": 3152
    },
    {
      "epoch": 0.4755297488877159,
      "grad_norm": 0.28713226071833503,
      "learning_rate": 2.9225469459231703e-07,
      "loss": 1.5498,
      "step": 3153
    },
    {
      "epoch": 0.4756805670763894,
      "grad_norm": 0.2525910632174739,
      "learning_rate": 2.921483914867897e-07,
      "loss": 1.5784,
      "step": 3154
    },
    {
      "epoch": 0.47583138526506297,
      "grad_norm": 0.3356412422958949,
      "learning_rate": 2.920420845309492e-07,
      "loss": 1.5583,
      "step": 3155
    },
    {
      "epoch": 0.47598220345373654,
      "grad_norm": 0.24307731624206946,
      "learning_rate": 2.9193577374866423e-07,
      "loss": 1.5684,
      "step": 3156
    },
    {
      "epoch": 0.47613302164241006,
      "grad_norm": 0.24358134785836383,
      "learning_rate": 2.9182945916380493e-07,
      "loss": 1.6196,
      "step": 3157
    },
    {
      "epoch": 0.47628383983108363,
      "grad_norm": 0.30551696235434356,
      "learning_rate": 2.9172314080024187e-07,
      "loss": 1.5296,
      "step": 3158
    },
    {
      "epoch": 0.4764346580197572,
      "grad_norm": 0.2881545017403535,
      "learning_rate": 2.9161681868184666e-07,
      "loss": 1.6224,
      "step": 3159
    },
    {
      "epoch": 0.4765854762084307,
      "grad_norm": 0.25304889949388404,
      "learning_rate": 2.915104928324917e-07,
      "loss": 1.5174,
      "step": 3160
    },
    {
      "epoch": 0.4767362943971043,
      "grad_norm": 0.24605270505538374,
      "learning_rate": 2.9140416327605025e-07,
      "loss": 1.6102,
      "step": 3161
    },
    {
      "epoch": 0.4768871125857779,
      "grad_norm": 0.25091348999395624,
      "learning_rate": 2.9129783003639657e-07,
      "loss": 1.5096,
      "step": 3162
    },
    {
      "epoch": 0.4770379307744514,
      "grad_norm": 0.24205045088318083,
      "learning_rate": 2.911914931374052e-07,
      "loss": 1.5645,
      "step": 3163
    },
    {
      "epoch": 0.47718874896312496,
      "grad_norm": 0.2478677211449069,
      "learning_rate": 2.910851526029522e-07,
      "loss": 1.547,
      "step": 3164
    },
    {
      "epoch": 0.4773395671517985,
      "grad_norm": 0.2635117483325081,
      "learning_rate": 2.9097880845691397e-07,
      "loss": 1.516,
      "step": 3165
    },
    {
      "epoch": 0.47749038534047206,
      "grad_norm": 0.24323263414281793,
      "learning_rate": 2.9087246072316787e-07,
      "loss": 1.5332,
      "step": 3166
    },
    {
      "epoch": 0.47764120352914563,
      "grad_norm": 0.29558092002313663,
      "learning_rate": 2.9076610942559204e-07,
      "loss": 1.6102,
      "step": 3167
    },
    {
      "epoch": 0.47779202171781915,
      "grad_norm": 0.24736869832808636,
      "learning_rate": 2.9065975458806553e-07,
      "loss": 1.5347,
      "step": 3168
    },
    {
      "epoch": 0.4779428399064927,
      "grad_norm": 0.2646388181945015,
      "learning_rate": 2.90553396234468e-07,
      "loss": 1.5861,
      "step": 3169
    },
    {
      "epoch": 0.4780936580951663,
      "grad_norm": 0.27381585079018494,
      "learning_rate": 2.904470343886801e-07,
      "loss": 1.5137,
      "step": 3170
    },
    {
      "epoch": 0.4782444762838398,
      "grad_norm": 0.2423324557476058,
      "learning_rate": 2.903406690745831e-07,
      "loss": 1.5832,
      "step": 3171
    },
    {
      "epoch": 0.4783952944725134,
      "grad_norm": 0.2517611124533566,
      "learning_rate": 2.9023430031605923e-07,
      "loss": 1.5537,
      "step": 3172
    },
    {
      "epoch": 0.47854611266118696,
      "grad_norm": 0.24805179563772367,
      "learning_rate": 2.901279281369911e-07,
      "loss": 1.5987,
      "step": 3173
    },
    {
      "epoch": 0.4786969308498605,
      "grad_norm": 0.26021823639773467,
      "learning_rate": 2.900215525612626e-07,
      "loss": 1.5365,
      "step": 3174
    },
    {
      "epoch": 0.47884774903853405,
      "grad_norm": 0.24792292288618362,
      "learning_rate": 2.899151736127581e-07,
      "loss": 1.5926,
      "step": 3175
    },
    {
      "epoch": 0.47899856722720763,
      "grad_norm": 0.2415848305018361,
      "learning_rate": 2.898087913153627e-07,
      "loss": 1.5433,
      "step": 3176
    },
    {
      "epoch": 0.47914938541588115,
      "grad_norm": 0.6260731509433154,
      "learning_rate": 2.8970240569296236e-07,
      "loss": 1.5663,
      "step": 3177
    },
    {
      "epoch": 0.4793002036045547,
      "grad_norm": 0.24539307819478479,
      "learning_rate": 2.8959601676944383e-07,
      "loss": 1.5906,
      "step": 3178
    },
    {
      "epoch": 0.47945102179322824,
      "grad_norm": 0.25790011668612844,
      "learning_rate": 2.8948962456869437e-07,
      "loss": 1.6438,
      "step": 3179
    },
    {
      "epoch": 0.4796018399819018,
      "grad_norm": 0.2505504043774616,
      "learning_rate": 2.8938322911460226e-07,
      "loss": 1.5964,
      "step": 3180
    },
    {
      "epoch": 0.4797526581705754,
      "grad_norm": 0.2434342014035931,
      "learning_rate": 2.8927683043105633e-07,
      "loss": 1.5113,
      "step": 3181
    },
    {
      "epoch": 0.4799034763592489,
      "grad_norm": 0.23872019889510285,
      "learning_rate": 2.8917042854194614e-07,
      "loss": 1.6068,
      "step": 3182
    },
    {
      "epoch": 0.4800542945479225,
      "grad_norm": 0.23845824444940553,
      "learning_rate": 2.890640234711621e-07,
      "loss": 1.5804,
      "step": 3183
    },
    {
      "epoch": 0.48020511273659605,
      "grad_norm": 0.3319557403459845,
      "learning_rate": 2.8895761524259515e-07,
      "loss": 1.5272,
      "step": 3184
    },
    {
      "epoch": 0.48035593092526957,
      "grad_norm": 0.24694649494124626,
      "learning_rate": 2.888512038801372e-07,
      "loss": 1.5647,
      "step": 3185
    },
    {
      "epoch": 0.48050674911394314,
      "grad_norm": 0.2461811353531077,
      "learning_rate": 2.8874478940768044e-07,
      "loss": 1.569,
      "step": 3186
    },
    {
      "epoch": 0.4806575673026167,
      "grad_norm": 0.25810955839113636,
      "learning_rate": 2.886383718491182e-07,
      "loss": 1.5937,
      "step": 3187
    },
    {
      "epoch": 0.48080838549129024,
      "grad_norm": 0.2366926661385369,
      "learning_rate": 2.8853195122834435e-07,
      "loss": 1.562,
      "step": 3188
    },
    {
      "epoch": 0.4809592036799638,
      "grad_norm": 0.2504882681866331,
      "learning_rate": 2.8842552756925333e-07,
      "loss": 1.598,
      "step": 3189
    },
    {
      "epoch": 0.4811100218686374,
      "grad_norm": 0.3337958326699907,
      "learning_rate": 2.883191008957404e-07,
      "loss": 1.5665,
      "step": 3190
    },
    {
      "epoch": 0.4812608400573109,
      "grad_norm": 0.23940063487306815,
      "learning_rate": 2.882126712317013e-07,
      "loss": 1.5629,
      "step": 3191
    },
    {
      "epoch": 0.4814116582459845,
      "grad_norm": 0.2541958915384889,
      "learning_rate": 2.8810623860103284e-07,
      "loss": 1.6082,
      "step": 3192
    },
    {
      "epoch": 0.481562476434658,
      "grad_norm": 0.24315381580335652,
      "learning_rate": 2.879998030276321e-07,
      "loss": 1.5353,
      "step": 3193
    },
    {
      "epoch": 0.48171329462333157,
      "grad_norm": 0.244890391464526,
      "learning_rate": 2.8789336453539697e-07,
      "loss": 1.5962,
      "step": 3194
    },
    {
      "epoch": 0.48186411281200514,
      "grad_norm": 0.3054327723434127,
      "learning_rate": 2.87786923148226e-07,
      "loss": 1.5922,
      "step": 3195
    },
    {
      "epoch": 0.48201493100067866,
      "grad_norm": 0.28288167217486754,
      "learning_rate": 2.876804788900184e-07,
      "loss": 1.5471,
      "step": 3196
    },
    {
      "epoch": 0.48216574918935223,
      "grad_norm": 0.24939943317378474,
      "learning_rate": 2.87574031784674e-07,
      "loss": 1.5403,
      "step": 3197
    },
    {
      "epoch": 0.4823165673780258,
      "grad_norm": 0.2631547162360356,
      "learning_rate": 2.8746758185609327e-07,
      "loss": 1.5974,
      "step": 3198
    },
    {
      "epoch": 0.4824673855666993,
      "grad_norm": 10.530002750056518,
      "learning_rate": 2.873611291281773e-07,
      "loss": 1.58,
      "step": 3199
    },
    {
      "epoch": 0.4826182037553729,
      "grad_norm": 0.26162315909216316,
      "learning_rate": 2.872546736248278e-07,
      "loss": 1.4948,
      "step": 3200
    },
    {
      "epoch": 0.4827690219440465,
      "grad_norm": 0.24212148008370052,
      "learning_rate": 2.8714821536994726e-07,
      "loss": 1.5124,
      "step": 3201
    },
    {
      "epoch": 0.48291984013272,
      "grad_norm": 0.2801617978718279,
      "learning_rate": 2.870417543874385e-07,
      "loss": 1.5935,
      "step": 3202
    },
    {
      "epoch": 0.48307065832139356,
      "grad_norm": 0.2387419566852273,
      "learning_rate": 2.869352907012053e-07,
      "loss": 1.5464,
      "step": 3203
    },
    {
      "epoch": 0.48322147651006714,
      "grad_norm": 0.26099184209771586,
      "learning_rate": 2.868288243351516e-07,
      "loss": 1.5626,
      "step": 3204
    },
    {
      "epoch": 0.48337229469874066,
      "grad_norm": 0.25288447335531195,
      "learning_rate": 2.867223553131824e-07,
      "loss": 1.5689,
      "step": 3205
    },
    {
      "epoch": 0.48352311288741423,
      "grad_norm": 0.39672504202397724,
      "learning_rate": 2.866158836592031e-07,
      "loss": 1.5479,
      "step": 3206
    },
    {
      "epoch": 0.48367393107608775,
      "grad_norm": 0.26974266115825624,
      "learning_rate": 2.865094093971195e-07,
      "loss": 1.5265,
      "step": 3207
    },
    {
      "epoch": 0.4838247492647613,
      "grad_norm": 0.24965044998372343,
      "learning_rate": 2.8640293255083836e-07,
      "loss": 1.6274,
      "step": 3208
    },
    {
      "epoch": 0.4839755674534349,
      "grad_norm": 0.24944590452093887,
      "learning_rate": 2.862964531442668e-07,
      "loss": 1.5188,
      "step": 3209
    },
    {
      "epoch": 0.4841263856421084,
      "grad_norm": 0.2565371544538948,
      "learning_rate": 2.8618997120131257e-07,
      "loss": 1.6013,
      "step": 3210
    },
    {
      "epoch": 0.484277203830782,
      "grad_norm": 0.33798667562659546,
      "learning_rate": 2.860834867458838e-07,
      "loss": 1.5895,
      "step": 3211
    },
    {
      "epoch": 0.48442802201945556,
      "grad_norm": 0.2812333198255085,
      "learning_rate": 2.859769998018896e-07,
      "loss": 1.5577,
      "step": 3212
    },
    {
      "epoch": 0.4845788402081291,
      "grad_norm": 0.2455805621250143,
      "learning_rate": 2.8587051039323907e-07,
      "loss": 1.5776,
      "step": 3213
    },
    {
      "epoch": 0.48472965839680265,
      "grad_norm": 0.26926124956469405,
      "learning_rate": 2.8576401854384254e-07,
      "loss": 1.5413,
      "step": 3214
    },
    {
      "epoch": 0.48488047658547623,
      "grad_norm": 0.2591508858213702,
      "learning_rate": 2.856575242776103e-07,
      "loss": 1.6702,
      "step": 3215
    },
    {
      "epoch": 0.48503129477414975,
      "grad_norm": 0.28225120745662957,
      "learning_rate": 2.855510276184535e-07,
      "loss": 1.5548,
      "step": 3216
    },
    {
      "epoch": 0.4851821129628233,
      "grad_norm": 0.25265566608438506,
      "learning_rate": 2.8544452859028366e-07,
      "loss": 1.6464,
      "step": 3217
    },
    {
      "epoch": 0.4853329311514969,
      "grad_norm": 0.2517714020018506,
      "learning_rate": 2.85338027217013e-07,
      "loss": 1.5778,
      "step": 3218
    },
    {
      "epoch": 0.4854837493401704,
      "grad_norm": 0.3158410705985503,
      "learning_rate": 2.852315235225542e-07,
      "loss": 1.5697,
      "step": 3219
    },
    {
      "epoch": 0.485634567528844,
      "grad_norm": 0.24993883929936955,
      "learning_rate": 2.8512501753082034e-07,
      "loss": 1.586,
      "step": 3220
    },
    {
      "epoch": 0.48578538571751756,
      "grad_norm": 0.24254905837014484,
      "learning_rate": 2.850185092657252e-07,
      "loss": 1.5374,
      "step": 3221
    },
    {
      "epoch": 0.4859362039061911,
      "grad_norm": 0.2527740538594686,
      "learning_rate": 2.84911998751183e-07,
      "loss": 1.6025,
      "step": 3222
    },
    {
      "epoch": 0.48608702209486465,
      "grad_norm": 0.25117848213803096,
      "learning_rate": 2.848054860111083e-07,
      "loss": 1.6301,
      "step": 3223
    },
    {
      "epoch": 0.48623784028353817,
      "grad_norm": 0.25643678570548345,
      "learning_rate": 2.8469897106941655e-07,
      "loss": 1.575,
      "step": 3224
    },
    {
      "epoch": 0.48638865847221174,
      "grad_norm": 0.25586334913961334,
      "learning_rate": 2.8459245395002325e-07,
      "loss": 1.5797,
      "step": 3225
    },
    {
      "epoch": 0.4865394766608853,
      "grad_norm": 0.25979692487366346,
      "learning_rate": 2.8448593467684465e-07,
      "loss": 1.5381,
      "step": 3226
    },
    {
      "epoch": 0.48669029484955884,
      "grad_norm": 0.2535970150070719,
      "learning_rate": 2.843794132737975e-07,
      "loss": 1.6001,
      "step": 3227
    },
    {
      "epoch": 0.4868411130382324,
      "grad_norm": 0.2596012177129744,
      "learning_rate": 2.842728897647989e-07,
      "loss": 1.5991,
      "step": 3228
    },
    {
      "epoch": 0.486991931226906,
      "grad_norm": 0.2942803486057231,
      "learning_rate": 2.8416636417376656e-07,
      "loss": 1.5302,
      "step": 3229
    },
    {
      "epoch": 0.4871427494155795,
      "grad_norm": 0.25708300455766936,
      "learning_rate": 2.840598365246184e-07,
      "loss": 1.5565,
      "step": 3230
    },
    {
      "epoch": 0.4872935676042531,
      "grad_norm": 0.8878456230619468,
      "learning_rate": 2.839533068412731e-07,
      "loss": 1.5992,
      "step": 3231
    },
    {
      "epoch": 0.48744438579292665,
      "grad_norm": 0.28130475511678493,
      "learning_rate": 2.8384677514764975e-07,
      "loss": 1.5777,
      "step": 3232
    },
    {
      "epoch": 0.48759520398160017,
      "grad_norm": 0.2500117516063452,
      "learning_rate": 2.8374024146766775e-07,
      "loss": 1.6402,
      "step": 3233
    },
    {
      "epoch": 0.48774602217027374,
      "grad_norm": 0.2514115286658896,
      "learning_rate": 2.836337058252469e-07,
      "loss": 1.5044,
      "step": 3234
    },
    {
      "epoch": 0.4878968403589473,
      "grad_norm": 0.25314414769145527,
      "learning_rate": 2.8352716824430764e-07,
      "loss": 1.5921,
      "step": 3235
    },
    {
      "epoch": 0.48804765854762083,
      "grad_norm": 0.24900053045118872,
      "learning_rate": 2.834206287487708e-07,
      "loss": 1.5998,
      "step": 3236
    },
    {
      "epoch": 0.4881984767362944,
      "grad_norm": 0.2439659282480114,
      "learning_rate": 2.8331408736255765e-07,
      "loss": 1.5143,
      "step": 3237
    },
    {
      "epoch": 0.4883492949249679,
      "grad_norm": 0.25941843445222496,
      "learning_rate": 2.8320754410958964e-07,
      "loss": 1.5795,
      "step": 3238
    },
    {
      "epoch": 0.4885001131136415,
      "grad_norm": 0.41423721903258276,
      "learning_rate": 2.8310099901378897e-07,
      "loss": 1.5428,
      "step": 3239
    },
    {
      "epoch": 0.4886509313023151,
      "grad_norm": 0.33465417546104953,
      "learning_rate": 2.829944520990782e-07,
      "loss": 1.5999,
      "step": 3240
    },
    {
      "epoch": 0.4888017494909886,
      "grad_norm": 0.3828687331908,
      "learning_rate": 2.8288790338937996e-07,
      "loss": 1.5753,
      "step": 3241
    },
    {
      "epoch": 0.48895256767966216,
      "grad_norm": 0.2832946698386233,
      "learning_rate": 2.827813529086178e-07,
      "loss": 1.5806,
      "step": 3242
    },
    {
      "epoch": 0.48910338586833574,
      "grad_norm": 0.30331355605832694,
      "learning_rate": 2.8267480068071525e-07,
      "loss": 1.6383,
      "step": 3243
    },
    {
      "epoch": 0.48925420405700926,
      "grad_norm": 0.25922187834565497,
      "learning_rate": 2.825682467295964e-07,
      "loss": 1.5128,
      "step": 3244
    },
    {
      "epoch": 0.48940502224568283,
      "grad_norm": 0.2527270720867368,
      "learning_rate": 2.8246169107918577e-07,
      "loss": 1.5834,
      "step": 3245
    },
    {
      "epoch": 0.4895558404343564,
      "grad_norm": 0.4260916861411897,
      "learning_rate": 2.823551337534082e-07,
      "loss": 1.5579,
      "step": 3246
    },
    {
      "epoch": 0.4897066586230299,
      "grad_norm": 0.2583051452559372,
      "learning_rate": 2.8224857477618896e-07,
      "loss": 1.5318,
      "step": 3247
    },
    {
      "epoch": 0.4898574768117035,
      "grad_norm": 0.25639898547576445,
      "learning_rate": 2.821420141714535e-07,
      "loss": 1.5402,
      "step": 3248
    },
    {
      "epoch": 0.49000829500037707,
      "grad_norm": 0.2561054616826835,
      "learning_rate": 2.8203545196312783e-07,
      "loss": 1.5973,
      "step": 3249
    },
    {
      "epoch": 0.4901591131890506,
      "grad_norm": 0.24645710856747305,
      "learning_rate": 2.819288881751384e-07,
      "loss": 1.6246,
      "step": 3250
    },
    {
      "epoch": 0.49030993137772416,
      "grad_norm": 0.295121881766085,
      "learning_rate": 2.8182232283141167e-07,
      "loss": 1.5685,
      "step": 3251
    },
    {
      "epoch": 0.4904607495663977,
      "grad_norm": 0.26948819881038233,
      "learning_rate": 2.8171575595587485e-07,
      "loss": 1.6287,
      "step": 3252
    },
    {
      "epoch": 0.49061156775507125,
      "grad_norm": 0.23821034619382012,
      "learning_rate": 2.816091875724552e-07,
      "loss": 1.5834,
      "step": 3253
    },
    {
      "epoch": 0.4907623859437448,
      "grad_norm": 0.23316130818787745,
      "learning_rate": 2.8150261770508043e-07,
      "loss": 1.5489,
      "step": 3254
    },
    {
      "epoch": 0.49091320413241835,
      "grad_norm": 0.2536546870683742,
      "learning_rate": 2.813960463776787e-07,
      "loss": 1.643,
      "step": 3255
    },
    {
      "epoch": 0.4910640223210919,
      "grad_norm": 0.2569158519683004,
      "learning_rate": 2.8128947361417824e-07,
      "loss": 1.6269,
      "step": 3256
    },
    {
      "epoch": 0.4912148405097655,
      "grad_norm": 0.24099574529841028,
      "learning_rate": 2.8118289943850775e-07,
      "loss": 1.6473,
      "step": 3257
    },
    {
      "epoch": 0.491365658698439,
      "grad_norm": 0.303913762146176,
      "learning_rate": 2.8107632387459633e-07,
      "loss": 1.5334,
      "step": 3258
    },
    {
      "epoch": 0.4915164768871126,
      "grad_norm": 0.29823244363900175,
      "learning_rate": 2.809697469463732e-07,
      "loss": 1.5865,
      "step": 3259
    },
    {
      "epoch": 0.49166729507578616,
      "grad_norm": 0.2751509847697237,
      "learning_rate": 2.8086316867776807e-07,
      "loss": 1.5668,
      "step": 3260
    },
    {
      "epoch": 0.4918181132644597,
      "grad_norm": 0.2523827456942733,
      "learning_rate": 2.8075658909271077e-07,
      "loss": 1.65,
      "step": 3261
    },
    {
      "epoch": 0.49196893145313325,
      "grad_norm": 0.3973065228061601,
      "learning_rate": 2.806500082151316e-07,
      "loss": 1.603,
      "step": 3262
    },
    {
      "epoch": 0.4921197496418068,
      "grad_norm": 0.24793582805269612,
      "learning_rate": 2.8054342606896096e-07,
      "loss": 1.534,
      "step": 3263
    },
    {
      "epoch": 0.49227056783048034,
      "grad_norm": 0.25163649978972913,
      "learning_rate": 2.804368426781298e-07,
      "loss": 1.6261,
      "step": 3264
    },
    {
      "epoch": 0.4924213860191539,
      "grad_norm": 0.2520544100110073,
      "learning_rate": 2.8033025806656907e-07,
      "loss": 1.6166,
      "step": 3265
    },
    {
      "epoch": 0.4925722042078275,
      "grad_norm": 0.3305439844269301,
      "learning_rate": 2.802236722582102e-07,
      "loss": 1.5444,
      "step": 3266
    },
    {
      "epoch": 0.492723022396501,
      "grad_norm": 0.2599777522374565,
      "learning_rate": 2.8011708527698476e-07,
      "loss": 1.5614,
      "step": 3267
    },
    {
      "epoch": 0.4928738405851746,
      "grad_norm": 0.25958720907226457,
      "learning_rate": 2.8001049714682467e-07,
      "loss": 1.5006,
      "step": 3268
    },
    {
      "epoch": 0.4930246587738481,
      "grad_norm": 0.254499677063202,
      "learning_rate": 2.79903907891662e-07,
      "loss": 1.5317,
      "step": 3269
    },
    {
      "epoch": 0.4931754769625217,
      "grad_norm": 0.26686274358282275,
      "learning_rate": 2.797973175354293e-07,
      "loss": 1.6463,
      "step": 3270
    },
    {
      "epoch": 0.49332629515119525,
      "grad_norm": 0.2699881617841487,
      "learning_rate": 2.79690726102059e-07,
      "loss": 1.633,
      "step": 3271
    },
    {
      "epoch": 0.49347711333986877,
      "grad_norm": 0.25557733518747255,
      "learning_rate": 2.7958413361548413e-07,
      "loss": 1.5208,
      "step": 3272
    },
    {
      "epoch": 0.49362793152854234,
      "grad_norm": 0.2821977085023352,
      "learning_rate": 2.7947754009963774e-07,
      "loss": 1.6234,
      "step": 3273
    },
    {
      "epoch": 0.4937787497172159,
      "grad_norm": 0.27398607201443076,
      "learning_rate": 2.7937094557845323e-07,
      "loss": 1.6299,
      "step": 3274
    },
    {
      "epoch": 0.49392956790588943,
      "grad_norm": 0.25014826686171554,
      "learning_rate": 2.792643500758641e-07,
      "loss": 1.6191,
      "step": 3275
    },
    {
      "epoch": 0.494080386094563,
      "grad_norm": 0.3087411622057582,
      "learning_rate": 2.7915775361580425e-07,
      "loss": 1.519,
      "step": 3276
    },
    {
      "epoch": 0.4942312042832366,
      "grad_norm": 0.2584795757804951,
      "learning_rate": 2.7905115622220756e-07,
      "loss": 1.6041,
      "step": 3277
    },
    {
      "epoch": 0.4943820224719101,
      "grad_norm": 0.2524777500062278,
      "learning_rate": 2.7894455791900835e-07,
      "loss": 1.5541,
      "step": 3278
    },
    {
      "epoch": 0.4945328406605837,
      "grad_norm": 0.24427578751511272,
      "learning_rate": 2.78837958730141e-07,
      "loss": 1.598,
      "step": 3279
    },
    {
      "epoch": 0.49468365884925725,
      "grad_norm": 0.42420902335941657,
      "learning_rate": 2.787313586795401e-07,
      "loss": 1.562,
      "step": 3280
    },
    {
      "epoch": 0.49483447703793076,
      "grad_norm": 0.2529363586213235,
      "learning_rate": 2.786247577911405e-07,
      "loss": 1.5088,
      "step": 3281
    },
    {
      "epoch": 0.49498529522660434,
      "grad_norm": 0.25403005821628705,
      "learning_rate": 2.785181560888773e-07,
      "loss": 1.6318,
      "step": 3282
    },
    {
      "epoch": 0.49513611341527786,
      "grad_norm": 0.2370884087039444,
      "learning_rate": 2.7841155359668557e-07,
      "loss": 1.5434,
      "step": 3283
    },
    {
      "epoch": 0.49528693160395143,
      "grad_norm": 0.25205579124353483,
      "learning_rate": 2.7830495033850063e-07,
      "loss": 1.5691,
      "step": 3284
    },
    {
      "epoch": 0.495437749792625,
      "grad_norm": 0.25856685206236524,
      "learning_rate": 2.7819834633825813e-07,
      "loss": 1.5557,
      "step": 3285
    },
    {
      "epoch": 0.4955885679812985,
      "grad_norm": 0.24533069937513297,
      "learning_rate": 2.780917416198938e-07,
      "loss": 1.5206,
      "step": 3286
    },
    {
      "epoch": 0.4957393861699721,
      "grad_norm": 0.24305954156589926,
      "learning_rate": 2.779851362073434e-07,
      "loss": 1.5174,
      "step": 3287
    },
    {
      "epoch": 0.49589020435864567,
      "grad_norm": 0.25054942239574796,
      "learning_rate": 2.778785301245429e-07,
      "loss": 1.5329,
      "step": 3288
    },
    {
      "epoch": 0.4960410225473192,
      "grad_norm": 2.0338205742667887,
      "learning_rate": 2.7777192339542867e-07,
      "loss": 1.5391,
      "step": 3289
    },
    {
      "epoch": 0.49619184073599276,
      "grad_norm": 0.25795490828469647,
      "learning_rate": 2.7766531604393684e-07,
      "loss": 1.5867,
      "step": 3290
    },
    {
      "epoch": 0.49634265892466634,
      "grad_norm": 0.2574914292698842,
      "learning_rate": 2.7755870809400396e-07,
      "loss": 1.5698,
      "step": 3291
    },
    {
      "epoch": 0.49649347711333985,
      "grad_norm": 0.7169073199522987,
      "learning_rate": 2.7745209956956653e-07,
      "loss": 1.5594,
      "step": 3292
    },
    {
      "epoch": 0.4966442953020134,
      "grad_norm": 0.26192159719759245,
      "learning_rate": 2.7734549049456145e-07,
      "loss": 1.5836,
      "step": 3293
    },
    {
      "epoch": 0.496795113490687,
      "grad_norm": 0.24903268730487435,
      "learning_rate": 2.772388808929254e-07,
      "loss": 1.511,
      "step": 3294
    },
    {
      "epoch": 0.4969459316793605,
      "grad_norm": 0.2459377727366106,
      "learning_rate": 2.771322707885953e-07,
      "loss": 1.5383,
      "step": 3295
    },
    {
      "epoch": 0.4970967498680341,
      "grad_norm": 0.24091784685121265,
      "learning_rate": 2.770256602055084e-07,
      "loss": 1.5624,
      "step": 3296
    },
    {
      "epoch": 0.4972475680567076,
      "grad_norm": 0.2444409709467392,
      "learning_rate": 2.7691904916760184e-07,
      "loss": 1.635,
      "step": 3297
    },
    {
      "epoch": 0.4973983862453812,
      "grad_norm": 0.25870568500425084,
      "learning_rate": 2.768124376988127e-07,
      "loss": 1.5313,
      "step": 3298
    },
    {
      "epoch": 0.49754920443405476,
      "grad_norm": 0.26148979055400845,
      "learning_rate": 2.767058258230787e-07,
      "loss": 1.5605,
      "step": 3299
    },
    {
      "epoch": 0.4977000226227283,
      "grad_norm": 0.26489712173147045,
      "learning_rate": 2.7659921356433695e-07,
      "loss": 1.5787,
      "step": 3300
    },
    {
      "epoch": 0.49785084081140185,
      "grad_norm": 0.24771529971415504,
      "learning_rate": 2.7649260094652535e-07,
      "loss": 1.5154,
      "step": 3301
    },
    {
      "epoch": 0.4980016590000754,
      "grad_norm": 0.25038035734122427,
      "learning_rate": 2.763859879935812e-07,
      "loss": 1.5992,
      "step": 3302
    },
    {
      "epoch": 0.49815247718874894,
      "grad_norm": 0.2533812623579808,
      "learning_rate": 2.762793747294425e-07,
      "loss": 1.5593,
      "step": 3303
    },
    {
      "epoch": 0.4983032953774225,
      "grad_norm": 0.29352075710368725,
      "learning_rate": 2.7617276117804683e-07,
      "loss": 1.6137,
      "step": 3304
    },
    {
      "epoch": 0.4984541135660961,
      "grad_norm": 0.25649552704037465,
      "learning_rate": 2.760661473633321e-07,
      "loss": 1.5203,
      "step": 3305
    },
    {
      "epoch": 0.4986049317547696,
      "grad_norm": 0.2538748736753708,
      "learning_rate": 2.7595953330923637e-07,
      "loss": 1.5523,
      "step": 3306
    },
    {
      "epoch": 0.4987557499434432,
      "grad_norm": 0.25538783591488673,
      "learning_rate": 2.758529190396974e-07,
      "loss": 1.5785,
      "step": 3307
    },
    {
      "epoch": 0.49890656813211676,
      "grad_norm": 0.2530921350682668,
      "learning_rate": 2.7574630457865316e-07,
      "loss": 1.4433,
      "step": 3308
    },
    {
      "epoch": 0.4990573863207903,
      "grad_norm": 0.25531060239975967,
      "learning_rate": 2.75639689950042e-07,
      "loss": 1.5884,
      "step": 3309
    },
    {
      "epoch": 0.49920820450946385,
      "grad_norm": 0.25898423712370006,
      "learning_rate": 2.7553307517780166e-07,
      "loss": 1.5225,
      "step": 3310
    },
    {
      "epoch": 0.4993590226981374,
      "grad_norm": 0.2469778206164802,
      "learning_rate": 2.754264602858705e-07,
      "loss": 1.5995,
      "step": 3311
    },
    {
      "epoch": 0.49950984088681094,
      "grad_norm": 0.2447786405845865,
      "learning_rate": 2.753198452981866e-07,
      "loss": 1.4748,
      "step": 3312
    },
    {
      "epoch": 0.4996606590754845,
      "grad_norm": 0.2765619723237778,
      "learning_rate": 2.75213230238688e-07,
      "loss": 1.6506,
      "step": 3313
    },
    {
      "epoch": 0.49981147726415803,
      "grad_norm": 0.24721530332690836,
      "learning_rate": 2.751066151313131e-07,
      "loss": 1.5378,
      "step": 3314
    },
    {
      "epoch": 0.4999622954528316,
      "grad_norm": 0.25053059640832,
      "learning_rate": 2.75e-07,
      "loss": 1.6209,
      "step": 3315
    },
    {
      "epoch": 0.5001131136415051,
      "grad_norm": 0.3042993127685436,
      "learning_rate": 2.7489338486868686e-07,
      "loss": 1.5954,
      "step": 3316
    },
    {
      "epoch": 0.5002639318301787,
      "grad_norm": 0.24973666347748574,
      "learning_rate": 2.7478676976131196e-07,
      "loss": 1.5605,
      "step": 3317
    },
    {
      "epoch": 0.5004147500188523,
      "grad_norm": 0.24988153493709508,
      "learning_rate": 2.7468015470181343e-07,
      "loss": 1.5431,
      "step": 3318
    },
    {
      "epoch": 0.5005655682075258,
      "grad_norm": 0.2865156977594693,
      "learning_rate": 2.7457353971412954e-07,
      "loss": 1.5388,
      "step": 3319
    },
    {
      "epoch": 0.5007163863961994,
      "grad_norm": 0.2527759398124433,
      "learning_rate": 2.7446692482219837e-07,
      "loss": 1.6225,
      "step": 3320
    },
    {
      "epoch": 0.5008672045848729,
      "grad_norm": 0.2693109111241276,
      "learning_rate": 2.7436031004995805e-07,
      "loss": 1.4975,
      "step": 3321
    },
    {
      "epoch": 0.5010180227735465,
      "grad_norm": 0.27130862309093945,
      "learning_rate": 2.742536954213468e-07,
      "loss": 1.5414,
      "step": 3322
    },
    {
      "epoch": 0.50116884096222,
      "grad_norm": 0.24706973863352966,
      "learning_rate": 2.7414708096030264e-07,
      "loss": 1.5393,
      "step": 3323
    },
    {
      "epoch": 0.5013196591508936,
      "grad_norm": 0.26812242254342294,
      "learning_rate": 2.7404046669076366e-07,
      "loss": 1.5682,
      "step": 3324
    },
    {
      "epoch": 0.5014704773395672,
      "grad_norm": 0.31047841476741905,
      "learning_rate": 2.7393385263666783e-07,
      "loss": 1.5423,
      "step": 3325
    },
    {
      "epoch": 0.5016212955282408,
      "grad_norm": 0.24735751950021653,
      "learning_rate": 2.738272388219532e-07,
      "loss": 1.5263,
      "step": 3326
    },
    {
      "epoch": 0.5017721137169142,
      "grad_norm": 0.25692232040787155,
      "learning_rate": 2.7372062527055755e-07,
      "loss": 1.5895,
      "step": 3327
    },
    {
      "epoch": 0.5019229319055878,
      "grad_norm": 0.2418566848114825,
      "learning_rate": 2.7361401200641884e-07,
      "loss": 1.5592,
      "step": 3328
    },
    {
      "epoch": 0.5020737500942614,
      "grad_norm": 0.23863554158574807,
      "learning_rate": 2.735073990534747e-07,
      "loss": 1.5274,
      "step": 3329
    },
    {
      "epoch": 0.5022245682829349,
      "grad_norm": 0.2982027483839481,
      "learning_rate": 2.734007864356631e-07,
      "loss": 1.6451,
      "step": 3330
    },
    {
      "epoch": 0.5023753864716085,
      "grad_norm": 0.24960726137909803,
      "learning_rate": 2.7329417417692134e-07,
      "loss": 1.5391,
      "step": 3331
    },
    {
      "epoch": 0.5025262046602821,
      "grad_norm": 0.2545436508039039,
      "learning_rate": 2.7318756230118726e-07,
      "loss": 1.649,
      "step": 3332
    },
    {
      "epoch": 0.5026770228489555,
      "grad_norm": 0.2513105990902941,
      "learning_rate": 2.7308095083239824e-07,
      "loss": 1.58,
      "step": 3333
    },
    {
      "epoch": 0.5028278410376291,
      "grad_norm": 0.24609194731011771,
      "learning_rate": 2.729743397944916e-07,
      "loss": 1.5532,
      "step": 3334
    },
    {
      "epoch": 0.5029786592263027,
      "grad_norm": 0.2943735804970433,
      "learning_rate": 2.7286772921140466e-07,
      "loss": 1.5885,
      "step": 3335
    },
    {
      "epoch": 0.5031294774149763,
      "grad_norm": 0.25308883997269344,
      "learning_rate": 2.7276111910707465e-07,
      "loss": 1.5336,
      "step": 3336
    },
    {
      "epoch": 0.5032802956036498,
      "grad_norm": 0.39632265194069904,
      "learning_rate": 2.726545095054386e-07,
      "loss": 1.5418,
      "step": 3337
    },
    {
      "epoch": 0.5034311137923233,
      "grad_norm": 0.2495909049144747,
      "learning_rate": 2.7254790043043344e-07,
      "loss": 1.5832,
      "step": 3338
    },
    {
      "epoch": 0.5035819319809969,
      "grad_norm": 0.254632877714315,
      "learning_rate": 2.7244129190599607e-07,
      "loss": 1.6509,
      "step": 3339
    },
    {
      "epoch": 0.5037327501696705,
      "grad_norm": 0.2748382970830817,
      "learning_rate": 2.723346839560632e-07,
      "loss": 1.6096,
      "step": 3340
    },
    {
      "epoch": 0.503883568358344,
      "grad_norm": 0.2699005878780138,
      "learning_rate": 2.7222807660457136e-07,
      "loss": 1.5777,
      "step": 3341
    },
    {
      "epoch": 0.5040343865470176,
      "grad_norm": 0.2553144305506177,
      "learning_rate": 2.721214698754571e-07,
      "loss": 1.5563,
      "step": 3342
    },
    {
      "epoch": 0.5041852047356912,
      "grad_norm": 0.2985671999866322,
      "learning_rate": 2.7201486379265663e-07,
      "loss": 1.567,
      "step": 3343
    },
    {
      "epoch": 0.5043360229243646,
      "grad_norm": 0.23768594996157277,
      "learning_rate": 2.7190825838010623e-07,
      "loss": 1.4942,
      "step": 3344
    },
    {
      "epoch": 0.5044868411130382,
      "grad_norm": 0.26881074903449925,
      "learning_rate": 2.7180165366174184e-07,
      "loss": 1.5781,
      "step": 3345
    },
    {
      "epoch": 0.5046376593017118,
      "grad_norm": 0.4033794521730484,
      "learning_rate": 2.7169504966149934e-07,
      "loss": 1.4971,
      "step": 3346
    },
    {
      "epoch": 0.5047884774903854,
      "grad_norm": 0.25578761257228577,
      "learning_rate": 2.7158844640331446e-07,
      "loss": 1.5625,
      "step": 3347
    },
    {
      "epoch": 0.5049392956790589,
      "grad_norm": 0.25244155440042493,
      "learning_rate": 2.714818439111227e-07,
      "loss": 1.6181,
      "step": 3348
    },
    {
      "epoch": 0.5050901138677324,
      "grad_norm": 0.34838110431199665,
      "learning_rate": 2.713752422088595e-07,
      "loss": 1.5672,
      "step": 3349
    },
    {
      "epoch": 0.505240932056406,
      "grad_norm": 0.29123625682680754,
      "learning_rate": 2.7126864132045995e-07,
      "loss": 1.5912,
      "step": 3350
    },
    {
      "epoch": 0.5053917502450795,
      "grad_norm": 0.2535329957922546,
      "learning_rate": 2.711620412698591e-07,
      "loss": 1.5847,
      "step": 3351
    },
    {
      "epoch": 0.5055425684337531,
      "grad_norm": 0.25675943816014185,
      "learning_rate": 2.710554420809917e-07,
      "loss": 1.5175,
      "step": 3352
    },
    {
      "epoch": 0.5056933866224267,
      "grad_norm": 0.2483481980707172,
      "learning_rate": 2.7094884377779246e-07,
      "loss": 1.6137,
      "step": 3353
    },
    {
      "epoch": 0.5058442048111003,
      "grad_norm": 0.2805203243287897,
      "learning_rate": 2.708422463841958e-07,
      "loss": 1.495,
      "step": 3354
    },
    {
      "epoch": 0.5059950229997737,
      "grad_norm": 0.24657785511710636,
      "learning_rate": 2.707356499241359e-07,
      "loss": 1.5425,
      "step": 3355
    },
    {
      "epoch": 0.5061458411884473,
      "grad_norm": 0.2804663396090381,
      "learning_rate": 2.706290544215468e-07,
      "loss": 1.6177,
      "step": 3356
    },
    {
      "epoch": 0.5062966593771209,
      "grad_norm": 0.2519727641215625,
      "learning_rate": 2.705224599003622e-07,
      "loss": 1.5697,
      "step": 3357
    },
    {
      "epoch": 0.5064474775657944,
      "grad_norm": 0.24965340619312307,
      "learning_rate": 2.704158663845159e-07,
      "loss": 1.4687,
      "step": 3358
    },
    {
      "epoch": 0.506598295754468,
      "grad_norm": 0.2572141299558103,
      "learning_rate": 2.70309273897941e-07,
      "loss": 1.5834,
      "step": 3359
    },
    {
      "epoch": 0.5067491139431416,
      "grad_norm": 0.25474124648079854,
      "learning_rate": 2.7020268246457074e-07,
      "loss": 1.5704,
      "step": 3360
    },
    {
      "epoch": 0.5068999321318151,
      "grad_norm": 0.25162282375325096,
      "learning_rate": 2.700960921083379e-07,
      "loss": 1.5595,
      "step": 3361
    },
    {
      "epoch": 0.5070507503204886,
      "grad_norm": 0.32619383144972003,
      "learning_rate": 2.6998950285317536e-07,
      "loss": 1.5805,
      "step": 3362
    },
    {
      "epoch": 0.5072015685091622,
      "grad_norm": 0.25540840064293446,
      "learning_rate": 2.6988291472301527e-07,
      "loss": 1.5823,
      "step": 3363
    },
    {
      "epoch": 0.5073523866978358,
      "grad_norm": 0.24902918521605308,
      "learning_rate": 2.6977632774178983e-07,
      "loss": 1.575,
      "step": 3364
    },
    {
      "epoch": 0.5075032048865094,
      "grad_norm": 0.243927216521824,
      "learning_rate": 2.6966974193343096e-07,
      "loss": 1.5108,
      "step": 3365
    },
    {
      "epoch": 0.5076540230751828,
      "grad_norm": 0.2359957641037577,
      "learning_rate": 2.695631573218703e-07,
      "loss": 1.5134,
      "step": 3366
    },
    {
      "epoch": 0.5078048412638564,
      "grad_norm": 0.2683406389292298,
      "learning_rate": 2.694565739310391e-07,
      "loss": 1.5073,
      "step": 3367
    },
    {
      "epoch": 0.50795565945253,
      "grad_norm": 0.2486889359494454,
      "learning_rate": 2.693499917848685e-07,
      "loss": 1.4811,
      "step": 3368
    },
    {
      "epoch": 0.5081064776412035,
      "grad_norm": 0.24544466385111835,
      "learning_rate": 2.692434109072893e-07,
      "loss": 1.5668,
      "step": 3369
    },
    {
      "epoch": 0.5082572958298771,
      "grad_norm": 0.2353367859389473,
      "learning_rate": 2.69136831322232e-07,
      "loss": 1.547,
      "step": 3370
    },
    {
      "epoch": 0.5084081140185507,
      "grad_norm": 0.24818901527921186,
      "learning_rate": 2.6903025305362686e-07,
      "loss": 1.5081,
      "step": 3371
    },
    {
      "epoch": 0.5085589322072241,
      "grad_norm": 0.3365507853358683,
      "learning_rate": 2.6892367612540375e-07,
      "loss": 1.5223,
      "step": 3372
    },
    {
      "epoch": 0.5087097503958977,
      "grad_norm": 0.2551285914497227,
      "learning_rate": 2.688171005614923e-07,
      "loss": 1.5723,
      "step": 3373
    },
    {
      "epoch": 0.5088605685845713,
      "grad_norm": 0.28271133388757014,
      "learning_rate": 2.687105263858218e-07,
      "loss": 1.5875,
      "step": 3374
    },
    {
      "epoch": 0.5090113867732449,
      "grad_norm": 0.24363473847958142,
      "learning_rate": 2.6860395362232136e-07,
      "loss": 1.5696,
      "step": 3375
    },
    {
      "epoch": 0.5091622049619184,
      "grad_norm": 0.246165245140535,
      "learning_rate": 2.684973822949196e-07,
      "loss": 1.6845,
      "step": 3376
    },
    {
      "epoch": 0.509313023150592,
      "grad_norm": 0.2526657502202713,
      "learning_rate": 2.683908124275449e-07,
      "loss": 1.5608,
      "step": 3377
    },
    {
      "epoch": 0.5094638413392655,
      "grad_norm": 0.2822124747596741,
      "learning_rate": 2.682842440441252e-07,
      "loss": 1.6233,
      "step": 3378
    },
    {
      "epoch": 0.509614659527939,
      "grad_norm": 0.27211605527560906,
      "learning_rate": 2.681776771685884e-07,
      "loss": 1.557,
      "step": 3379
    },
    {
      "epoch": 0.5097654777166126,
      "grad_norm": 0.25618395269705335,
      "learning_rate": 2.680711118248617e-07,
      "loss": 1.5103,
      "step": 3380
    },
    {
      "epoch": 0.5099162959052862,
      "grad_norm": 0.25714012472043773,
      "learning_rate": 2.679645480368722e-07,
      "loss": 1.562,
      "step": 3381
    },
    {
      "epoch": 0.5100671140939598,
      "grad_norm": 0.2551121867483003,
      "learning_rate": 2.678579858285466e-07,
      "loss": 1.5021,
      "step": 3382
    },
    {
      "epoch": 0.5102179322826332,
      "grad_norm": 0.279795512750936,
      "learning_rate": 2.677514252238111e-07,
      "loss": 1.5977,
      "step": 3383
    },
    {
      "epoch": 0.5103687504713068,
      "grad_norm": 0.2667401842863274,
      "learning_rate": 2.6764486624659183e-07,
      "loss": 1.4919,
      "step": 3384
    },
    {
      "epoch": 0.5105195686599804,
      "grad_norm": 0.24870196970524328,
      "learning_rate": 2.6753830892081426e-07,
      "loss": 1.5754,
      "step": 3385
    },
    {
      "epoch": 0.510670386848654,
      "grad_norm": 0.2638389192651744,
      "learning_rate": 2.6743175327040363e-07,
      "loss": 1.5756,
      "step": 3386
    },
    {
      "epoch": 0.5108212050373275,
      "grad_norm": 0.4072165688694835,
      "learning_rate": 2.6732519931928483e-07,
      "loss": 1.6453,
      "step": 3387
    },
    {
      "epoch": 0.5109720232260011,
      "grad_norm": 0.2422848523750943,
      "learning_rate": 2.6721864709138225e-07,
      "loss": 1.4672,
      "step": 3388
    },
    {
      "epoch": 0.5111228414146746,
      "grad_norm": 0.40391218423175,
      "learning_rate": 2.6711209661062006e-07,
      "loss": 1.5901,
      "step": 3389
    },
    {
      "epoch": 0.5112736596033481,
      "grad_norm": 0.30394247097694993,
      "learning_rate": 2.6700554790092187e-07,
      "loss": 1.5573,
      "step": 3390
    },
    {
      "epoch": 0.5114244777920217,
      "grad_norm": 0.42617749918711456,
      "learning_rate": 2.6689900098621106e-07,
      "loss": 1.579,
      "step": 3391
    },
    {
      "epoch": 0.5115752959806953,
      "grad_norm": 0.2804842346156845,
      "learning_rate": 2.667924558904104e-07,
      "loss": 1.5556,
      "step": 3392
    },
    {
      "epoch": 0.5117261141693689,
      "grad_norm": 0.248293866650627,
      "learning_rate": 2.6668591263744243e-07,
      "loss": 1.625,
      "step": 3393
    },
    {
      "epoch": 0.5118769323580423,
      "grad_norm": 0.25266486877579375,
      "learning_rate": 2.665793712512293e-07,
      "loss": 1.5812,
      "step": 3394
    },
    {
      "epoch": 0.5120277505467159,
      "grad_norm": 0.25281005618560276,
      "learning_rate": 2.664728317556924e-07,
      "loss": 1.5527,
      "step": 3395
    },
    {
      "epoch": 0.5121785687353895,
      "grad_norm": 0.26245082973756306,
      "learning_rate": 2.6636629417475314e-07,
      "loss": 1.6174,
      "step": 3396
    },
    {
      "epoch": 0.512329386924063,
      "grad_norm": 0.24176224348249375,
      "learning_rate": 2.662597585323324e-07,
      "loss": 1.5474,
      "step": 3397
    },
    {
      "epoch": 0.5124802051127366,
      "grad_norm": 0.27145047565218094,
      "learning_rate": 2.6615322485235023e-07,
      "loss": 1.5874,
      "step": 3398
    },
    {
      "epoch": 0.5126310233014102,
      "grad_norm": 0.2457872345203949,
      "learning_rate": 2.660466931587269e-07,
      "loss": 1.5548,
      "step": 3399
    },
    {
      "epoch": 0.5127818414900837,
      "grad_norm": 0.2561704235833319,
      "learning_rate": 2.659401634753817e-07,
      "loss": 1.5722,
      "step": 3400
    },
    {
      "epoch": 0.5129326596787572,
      "grad_norm": 0.257556030514651,
      "learning_rate": 2.658336358262335e-07,
      "loss": 1.5961,
      "step": 3401
    },
    {
      "epoch": 0.5130834778674308,
      "grad_norm": 0.2524081368312293,
      "learning_rate": 2.6572711023520113e-07,
      "loss": 1.6134,
      "step": 3402
    },
    {
      "epoch": 0.5132342960561044,
      "grad_norm": 0.2587914057008669,
      "learning_rate": 2.656205867262025e-07,
      "loss": 1.6208,
      "step": 3403
    },
    {
      "epoch": 0.513385114244778,
      "grad_norm": 0.3278007542839679,
      "learning_rate": 2.6551406532315533e-07,
      "loss": 1.6417,
      "step": 3404
    },
    {
      "epoch": 0.5135359324334515,
      "grad_norm": 0.24718450490251231,
      "learning_rate": 2.6540754604997683e-07,
      "loss": 1.4789,
      "step": 3405
    },
    {
      "epoch": 0.513686750622125,
      "grad_norm": 0.2589284027537417,
      "learning_rate": 2.6530102893058353e-07,
      "loss": 1.5179,
      "step": 3406
    },
    {
      "epoch": 0.5138375688107986,
      "grad_norm": 0.2747977819875336,
      "learning_rate": 2.6519451398889177e-07,
      "loss": 1.5357,
      "step": 3407
    },
    {
      "epoch": 0.5139883869994721,
      "grad_norm": 0.24735369688127637,
      "learning_rate": 2.650880012488171e-07,
      "loss": 1.5487,
      "step": 3408
    },
    {
      "epoch": 0.5141392051881457,
      "grad_norm": 0.24664274099541278,
      "learning_rate": 2.6498149073427485e-07,
      "loss": 1.5351,
      "step": 3409
    },
    {
      "epoch": 0.5142900233768193,
      "grad_norm": 0.2590613272316827,
      "learning_rate": 2.6487498246917974e-07,
      "loss": 1.5629,
      "step": 3410
    },
    {
      "epoch": 0.5144408415654927,
      "grad_norm": 0.3705828058823599,
      "learning_rate": 2.647684764774459e-07,
      "loss": 1.5627,
      "step": 3411
    },
    {
      "epoch": 0.5145916597541663,
      "grad_norm": 0.24319145857519242,
      "learning_rate": 2.6466197278298703e-07,
      "loss": 1.4684,
      "step": 3412
    },
    {
      "epoch": 0.5147424779428399,
      "grad_norm": 0.2472923219641013,
      "learning_rate": 2.645554714097164e-07,
      "loss": 1.5518,
      "step": 3413
    },
    {
      "epoch": 0.5148932961315135,
      "grad_norm": 0.290010629764668,
      "learning_rate": 2.644489723815466e-07,
      "loss": 1.5554,
      "step": 3414
    },
    {
      "epoch": 0.515044114320187,
      "grad_norm": 0.48259758803379066,
      "learning_rate": 2.643424757223898e-07,
      "loss": 1.5199,
      "step": 3415
    },
    {
      "epoch": 0.5151949325088606,
      "grad_norm": 0.2700103396991272,
      "learning_rate": 2.6423598145615755e-07,
      "loss": 1.5404,
      "step": 3416
    },
    {
      "epoch": 0.5153457506975341,
      "grad_norm": 0.27260193450195136,
      "learning_rate": 2.641294896067609e-07,
      "loss": 1.5869,
      "step": 3417
    },
    {
      "epoch": 0.5154965688862077,
      "grad_norm": 0.25736491966765584,
      "learning_rate": 2.6402300019811056e-07,
      "loss": 1.4908,
      "step": 3418
    },
    {
      "epoch": 0.5156473870748812,
      "grad_norm": 0.2617846338780766,
      "learning_rate": 2.639165132541162e-07,
      "loss": 1.4805,
      "step": 3419
    },
    {
      "epoch": 0.5157982052635548,
      "grad_norm": 0.25187905256722865,
      "learning_rate": 2.6381002879868746e-07,
      "loss": 1.583,
      "step": 3420
    },
    {
      "epoch": 0.5159490234522284,
      "grad_norm": 0.24428837350616164,
      "learning_rate": 2.6370354685573315e-07,
      "loss": 1.4823,
      "step": 3421
    },
    {
      "epoch": 0.516099841640902,
      "grad_norm": 0.35219293961394166,
      "learning_rate": 2.635970674491616e-07,
      "loss": 1.5976,
      "step": 3422
    },
    {
      "epoch": 0.5162506598295754,
      "grad_norm": 0.24748325128883544,
      "learning_rate": 2.6349059060288047e-07,
      "loss": 1.5116,
      "step": 3423
    },
    {
      "epoch": 0.516401478018249,
      "grad_norm": 0.26782624643815234,
      "learning_rate": 2.6338411634079695e-07,
      "loss": 1.6013,
      "step": 3424
    },
    {
      "epoch": 0.5165522962069226,
      "grad_norm": 0.2516196167397587,
      "learning_rate": 2.6327764468681755e-07,
      "loss": 1.5371,
      "step": 3425
    },
    {
      "epoch": 0.5167031143955961,
      "grad_norm": 0.37290749259523764,
      "learning_rate": 2.6317117566484836e-07,
      "loss": 1.4911,
      "step": 3426
    },
    {
      "epoch": 0.5168539325842697,
      "grad_norm": 0.251509966071061,
      "learning_rate": 2.6306470929879475e-07,
      "loss": 1.5656,
      "step": 3427
    },
    {
      "epoch": 0.5170047507729432,
      "grad_norm": 0.25849330500463785,
      "learning_rate": 2.629582456125615e-07,
      "loss": 1.5666,
      "step": 3428
    },
    {
      "epoch": 0.5171555689616167,
      "grad_norm": 0.24393511130890536,
      "learning_rate": 2.628517846300527e-07,
      "loss": 1.5653,
      "step": 3429
    },
    {
      "epoch": 0.5173063871502903,
      "grad_norm": 0.24229616697656609,
      "learning_rate": 2.6274532637517215e-07,
      "loss": 1.4901,
      "step": 3430
    },
    {
      "epoch": 0.5174572053389639,
      "grad_norm": 0.258315777973691,
      "learning_rate": 2.626388708718227e-07,
      "loss": 1.5753,
      "step": 3431
    },
    {
      "epoch": 0.5176080235276375,
      "grad_norm": 0.3039069236998845,
      "learning_rate": 2.6253241814390676e-07,
      "loss": 1.6208,
      "step": 3432
    },
    {
      "epoch": 0.517758841716311,
      "grad_norm": 0.2835486242320549,
      "learning_rate": 2.62425968215326e-07,
      "loss": 1.6249,
      "step": 3433
    },
    {
      "epoch": 0.5179096599049845,
      "grad_norm": 0.24902980508283476,
      "learning_rate": 2.6231952110998156e-07,
      "loss": 1.6022,
      "step": 3434
    },
    {
      "epoch": 0.5180604780936581,
      "grad_norm": 0.2492393031975446,
      "learning_rate": 2.6221307685177395e-07,
      "loss": 1.5974,
      "step": 3435
    },
    {
      "epoch": 0.5182112962823316,
      "grad_norm": 0.23794462105076858,
      "learning_rate": 2.62106635464603e-07,
      "loss": 1.5895,
      "step": 3436
    },
    {
      "epoch": 0.5183621144710052,
      "grad_norm": 0.4444198515386016,
      "learning_rate": 2.6200019697236784e-07,
      "loss": 1.5255,
      "step": 3437
    },
    {
      "epoch": 0.5185129326596788,
      "grad_norm": 0.253090569717866,
      "learning_rate": 2.6189376139896714e-07,
      "loss": 1.5699,
      "step": 3438
    },
    {
      "epoch": 0.5186637508483523,
      "grad_norm": 0.2393039961899078,
      "learning_rate": 2.617873287682986e-07,
      "loss": 1.5549,
      "step": 3439
    },
    {
      "epoch": 0.5188145690370258,
      "grad_norm": 0.25196990371497024,
      "learning_rate": 2.616808991042596e-07,
      "loss": 1.5806,
      "step": 3440
    },
    {
      "epoch": 0.5189653872256994,
      "grad_norm": 0.2750144429593545,
      "learning_rate": 2.6157447243074664e-07,
      "loss": 1.6465,
      "step": 3441
    },
    {
      "epoch": 0.519116205414373,
      "grad_norm": 0.272366716894923,
      "learning_rate": 2.6146804877165563e-07,
      "loss": 1.491,
      "step": 3442
    },
    {
      "epoch": 0.5192670236030466,
      "grad_norm": 0.31214694654172437,
      "learning_rate": 2.6136162815088175e-07,
      "loss": 1.5951,
      "step": 3443
    },
    {
      "epoch": 0.5194178417917201,
      "grad_norm": 0.2621271805612479,
      "learning_rate": 2.6125521059231954e-07,
      "loss": 1.5426,
      "step": 3444
    },
    {
      "epoch": 0.5195686599803936,
      "grad_norm": 0.271740818612056,
      "learning_rate": 2.6114879611986284e-07,
      "loss": 1.5348,
      "step": 3445
    },
    {
      "epoch": 0.5197194781690672,
      "grad_norm": 0.23973127661834764,
      "learning_rate": 2.610423847574048e-07,
      "loss": 1.5337,
      "step": 3446
    },
    {
      "epoch": 0.5198702963577407,
      "grad_norm": 0.26665725866133855,
      "learning_rate": 2.6093597652883784e-07,
      "loss": 1.6278,
      "step": 3447
    },
    {
      "epoch": 0.5200211145464143,
      "grad_norm": 0.26003632228083834,
      "learning_rate": 2.6082957145805384e-07,
      "loss": 1.6183,
      "step": 3448
    },
    {
      "epoch": 0.5201719327350879,
      "grad_norm": 0.259081350793201,
      "learning_rate": 2.607231695689437e-07,
      "loss": 1.5933,
      "step": 3449
    },
    {
      "epoch": 0.5203227509237615,
      "grad_norm": 0.2704935949422631,
      "learning_rate": 2.6061677088539777e-07,
      "loss": 1.5816,
      "step": 3450
    },
    {
      "epoch": 0.5204735691124349,
      "grad_norm": 0.5958003619919746,
      "learning_rate": 2.605103754313056e-07,
      "loss": 1.5625,
      "step": 3451
    },
    {
      "epoch": 0.5206243873011085,
      "grad_norm": 0.25068661919329216,
      "learning_rate": 2.604039832305562e-07,
      "loss": 1.565,
      "step": 3452
    },
    {
      "epoch": 0.5207752054897821,
      "grad_norm": 0.23637743445577852,
      "learning_rate": 2.602975943070376e-07,
      "loss": 1.5447,
      "step": 3453
    },
    {
      "epoch": 0.5209260236784556,
      "grad_norm": 0.2771298053934381,
      "learning_rate": 2.601912086846373e-07,
      "loss": 1.5708,
      "step": 3454
    },
    {
      "epoch": 0.5210768418671292,
      "grad_norm": 0.2645802923684436,
      "learning_rate": 2.600848263872419e-07,
      "loss": 1.6231,
      "step": 3455
    },
    {
      "epoch": 0.5212276600558027,
      "grad_norm": 0.2570476354424804,
      "learning_rate": 2.5997844743873735e-07,
      "loss": 1.5465,
      "step": 3456
    },
    {
      "epoch": 0.5213784782444763,
      "grad_norm": 0.24647483848346644,
      "learning_rate": 2.5987207186300884e-07,
      "loss": 1.4589,
      "step": 3457
    },
    {
      "epoch": 0.5215292964331498,
      "grad_norm": 0.2492627062540946,
      "learning_rate": 2.597656996839408e-07,
      "loss": 1.5618,
      "step": 3458
    },
    {
      "epoch": 0.5216801146218234,
      "grad_norm": 0.24356023524839673,
      "learning_rate": 2.596593309254168e-07,
      "loss": 1.5075,
      "step": 3459
    },
    {
      "epoch": 0.521830932810497,
      "grad_norm": 0.2570979238960611,
      "learning_rate": 2.5955296561131985e-07,
      "loss": 1.5627,
      "step": 3460
    },
    {
      "epoch": 0.5219817509991705,
      "grad_norm": 0.2737835754830638,
      "learning_rate": 2.5944660376553194e-07,
      "loss": 1.5675,
      "step": 3461
    },
    {
      "epoch": 0.522132569187844,
      "grad_norm": 0.25765438745021246,
      "learning_rate": 2.5934024541193444e-07,
      "loss": 1.525,
      "step": 3462
    },
    {
      "epoch": 0.5222833873765176,
      "grad_norm": 0.29237538114546824,
      "learning_rate": 2.5923389057440793e-07,
      "loss": 1.5651,
      "step": 3463
    },
    {
      "epoch": 0.5224342055651912,
      "grad_norm": 0.24218855295460975,
      "learning_rate": 2.591275392768321e-07,
      "loss": 1.5282,
      "step": 3464
    },
    {
      "epoch": 0.5225850237538647,
      "grad_norm": 0.31804114366062597,
      "learning_rate": 2.5902119154308606e-07,
      "loss": 1.6204,
      "step": 3465
    },
    {
      "epoch": 0.5227358419425383,
      "grad_norm": 0.26055067066311277,
      "learning_rate": 2.589148473970478e-07,
      "loss": 1.5551,
      "step": 3466
    },
    {
      "epoch": 0.5228866601312119,
      "grad_norm": 0.25097316862996666,
      "learning_rate": 2.5880850686259475e-07,
      "loss": 1.5746,
      "step": 3467
    },
    {
      "epoch": 0.5230374783198853,
      "grad_norm": 0.2506066181626998,
      "learning_rate": 2.587021699636035e-07,
      "loss": 1.5946,
      "step": 3468
    },
    {
      "epoch": 0.5231882965085589,
      "grad_norm": 0.2522370225688385,
      "learning_rate": 2.5859583672394967e-07,
      "loss": 1.5721,
      "step": 3469
    },
    {
      "epoch": 0.5233391146972325,
      "grad_norm": 0.3068255943113725,
      "learning_rate": 2.584895071675083e-07,
      "loss": 1.5739,
      "step": 3470
    },
    {
      "epoch": 0.5234899328859061,
      "grad_norm": 0.23850945286400205,
      "learning_rate": 2.583831813181533e-07,
      "loss": 1.5051,
      "step": 3471
    },
    {
      "epoch": 0.5236407510745796,
      "grad_norm": 0.2584496902340964,
      "learning_rate": 2.5827685919975816e-07,
      "loss": 1.5743,
      "step": 3472
    },
    {
      "epoch": 0.5237915692632531,
      "grad_norm": 0.25513878483580066,
      "learning_rate": 2.581705408361951e-07,
      "loss": 1.5676,
      "step": 3473
    },
    {
      "epoch": 0.5239423874519267,
      "grad_norm": 0.26447952862161583,
      "learning_rate": 2.580642262513358e-07,
      "loss": 1.5518,
      "step": 3474
    },
    {
      "epoch": 0.5240932056406002,
      "grad_norm": 0.2662577459383339,
      "learning_rate": 2.5795791546905087e-07,
      "loss": 1.5673,
      "step": 3475
    },
    {
      "epoch": 0.5242440238292738,
      "grad_norm": 0.32092647569567506,
      "learning_rate": 2.578516085132103e-07,
      "loss": 1.5428,
      "step": 3476
    },
    {
      "epoch": 0.5243948420179474,
      "grad_norm": 0.2498482740773882,
      "learning_rate": 2.5774530540768294e-07,
      "loss": 1.5802,
      "step": 3477
    },
    {
      "epoch": 0.524545660206621,
      "grad_norm": 0.2546959664545933,
      "learning_rate": 2.576390061763371e-07,
      "loss": 1.5317,
      "step": 3478
    },
    {
      "epoch": 0.5246964783952944,
      "grad_norm": 0.25001127382963,
      "learning_rate": 2.5753271084304e-07,
      "loss": 1.5412,
      "step": 3479
    },
    {
      "epoch": 0.524847296583968,
      "grad_norm": 0.25093090770390275,
      "learning_rate": 2.574264194316579e-07,
      "loss": 1.5312,
      "step": 3480
    },
    {
      "epoch": 0.5249981147726416,
      "grad_norm": 0.25287177921984955,
      "learning_rate": 2.573201319660566e-07,
      "loss": 1.5148,
      "step": 3481
    },
    {
      "epoch": 0.5251489329613152,
      "grad_norm": 0.24869925307935759,
      "learning_rate": 2.572138484701005e-07,
      "loss": 1.562,
      "step": 3482
    },
    {
      "epoch": 0.5252997511499887,
      "grad_norm": 0.2611394563145696,
      "learning_rate": 2.5710756896765337e-07,
      "loss": 1.5959,
      "step": 3483
    },
    {
      "epoch": 0.5254505693386622,
      "grad_norm": 0.24753523535044233,
      "learning_rate": 2.5700129348257814e-07,
      "loss": 1.5496,
      "step": 3484
    },
    {
      "epoch": 0.5256013875273358,
      "grad_norm": 0.5874019514703901,
      "learning_rate": 2.568950220387367e-07,
      "loss": 1.5877,
      "step": 3485
    },
    {
      "epoch": 0.5257522057160093,
      "grad_norm": 0.24725630624083114,
      "learning_rate": 2.5678875465999e-07,
      "loss": 1.5123,
      "step": 3486
    },
    {
      "epoch": 0.5259030239046829,
      "grad_norm": 0.25546325957175015,
      "learning_rate": 2.5668249137019837e-07,
      "loss": 1.6225,
      "step": 3487
    },
    {
      "epoch": 0.5260538420933565,
      "grad_norm": 0.2771144622831777,
      "learning_rate": 2.5657623219322084e-07,
      "loss": 1.5953,
      "step": 3488
    },
    {
      "epoch": 0.5262046602820301,
      "grad_norm": 0.25916392737960275,
      "learning_rate": 2.564699771529158e-07,
      "loss": 1.5079,
      "step": 3489
    },
    {
      "epoch": 0.5263554784707035,
      "grad_norm": 0.2527559440958871,
      "learning_rate": 2.5636372627314046e-07,
      "loss": 1.5888,
      "step": 3490
    },
    {
      "epoch": 0.5265062966593771,
      "grad_norm": 0.24067603964044604,
      "learning_rate": 2.5625747957775133e-07,
      "loss": 1.575,
      "step": 3491
    },
    {
      "epoch": 0.5266571148480507,
      "grad_norm": 0.24936834335229452,
      "learning_rate": 2.5615123709060395e-07,
      "loss": 1.5443,
      "step": 3492
    },
    {
      "epoch": 0.5268079330367242,
      "grad_norm": 0.24905887261192292,
      "learning_rate": 2.560449988355526e-07,
      "loss": 1.5863,
      "step": 3493
    },
    {
      "epoch": 0.5269587512253978,
      "grad_norm": 0.27961097872082313,
      "learning_rate": 2.5593876483645125e-07,
      "loss": 1.5328,
      "step": 3494
    },
    {
      "epoch": 0.5271095694140714,
      "grad_norm": 0.26027400008482726,
      "learning_rate": 2.5583253511715225e-07,
      "loss": 1.5847,
      "step": 3495
    },
    {
      "epoch": 0.5272603876027449,
      "grad_norm": 0.3010149921067354,
      "learning_rate": 2.5572630970150727e-07,
      "loss": 1.5505,
      "step": 3496
    },
    {
      "epoch": 0.5274112057914184,
      "grad_norm": 0.24462758422642664,
      "learning_rate": 2.556200886133672e-07,
      "loss": 1.5536,
      "step": 3497
    },
    {
      "epoch": 0.527562023980092,
      "grad_norm": 0.24902558953686316,
      "learning_rate": 2.5551387187658154e-07,
      "loss": 1.5946,
      "step": 3498
    },
    {
      "epoch": 0.5277128421687656,
      "grad_norm": 0.2613620026441812,
      "learning_rate": 2.554076595149991e-07,
      "loss": 1.5872,
      "step": 3499
    },
    {
      "epoch": 0.5278636603574391,
      "grad_norm": 0.31161334990915235,
      "learning_rate": 2.553014515524678e-07,
      "loss": 1.5581,
      "step": 3500
    },
    {
      "epoch": 0.5280144785461126,
      "grad_norm": 0.2460958270425893,
      "learning_rate": 2.5519524801283427e-07,
      "loss": 1.5814,
      "step": 3501
    },
    {
      "epoch": 0.5281652967347862,
      "grad_norm": 0.27346749530025577,
      "learning_rate": 2.5508904891994436e-07,
      "loss": 1.5826,
      "step": 3502
    },
    {
      "epoch": 0.5283161149234598,
      "grad_norm": 0.23930263953290692,
      "learning_rate": 2.549828542976428e-07,
      "loss": 1.589,
      "step": 3503
    },
    {
      "epoch": 0.5284669331121333,
      "grad_norm": 0.26860395403048765,
      "learning_rate": 2.5487666416977345e-07,
      "loss": 1.5285,
      "step": 3504
    },
    {
      "epoch": 0.5286177513008069,
      "grad_norm": 0.23913528179331986,
      "learning_rate": 2.5477047856017903e-07,
      "loss": 1.5009,
      "step": 3505
    },
    {
      "epoch": 0.5287685694894805,
      "grad_norm": 0.24512190716848994,
      "learning_rate": 2.546642974927013e-07,
      "loss": 1.519,
      "step": 3506
    },
    {
      "epoch": 0.5289193876781539,
      "grad_norm": 0.2974454301082803,
      "learning_rate": 2.5455812099118105e-07,
      "loss": 1.5193,
      "step": 3507
    },
    {
      "epoch": 0.5290702058668275,
      "grad_norm": 0.24362165691660376,
      "learning_rate": 2.5445194907945797e-07,
      "loss": 1.6427,
      "step": 3508
    },
    {
      "epoch": 0.5292210240555011,
      "grad_norm": 0.24062987703813193,
      "learning_rate": 2.543457817813708e-07,
      "loss": 1.5303,
      "step": 3509
    },
    {
      "epoch": 0.5293718422441747,
      "grad_norm": 0.25073290950057436,
      "learning_rate": 2.5423961912075707e-07,
      "loss": 1.5383,
      "step": 3510
    },
    {
      "epoch": 0.5295226604328482,
      "grad_norm": 0.24069293518381765,
      "learning_rate": 2.5413346112145343e-07,
      "loss": 1.5439,
      "step": 3511
    },
    {
      "epoch": 0.5296734786215218,
      "grad_norm": 0.245924529616333,
      "learning_rate": 2.540273078072956e-07,
      "loss": 1.6223,
      "step": 3512
    },
    {
      "epoch": 0.5298242968101953,
      "grad_norm": 0.28547544148896653,
      "learning_rate": 2.539211592021178e-07,
      "loss": 1.5504,
      "step": 3513
    },
    {
      "epoch": 0.5299751149988688,
      "grad_norm": 0.26993529938387667,
      "learning_rate": 2.538150153297537e-07,
      "loss": 1.594,
      "step": 3514
    },
    {
      "epoch": 0.5301259331875424,
      "grad_norm": 0.251950812081441,
      "learning_rate": 2.537088762140357e-07,
      "loss": 1.6449,
      "step": 3515
    },
    {
      "epoch": 0.530276751376216,
      "grad_norm": 0.2618791960227477,
      "learning_rate": 2.5360274187879497e-07,
      "loss": 1.5748,
      "step": 3516
    },
    {
      "epoch": 0.5304275695648896,
      "grad_norm": 0.2782190715307245,
      "learning_rate": 2.5349661234786185e-07,
      "loss": 1.5571,
      "step": 3517
    },
    {
      "epoch": 0.530578387753563,
      "grad_norm": 0.2675706137635166,
      "learning_rate": 2.5339048764506555e-07,
      "loss": 1.5349,
      "step": 3518
    },
    {
      "epoch": 0.5307292059422366,
      "grad_norm": 0.26581608327045403,
      "learning_rate": 2.5328436779423406e-07,
      "loss": 1.627,
      "step": 3519
    },
    {
      "epoch": 0.5308800241309102,
      "grad_norm": 0.27574459162247106,
      "learning_rate": 2.5317825281919453e-07,
      "loss": 1.587,
      "step": 3520
    },
    {
      "epoch": 0.5310308423195838,
      "grad_norm": 0.2516827300261559,
      "learning_rate": 2.530721427437727e-07,
      "loss": 1.5853,
      "step": 3521
    },
    {
      "epoch": 0.5311816605082573,
      "grad_norm": 0.28610110489774726,
      "learning_rate": 2.529660375917934e-07,
      "loss": 1.5742,
      "step": 3522
    },
    {
      "epoch": 0.5313324786969309,
      "grad_norm": 0.25757914000919657,
      "learning_rate": 2.528599373870804e-07,
      "loss": 1.5944,
      "step": 3523
    },
    {
      "epoch": 0.5314832968856044,
      "grad_norm": 0.26423042771554117,
      "learning_rate": 2.5275384215345627e-07,
      "loss": 1.5864,
      "step": 3524
    },
    {
      "epoch": 0.5316341150742779,
      "grad_norm": 0.2722270107839057,
      "learning_rate": 2.526477519147424e-07,
      "loss": 1.5289,
      "step": 3525
    },
    {
      "epoch": 0.5317849332629515,
      "grad_norm": 0.29345796456266554,
      "learning_rate": 2.5254166669475926e-07,
      "loss": 1.623,
      "step": 3526
    },
    {
      "epoch": 0.5319357514516251,
      "grad_norm": 0.25898910833468286,
      "learning_rate": 2.524355865173259e-07,
      "loss": 1.5909,
      "step": 3527
    },
    {
      "epoch": 0.5320865696402987,
      "grad_norm": 0.32284026703302277,
      "learning_rate": 2.523295114062606e-07,
      "loss": 1.6422,
      "step": 3528
    },
    {
      "epoch": 0.5322373878289721,
      "grad_norm": 0.26149293616511654,
      "learning_rate": 2.522234413853802e-07,
      "loss": 1.5461,
      "step": 3529
    },
    {
      "epoch": 0.5323882060176457,
      "grad_norm": 0.2521176471990762,
      "learning_rate": 2.5211737647850053e-07,
      "loss": 1.5801,
      "step": 3530
    },
    {
      "epoch": 0.5325390242063193,
      "grad_norm": 0.33083588263990305,
      "learning_rate": 2.520113167094363e-07,
      "loss": 1.5678,
      "step": 3531
    },
    {
      "epoch": 0.5326898423949928,
      "grad_norm": 0.24473178120743763,
      "learning_rate": 2.5190526210200093e-07,
      "loss": 1.5813,
      "step": 3532
    },
    {
      "epoch": 0.5328406605836664,
      "grad_norm": 0.25503783863822016,
      "learning_rate": 2.517992126800068e-07,
      "loss": 1.5506,
      "step": 3533
    },
    {
      "epoch": 0.53299147877234,
      "grad_norm": 0.25021014615026094,
      "learning_rate": 2.5169316846726515e-07,
      "loss": 1.5091,
      "step": 3534
    },
    {
      "epoch": 0.5331422969610135,
      "grad_norm": 0.2541044663955351,
      "learning_rate": 2.515871294875859e-07,
      "loss": 1.5303,
      "step": 3535
    },
    {
      "epoch": 0.533293115149687,
      "grad_norm": 0.37817575805690823,
      "learning_rate": 2.51481095764778e-07,
      "loss": 1.5387,
      "step": 3536
    },
    {
      "epoch": 0.5334439333383606,
      "grad_norm": 0.24535242253342482,
      "learning_rate": 2.5137506732264895e-07,
      "loss": 1.5676,
      "step": 3537
    },
    {
      "epoch": 0.5335947515270342,
      "grad_norm": 0.2733669094173515,
      "learning_rate": 2.512690441850054e-07,
      "loss": 1.5428,
      "step": 3538
    },
    {
      "epoch": 0.5337455697157077,
      "grad_norm": 0.25954789607178197,
      "learning_rate": 2.5116302637565255e-07,
      "loss": 1.5862,
      "step": 3539
    },
    {
      "epoch": 0.5338963879043813,
      "grad_norm": 0.309637144793164,
      "learning_rate": 2.5105701391839446e-07,
      "loss": 1.5611,
      "step": 3540
    },
    {
      "epoch": 0.5340472060930548,
      "grad_norm": 0.2420016020049591,
      "learning_rate": 2.509510068370341e-07,
      "loss": 1.5846,
      "step": 3541
    },
    {
      "epoch": 0.5341980242817284,
      "grad_norm": 0.26169552956251124,
      "learning_rate": 2.5084500515537304e-07,
      "loss": 1.573,
      "step": 3542
    },
    {
      "epoch": 0.5343488424704019,
      "grad_norm": 0.256453723857227,
      "learning_rate": 2.5073900889721174e-07,
      "loss": 1.5034,
      "step": 3543
    },
    {
      "epoch": 0.5344996606590755,
      "grad_norm": 0.27813789544137113,
      "learning_rate": 2.506330180863496e-07,
      "loss": 1.5684,
      "step": 3544
    },
    {
      "epoch": 0.5346504788477491,
      "grad_norm": 0.25234294332864915,
      "learning_rate": 2.5052703274658446e-07,
      "loss": 1.5334,
      "step": 3545
    },
    {
      "epoch": 0.5348012970364225,
      "grad_norm": 0.2625582654158137,
      "learning_rate": 2.5042105290171326e-07,
      "loss": 1.582,
      "step": 3546
    },
    {
      "epoch": 0.5349521152250961,
      "grad_norm": 0.2367841824021432,
      "learning_rate": 2.5031507857553154e-07,
      "loss": 1.5782,
      "step": 3547
    },
    {
      "epoch": 0.5351029334137697,
      "grad_norm": 0.2562501960561911,
      "learning_rate": 2.5020910979183346e-07,
      "loss": 1.596,
      "step": 3548
    },
    {
      "epoch": 0.5352537516024433,
      "grad_norm": 0.4601915570204031,
      "learning_rate": 2.5010314657441235e-07,
      "loss": 1.5404,
      "step": 3549
    },
    {
      "epoch": 0.5354045697911168,
      "grad_norm": 0.24387253387053395,
      "learning_rate": 2.499971889470598e-07,
      "loss": 1.5421,
      "step": 3550
    },
    {
      "epoch": 0.5355553879797904,
      "grad_norm": 0.265517357013569,
      "learning_rate": 2.498912369335666e-07,
      "loss": 1.611,
      "step": 3551
    },
    {
      "epoch": 0.5357062061684639,
      "grad_norm": 0.36906846802969784,
      "learning_rate": 2.497852905577219e-07,
      "loss": 1.5658,
      "step": 3552
    },
    {
      "epoch": 0.5358570243571374,
      "grad_norm": 0.259593341801903,
      "learning_rate": 2.496793498433138e-07,
      "loss": 1.5748,
      "step": 3553
    },
    {
      "epoch": 0.536007842545811,
      "grad_norm": 0.3200271372913328,
      "learning_rate": 2.495734148141291e-07,
      "loss": 1.4802,
      "step": 3554
    },
    {
      "epoch": 0.5361586607344846,
      "grad_norm": 0.2905874652597384,
      "learning_rate": 2.4946748549395325e-07,
      "loss": 1.5384,
      "step": 3555
    },
    {
      "epoch": 0.5363094789231582,
      "grad_norm": 0.25318479822579476,
      "learning_rate": 2.4936156190657047e-07,
      "loss": 1.5881,
      "step": 3556
    },
    {
      "epoch": 0.5364602971118317,
      "grad_norm": 0.2576415973112576,
      "learning_rate": 2.492556440757638e-07,
      "loss": 1.5923,
      "step": 3557
    },
    {
      "epoch": 0.5366111153005052,
      "grad_norm": 0.2583235905639093,
      "learning_rate": 2.491497320253147e-07,
      "loss": 1.6253,
      "step": 3558
    },
    {
      "epoch": 0.5367619334891788,
      "grad_norm": 0.24540277572881986,
      "learning_rate": 2.4904382577900367e-07,
      "loss": 1.5298,
      "step": 3559
    },
    {
      "epoch": 0.5369127516778524,
      "grad_norm": 0.24986792681543368,
      "learning_rate": 2.4893792536060966e-07,
      "loss": 1.5946,
      "step": 3560
    },
    {
      "epoch": 0.5370635698665259,
      "grad_norm": 0.2513171723500393,
      "learning_rate": 2.488320307939104e-07,
      "loss": 1.5697,
      "step": 3561
    },
    {
      "epoch": 0.5372143880551995,
      "grad_norm": 0.3112567416606474,
      "learning_rate": 2.487261421026823e-07,
      "loss": 1.5535,
      "step": 3562
    },
    {
      "epoch": 0.537365206243873,
      "grad_norm": 0.2575564935209731,
      "learning_rate": 2.486202593107006e-07,
      "loss": 1.6151,
      "step": 3563
    },
    {
      "epoch": 0.5375160244325465,
      "grad_norm": 0.2567580878163092,
      "learning_rate": 2.485143824417389e-07,
      "loss": 1.5657,
      "step": 3564
    },
    {
      "epoch": 0.5376668426212201,
      "grad_norm": 0.24389373835248265,
      "learning_rate": 2.4840851151956957e-07,
      "loss": 1.6059,
      "step": 3565
    },
    {
      "epoch": 0.5378176608098937,
      "grad_norm": 0.25309193545332304,
      "learning_rate": 2.483026465679639e-07,
      "loss": 1.5418,
      "step": 3566
    },
    {
      "epoch": 0.5379684789985673,
      "grad_norm": 0.2557894707789005,
      "learning_rate": 2.481967876106917e-07,
      "loss": 1.5498,
      "step": 3567
    },
    {
      "epoch": 0.5381192971872408,
      "grad_norm": 0.24582848285938932,
      "learning_rate": 2.480909346715211e-07,
      "loss": 1.5359,
      "step": 3568
    },
    {
      "epoch": 0.5382701153759143,
      "grad_norm": 0.25466688551768263,
      "learning_rate": 2.479850877742195e-07,
      "loss": 1.5738,
      "step": 3569
    },
    {
      "epoch": 0.5384209335645879,
      "grad_norm": 0.25479550425895725,
      "learning_rate": 2.4787924694255236e-07,
      "loss": 1.5369,
      "step": 3570
    },
    {
      "epoch": 0.5385717517532614,
      "grad_norm": 0.242997356531885,
      "learning_rate": 2.4777341220028416e-07,
      "loss": 1.572,
      "step": 3571
    },
    {
      "epoch": 0.538722569941935,
      "grad_norm": 0.2436254587281785,
      "learning_rate": 2.4766758357117784e-07,
      "loss": 1.536,
      "step": 3572
    },
    {
      "epoch": 0.5388733881306086,
      "grad_norm": 0.3239443419270094,
      "learning_rate": 2.4756176107899494e-07,
      "loss": 1.5273,
      "step": 3573
    },
    {
      "epoch": 0.539024206319282,
      "grad_norm": 0.25572158284117197,
      "learning_rate": 2.4745594474749585e-07,
      "loss": 1.4892,
      "step": 3574
    },
    {
      "epoch": 0.5391750245079556,
      "grad_norm": 0.2625148394944358,
      "learning_rate": 2.4735013460043933e-07,
      "loss": 1.6197,
      "step": 3575
    },
    {
      "epoch": 0.5393258426966292,
      "grad_norm": 0.24336310043788423,
      "learning_rate": 2.4724433066158283e-07,
      "loss": 1.653,
      "step": 3576
    },
    {
      "epoch": 0.5394766608853028,
      "grad_norm": 0.25107677729321576,
      "learning_rate": 2.471385329546825e-07,
      "loss": 1.6108,
      "step": 3577
    },
    {
      "epoch": 0.5396274790739763,
      "grad_norm": 0.25565257098025534,
      "learning_rate": 2.4703274150349285e-07,
      "loss": 1.5368,
      "step": 3578
    },
    {
      "epoch": 0.5397782972626499,
      "grad_norm": 0.2516168212131607,
      "learning_rate": 2.4692695633176727e-07,
      "loss": 1.6004,
      "step": 3579
    },
    {
      "epoch": 0.5399291154513234,
      "grad_norm": 0.357565028404368,
      "learning_rate": 2.4682117746325753e-07,
      "loss": 1.5574,
      "step": 3580
    },
    {
      "epoch": 0.540079933639997,
      "grad_norm": 0.24086631336091513,
      "learning_rate": 2.467154049217142e-07,
      "loss": 1.5089,
      "step": 3581
    },
    {
      "epoch": 0.5402307518286705,
      "grad_norm": 0.3350214806976764,
      "learning_rate": 2.466096387308861e-07,
      "loss": 1.6208,
      "step": 3582
    },
    {
      "epoch": 0.5403815700173441,
      "grad_norm": 0.249638578712114,
      "learning_rate": 2.4650387891452096e-07,
      "loss": 1.59,
      "step": 3583
    },
    {
      "epoch": 0.5405323882060177,
      "grad_norm": 0.2542986654698333,
      "learning_rate": 2.463981254963649e-07,
      "loss": 1.543,
      "step": 3584
    },
    {
      "epoch": 0.5406832063946913,
      "grad_norm": 0.26067240069467273,
      "learning_rate": 2.462923785001627e-07,
      "loss": 1.5852,
      "step": 3585
    },
    {
      "epoch": 0.5408340245833647,
      "grad_norm": 0.2463546732060017,
      "learning_rate": 2.461866379496575e-07,
      "loss": 1.5081,
      "step": 3586
    },
    {
      "epoch": 0.5409848427720383,
      "grad_norm": 0.23489869798759375,
      "learning_rate": 2.460809038685912e-07,
      "loss": 1.5447,
      "step": 3587
    },
    {
      "epoch": 0.5411356609607119,
      "grad_norm": 0.24221533397665837,
      "learning_rate": 2.459751762807043e-07,
      "loss": 1.5746,
      "step": 3588
    },
    {
      "epoch": 0.5412864791493854,
      "grad_norm": 0.24791504924382135,
      "learning_rate": 2.458694552097356e-07,
      "loss": 1.5382,
      "step": 3589
    },
    {
      "epoch": 0.541437297338059,
      "grad_norm": 0.2574534029734148,
      "learning_rate": 2.4576374067942256e-07,
      "loss": 1.6216,
      "step": 3590
    },
    {
      "epoch": 0.5415881155267325,
      "grad_norm": 0.2679891830487016,
      "learning_rate": 2.4565803271350117e-07,
      "loss": 1.5621,
      "step": 3591
    },
    {
      "epoch": 0.541738933715406,
      "grad_norm": 0.34319500827264665,
      "learning_rate": 2.4555233133570595e-07,
      "loss": 1.4834,
      "step": 3592
    },
    {
      "epoch": 0.5418897519040796,
      "grad_norm": 0.24811600901697226,
      "learning_rate": 2.4544663656977004e-07,
      "loss": 1.5348,
      "step": 3593
    },
    {
      "epoch": 0.5420405700927532,
      "grad_norm": 0.23934457658158845,
      "learning_rate": 2.453409484394248e-07,
      "loss": 1.5886,
      "step": 3594
    },
    {
      "epoch": 0.5421913882814268,
      "grad_norm": 0.2466036375870887,
      "learning_rate": 2.452352669684005e-07,
      "loss": 1.5959,
      "step": 3595
    },
    {
      "epoch": 0.5423422064701003,
      "grad_norm": 0.2514097237587263,
      "learning_rate": 2.4512959218042556e-07,
      "loss": 1.4942,
      "step": 3596
    },
    {
      "epoch": 0.5424930246587738,
      "grad_norm": 0.4800774121409877,
      "learning_rate": 2.450239240992271e-07,
      "loss": 1.5149,
      "step": 3597
    },
    {
      "epoch": 0.5426438428474474,
      "grad_norm": 0.254282445301358,
      "learning_rate": 2.4491826274853074e-07,
      "loss": 1.6246,
      "step": 3598
    },
    {
      "epoch": 0.542794661036121,
      "grad_norm": 0.2580213491036497,
      "learning_rate": 2.448126081520604e-07,
      "loss": 1.5342,
      "step": 3599
    },
    {
      "epoch": 0.5429454792247945,
      "grad_norm": 0.25683975686626387,
      "learning_rate": 2.4470696033353866e-07,
      "loss": 1.5651,
      "step": 3600
    },
    {
      "epoch": 0.5430962974134681,
      "grad_norm": 0.24544494870168712,
      "learning_rate": 2.4460131931668664e-07,
      "loss": 1.5922,
      "step": 3601
    },
    {
      "epoch": 0.5432471156021417,
      "grad_norm": 0.2567332109666076,
      "learning_rate": 2.4449568512522367e-07,
      "loss": 1.5531,
      "step": 3602
    },
    {
      "epoch": 0.5433979337908151,
      "grad_norm": 0.26764572837580614,
      "learning_rate": 2.4439005778286785e-07,
      "loss": 1.5531,
      "step": 3603
    },
    {
      "epoch": 0.5435487519794887,
      "grad_norm": 0.3361771588722897,
      "learning_rate": 2.4428443731333545e-07,
      "loss": 1.6069,
      "step": 3604
    },
    {
      "epoch": 0.5436995701681623,
      "grad_norm": 0.2841049813886416,
      "learning_rate": 2.441788237403414e-07,
      "loss": 1.5296,
      "step": 3605
    },
    {
      "epoch": 0.5438503883568359,
      "grad_norm": 0.2545745313697678,
      "learning_rate": 2.440732170875991e-07,
      "loss": 1.5929,
      "step": 3606
    },
    {
      "epoch": 0.5440012065455094,
      "grad_norm": 0.25337715389509585,
      "learning_rate": 2.439676173788202e-07,
      "loss": 1.5714,
      "step": 3607
    },
    {
      "epoch": 0.5441520247341829,
      "grad_norm": 0.2446314960927492,
      "learning_rate": 2.43862024637715e-07,
      "loss": 1.5205,
      "step": 3608
    },
    {
      "epoch": 0.5443028429228565,
      "grad_norm": 0.2583744288497405,
      "learning_rate": 2.43756438887992e-07,
      "loss": 1.5997,
      "step": 3609
    },
    {
      "epoch": 0.54445366111153,
      "grad_norm": 0.2520486322377621,
      "learning_rate": 2.4365086015335853e-07,
      "loss": 1.5467,
      "step": 3610
    },
    {
      "epoch": 0.5446044793002036,
      "grad_norm": 0.2531732482880286,
      "learning_rate": 2.4354528845751986e-07,
      "loss": 1.5206,
      "step": 3611
    },
    {
      "epoch": 0.5447552974888772,
      "grad_norm": 0.2483682771543181,
      "learning_rate": 2.4343972382418e-07,
      "loss": 1.6046,
      "step": 3612
    },
    {
      "epoch": 0.5449061156775508,
      "grad_norm": 0.24439864855994065,
      "learning_rate": 2.4333416627704136e-07,
      "loss": 1.4808,
      "step": 3613
    },
    {
      "epoch": 0.5450569338662242,
      "grad_norm": 0.3600198536008531,
      "learning_rate": 2.4322861583980447e-07,
      "loss": 1.529,
      "step": 3614
    },
    {
      "epoch": 0.5452077520548978,
      "grad_norm": 0.2490329476442969,
      "learning_rate": 2.431230725361687e-07,
      "loss": 1.542,
      "step": 3615
    },
    {
      "epoch": 0.5453585702435714,
      "grad_norm": 0.2473844578166116,
      "learning_rate": 2.430175363898315e-07,
      "loss": 1.5114,
      "step": 3616
    },
    {
      "epoch": 0.545509388432245,
      "grad_norm": 0.23918089282027064,
      "learning_rate": 2.4291200742448873e-07,
      "loss": 1.5595,
      "step": 3617
    },
    {
      "epoch": 0.5456602066209185,
      "grad_norm": 0.23739710855690835,
      "learning_rate": 2.4280648566383486e-07,
      "loss": 1.5684,
      "step": 3618
    },
    {
      "epoch": 0.545811024809592,
      "grad_norm": 0.2486525885290028,
      "learning_rate": 2.427009711315626e-07,
      "loss": 1.5525,
      "step": 3619
    },
    {
      "epoch": 0.5459618429982656,
      "grad_norm": 0.24953932351497338,
      "learning_rate": 2.425954638513629e-07,
      "loss": 1.5736,
      "step": 3620
    },
    {
      "epoch": 0.5461126611869391,
      "grad_norm": 0.31864413884135867,
      "learning_rate": 2.424899638469253e-07,
      "loss": 1.5432,
      "step": 3621
    },
    {
      "epoch": 0.5462634793756127,
      "grad_norm": 0.2517109482883882,
      "learning_rate": 2.4238447114193757e-07,
      "loss": 1.5364,
      "step": 3622
    },
    {
      "epoch": 0.5464142975642863,
      "grad_norm": 0.24347151696480387,
      "learning_rate": 2.4227898576008596e-07,
      "loss": 1.5847,
      "step": 3623
    },
    {
      "epoch": 0.5465651157529599,
      "grad_norm": 0.23677989847203543,
      "learning_rate": 2.4217350772505503e-07,
      "loss": 1.4737,
      "step": 3624
    },
    {
      "epoch": 0.5467159339416333,
      "grad_norm": 0.2468086290035077,
      "learning_rate": 2.4206803706052757e-07,
      "loss": 1.5888,
      "step": 3625
    },
    {
      "epoch": 0.5468667521303069,
      "grad_norm": 0.2606802458058856,
      "learning_rate": 2.4196257379018485e-07,
      "loss": 1.5506,
      "step": 3626
    },
    {
      "epoch": 0.5470175703189805,
      "grad_norm": 0.23904823371329273,
      "learning_rate": 2.4185711793770655e-07,
      "loss": 1.5732,
      "step": 3627
    },
    {
      "epoch": 0.547168388507654,
      "grad_norm": 0.36528586683365055,
      "learning_rate": 2.4175166952677045e-07,
      "loss": 1.5987,
      "step": 3628
    },
    {
      "epoch": 0.5473192066963276,
      "grad_norm": 0.25159845930818714,
      "learning_rate": 2.416462285810529e-07,
      "loss": 1.5513,
      "step": 3629
    },
    {
      "epoch": 0.5474700248850012,
      "grad_norm": 0.24853920826770132,
      "learning_rate": 2.415407951242283e-07,
      "loss": 1.5624,
      "step": 3630
    },
    {
      "epoch": 0.5476208430736746,
      "grad_norm": 0.23867638541766906,
      "learning_rate": 2.414353691799698e-07,
      "loss": 1.6519,
      "step": 3631
    },
    {
      "epoch": 0.5477716612623482,
      "grad_norm": 0.28298887384204163,
      "learning_rate": 2.413299507719483e-07,
      "loss": 1.6201,
      "step": 3632
    },
    {
      "epoch": 0.5479224794510218,
      "grad_norm": 0.2506146374102195,
      "learning_rate": 2.412245399238335e-07,
      "loss": 1.5457,
      "step": 3633
    },
    {
      "epoch": 0.5480732976396954,
      "grad_norm": 0.24805360609832733,
      "learning_rate": 2.411191366592931e-07,
      "loss": 1.6349,
      "step": 3634
    },
    {
      "epoch": 0.5482241158283689,
      "grad_norm": 0.25391443799219615,
      "learning_rate": 2.4101374100199315e-07,
      "loss": 1.5485,
      "step": 3635
    },
    {
      "epoch": 0.5483749340170424,
      "grad_norm": 0.2979419731401153,
      "learning_rate": 2.409083529755983e-07,
      "loss": 1.5457,
      "step": 3636
    },
    {
      "epoch": 0.548525752205716,
      "grad_norm": 0.25061725613115277,
      "learning_rate": 2.40802972603771e-07,
      "loss": 1.5062,
      "step": 3637
    },
    {
      "epoch": 0.5486765703943896,
      "grad_norm": 0.3675582934544481,
      "learning_rate": 2.406975999101723e-07,
      "loss": 1.514,
      "step": 3638
    },
    {
      "epoch": 0.5488273885830631,
      "grad_norm": 0.2337542475694023,
      "learning_rate": 2.4059223491846135e-07,
      "loss": 1.5166,
      "step": 3639
    },
    {
      "epoch": 0.5489782067717367,
      "grad_norm": 0.25198011405401805,
      "learning_rate": 2.4048687765229574e-07,
      "loss": 1.5625,
      "step": 3640
    },
    {
      "epoch": 0.5491290249604103,
      "grad_norm": 0.24638371098340794,
      "learning_rate": 2.403815281353313e-07,
      "loss": 1.578,
      "step": 3641
    },
    {
      "epoch": 0.5492798431490837,
      "grad_norm": 0.2563153278271117,
      "learning_rate": 2.402761863912219e-07,
      "loss": 1.6488,
      "step": 3642
    },
    {
      "epoch": 0.5494306613377573,
      "grad_norm": 0.2580106370971055,
      "learning_rate": 2.4017085244362e-07,
      "loss": 1.6231,
      "step": 3643
    },
    {
      "epoch": 0.5495814795264309,
      "grad_norm": 0.24455256267634917,
      "learning_rate": 2.4006552631617604e-07,
      "loss": 1.5983,
      "step": 3644
    },
    {
      "epoch": 0.5497322977151045,
      "grad_norm": 0.39325417509774485,
      "learning_rate": 2.399602080325388e-07,
      "loss": 1.5503,
      "step": 3645
    },
    {
      "epoch": 0.549883115903778,
      "grad_norm": 0.2441326961183391,
      "learning_rate": 2.3985489761635543e-07,
      "loss": 1.5835,
      "step": 3646
    },
    {
      "epoch": 0.5500339340924516,
      "grad_norm": 0.26732110100009343,
      "learning_rate": 2.3974959509127095e-07,
      "loss": 1.5572,
      "step": 3647
    },
    {
      "epoch": 0.5501847522811251,
      "grad_norm": 0.2643667705732118,
      "learning_rate": 2.3964430048092903e-07,
      "loss": 1.5744,
      "step": 3648
    },
    {
      "epoch": 0.5503355704697986,
      "grad_norm": 0.25634976898777334,
      "learning_rate": 2.3953901380897137e-07,
      "loss": 1.5818,
      "step": 3649
    },
    {
      "epoch": 0.5504863886584722,
      "grad_norm": 0.24104439553659085,
      "learning_rate": 2.3943373509903776e-07,
      "loss": 1.5567,
      "step": 3650
    },
    {
      "epoch": 0.5506372068471458,
      "grad_norm": 0.27330546395370564,
      "learning_rate": 2.393284643747665e-07,
      "loss": 1.5864,
      "step": 3651
    },
    {
      "epoch": 0.5507880250358194,
      "grad_norm": 0.24567026735933592,
      "learning_rate": 2.3922320165979384e-07,
      "loss": 1.5429,
      "step": 3652
    },
    {
      "epoch": 0.5509388432244928,
      "grad_norm": 0.25877285545727613,
      "learning_rate": 2.3911794697775427e-07,
      "loss": 1.5658,
      "step": 3653
    },
    {
      "epoch": 0.5510896614131664,
      "grad_norm": 0.2829892531211758,
      "learning_rate": 2.390127003522807e-07,
      "loss": 1.5716,
      "step": 3654
    },
    {
      "epoch": 0.55124047960184,
      "grad_norm": 0.26420734964587744,
      "learning_rate": 2.3890746180700393e-07,
      "loss": 1.5965,
      "step": 3655
    },
    {
      "epoch": 0.5513912977905135,
      "grad_norm": 0.2813925312118151,
      "learning_rate": 2.388022313655531e-07,
      "loss": 1.5952,
      "step": 3656
    },
    {
      "epoch": 0.5515421159791871,
      "grad_norm": 0.27421147776570304,
      "learning_rate": 2.3869700905155555e-07,
      "loss": 1.5298,
      "step": 3657
    },
    {
      "epoch": 0.5516929341678607,
      "grad_norm": 0.24612321380173793,
      "learning_rate": 2.385917948886367e-07,
      "loss": 1.5782,
      "step": 3658
    },
    {
      "epoch": 0.5518437523565342,
      "grad_norm": 0.3010225001430544,
      "learning_rate": 2.384865889004203e-07,
      "loss": 1.5167,
      "step": 3659
    },
    {
      "epoch": 0.5519945705452077,
      "grad_norm": 0.25818104168215283,
      "learning_rate": 2.3838139111052796e-07,
      "loss": 1.5353,
      "step": 3660
    },
    {
      "epoch": 0.5521453887338813,
      "grad_norm": 0.2594370849630129,
      "learning_rate": 2.3827620154257982e-07,
      "loss": 1.5772,
      "step": 3661
    },
    {
      "epoch": 0.5522962069225549,
      "grad_norm": 0.2581055323738375,
      "learning_rate": 2.3817102022019399e-07,
      "loss": 1.5492,
      "step": 3662
    },
    {
      "epoch": 0.5524470251112285,
      "grad_norm": 0.380152434425543,
      "learning_rate": 2.3806584716698663e-07,
      "loss": 1.5527,
      "step": 3663
    },
    {
      "epoch": 0.5525978432999019,
      "grad_norm": 0.24856846332942859,
      "learning_rate": 2.379606824065723e-07,
      "loss": 1.5547,
      "step": 3664
    },
    {
      "epoch": 0.5527486614885755,
      "grad_norm": 0.2497454185942363,
      "learning_rate": 2.3785552596256343e-07,
      "loss": 1.5996,
      "step": 3665
    },
    {
      "epoch": 0.5528994796772491,
      "grad_norm": 0.31191475381343176,
      "learning_rate": 2.377503778585707e-07,
      "loss": 1.5698,
      "step": 3666
    },
    {
      "epoch": 0.5530502978659226,
      "grad_norm": 0.25142526486928735,
      "learning_rate": 2.3764523811820308e-07,
      "loss": 1.5149,
      "step": 3667
    },
    {
      "epoch": 0.5532011160545962,
      "grad_norm": 0.27740314564154706,
      "learning_rate": 2.375401067650673e-07,
      "loss": 1.5717,
      "step": 3668
    },
    {
      "epoch": 0.5533519342432698,
      "grad_norm": 0.2549316680761266,
      "learning_rate": 2.3743498382276855e-07,
      "loss": 1.5578,
      "step": 3669
    },
    {
      "epoch": 0.5535027524319432,
      "grad_norm": 0.23933092939208836,
      "learning_rate": 2.3732986931490996e-07,
      "loss": 1.5491,
      "step": 3670
    },
    {
      "epoch": 0.5536535706206168,
      "grad_norm": 0.3834995115356936,
      "learning_rate": 2.3722476326509272e-07,
      "loss": 1.5602,
      "step": 3671
    },
    {
      "epoch": 0.5538043888092904,
      "grad_norm": 0.37334563762423256,
      "learning_rate": 2.3711966569691634e-07,
      "loss": 1.6239,
      "step": 3672
    },
    {
      "epoch": 0.553955206997964,
      "grad_norm": 0.2395634198444132,
      "learning_rate": 2.3701457663397812e-07,
      "loss": 1.5575,
      "step": 3673
    },
    {
      "epoch": 0.5541060251866375,
      "grad_norm": 0.26102920902338,
      "learning_rate": 2.3690949609987375e-07,
      "loss": 1.5986,
      "step": 3674
    },
    {
      "epoch": 0.5542568433753111,
      "grad_norm": 0.8757806837521017,
      "learning_rate": 2.3680442411819684e-07,
      "loss": 1.5603,
      "step": 3675
    },
    {
      "epoch": 0.5544076615639846,
      "grad_norm": 0.3097578413639605,
      "learning_rate": 2.3669936071253905e-07,
      "loss": 1.5871,
      "step": 3676
    },
    {
      "epoch": 0.5545584797526582,
      "grad_norm": 0.2415676421450433,
      "learning_rate": 2.3659430590649032e-07,
      "loss": 1.6083,
      "step": 3677
    },
    {
      "epoch": 0.5547092979413317,
      "grad_norm": 0.25008839522250964,
      "learning_rate": 2.364892597236383e-07,
      "loss": 1.5743,
      "step": 3678
    },
    {
      "epoch": 0.5548601161300053,
      "grad_norm": 0.28816170962884513,
      "learning_rate": 2.3638422218756903e-07,
      "loss": 1.5044,
      "step": 3679
    },
    {
      "epoch": 0.5550109343186789,
      "grad_norm": 0.259771854122331,
      "learning_rate": 2.3627919332186657e-07,
      "loss": 1.517,
      "step": 3680
    },
    {
      "epoch": 0.5551617525073523,
      "grad_norm": 0.2921177162408171,
      "learning_rate": 2.3617417315011285e-07,
      "loss": 1.5945,
      "step": 3681
    },
    {
      "epoch": 0.5553125706960259,
      "grad_norm": 0.25867029639801753,
      "learning_rate": 2.3606916169588804e-07,
      "loss": 1.5291,
      "step": 3682
    },
    {
      "epoch": 0.5554633888846995,
      "grad_norm": 0.2432258692914303,
      "learning_rate": 2.359641589827701e-07,
      "loss": 1.5735,
      "step": 3683
    },
    {
      "epoch": 0.5556142070733731,
      "grad_norm": 0.25144434374534613,
      "learning_rate": 2.358591650343354e-07,
      "loss": 1.5869,
      "step": 3684
    },
    {
      "epoch": 0.5557650252620466,
      "grad_norm": 0.2548083114866012,
      "learning_rate": 2.357541798741581e-07,
      "loss": 1.4551,
      "step": 3685
    },
    {
      "epoch": 0.5559158434507202,
      "grad_norm": 0.2580466394469013,
      "learning_rate": 2.3564920352581025e-07,
      "loss": 1.5541,
      "step": 3686
    },
    {
      "epoch": 0.5560666616393937,
      "grad_norm": 0.24127002626442975,
      "learning_rate": 2.355442360128624e-07,
      "loss": 1.5381,
      "step": 3687
    },
    {
      "epoch": 0.5562174798280672,
      "grad_norm": 0.24253121564118715,
      "learning_rate": 2.3543927735888248e-07,
      "loss": 1.5691,
      "step": 3688
    },
    {
      "epoch": 0.5563682980167408,
      "grad_norm": 0.2569879120797617,
      "learning_rate": 2.35334327587437e-07,
      "loss": 1.5691,
      "step": 3689
    },
    {
      "epoch": 0.5565191162054144,
      "grad_norm": 0.23527231150376013,
      "learning_rate": 2.352293867220902e-07,
      "loss": 1.5975,
      "step": 3690
    },
    {
      "epoch": 0.556669934394088,
      "grad_norm": 0.25058550713662187,
      "learning_rate": 2.3512445478640425e-07,
      "loss": 1.5556,
      "step": 3691
    },
    {
      "epoch": 0.5568207525827615,
      "grad_norm": 0.2446541392046704,
      "learning_rate": 2.3501953180393958e-07,
      "loss": 1.6593,
      "step": 3692
    },
    {
      "epoch": 0.556971570771435,
      "grad_norm": 0.25115483375925146,
      "learning_rate": 2.3491461779825433e-07,
      "loss": 1.5412,
      "step": 3693
    },
    {
      "epoch": 0.5571223889601086,
      "grad_norm": 0.2520219702329995,
      "learning_rate": 2.348097127929048e-07,
      "loss": 1.5735,
      "step": 3694
    },
    {
      "epoch": 0.5572732071487821,
      "grad_norm": 0.2594579956511486,
      "learning_rate": 2.3470481681144526e-07,
      "loss": 1.5407,
      "step": 3695
    },
    {
      "epoch": 0.5574240253374557,
      "grad_norm": 0.2632419593584636,
      "learning_rate": 2.3459992987742783e-07,
      "loss": 1.5122,
      "step": 3696
    },
    {
      "epoch": 0.5575748435261293,
      "grad_norm": 0.24066331709923033,
      "learning_rate": 2.3449505201440272e-07,
      "loss": 1.5542,
      "step": 3697
    },
    {
      "epoch": 0.5577256617148028,
      "grad_norm": 0.2670785975264236,
      "learning_rate": 2.3439018324591815e-07,
      "loss": 1.5664,
      "step": 3698
    },
    {
      "epoch": 0.5578764799034763,
      "grad_norm": 0.2566681108304461,
      "learning_rate": 2.3428532359552006e-07,
      "loss": 1.4981,
      "step": 3699
    },
    {
      "epoch": 0.5580272980921499,
      "grad_norm": 0.3193962578119453,
      "learning_rate": 2.341804730867526e-07,
      "loss": 1.5459,
      "step": 3700
    },
    {
      "epoch": 0.5581781162808235,
      "grad_norm": 0.24338695588518144,
      "learning_rate": 2.3407563174315764e-07,
      "loss": 1.6626,
      "step": 3701
    },
    {
      "epoch": 0.558328934469497,
      "grad_norm": 0.26531266120238695,
      "learning_rate": 2.3397079958827527e-07,
      "loss": 1.5394,
      "step": 3702
    },
    {
      "epoch": 0.5584797526581706,
      "grad_norm": 0.25190397601521103,
      "learning_rate": 2.3386597664564335e-07,
      "loss": 1.5861,
      "step": 3703
    },
    {
      "epoch": 0.5586305708468441,
      "grad_norm": 0.2446769652456106,
      "learning_rate": 2.337611629387975e-07,
      "loss": 1.5025,
      "step": 3704
    },
    {
      "epoch": 0.5587813890355177,
      "grad_norm": 0.25365522039409677,
      "learning_rate": 2.3365635849127163e-07,
      "loss": 1.5669,
      "step": 3705
    },
    {
      "epoch": 0.5589322072241912,
      "grad_norm": 0.25214964867523426,
      "learning_rate": 2.3355156332659743e-07,
      "loss": 1.5295,
      "step": 3706
    },
    {
      "epoch": 0.5590830254128648,
      "grad_norm": 0.2653556219624336,
      "learning_rate": 2.3344677746830426e-07,
      "loss": 1.5761,
      "step": 3707
    },
    {
      "epoch": 0.5592338436015384,
      "grad_norm": 0.2480112212862208,
      "learning_rate": 2.333420009399198e-07,
      "loss": 1.5422,
      "step": 3708
    },
    {
      "epoch": 0.5593846617902118,
      "grad_norm": 0.24583448225831397,
      "learning_rate": 2.3323723376496928e-07,
      "loss": 1.5547,
      "step": 3709
    },
    {
      "epoch": 0.5595354799788854,
      "grad_norm": 0.26179366693271394,
      "learning_rate": 2.3313247596697604e-07,
      "loss": 1.5524,
      "step": 3710
    },
    {
      "epoch": 0.559686298167559,
      "grad_norm": 0.2810880095236748,
      "learning_rate": 2.330277275694613e-07,
      "loss": 1.5449,
      "step": 3711
    },
    {
      "epoch": 0.5598371163562326,
      "grad_norm": 0.28865368321909335,
      "learning_rate": 2.3292298859594407e-07,
      "loss": 1.5111,
      "step": 3712
    },
    {
      "epoch": 0.5599879345449061,
      "grad_norm": 0.28220169530861183,
      "learning_rate": 2.3281825906994134e-07,
      "loss": 1.5148,
      "step": 3713
    },
    {
      "epoch": 0.5601387527335797,
      "grad_norm": 0.8659363874227264,
      "learning_rate": 2.3271353901496786e-07,
      "loss": 1.5034,
      "step": 3714
    },
    {
      "epoch": 0.5602895709222532,
      "grad_norm": 0.24744321943675116,
      "learning_rate": 2.3260882845453643e-07,
      "loss": 1.5889,
      "step": 3715
    },
    {
      "epoch": 0.5604403891109268,
      "grad_norm": 0.256305715829819,
      "learning_rate": 2.325041274121576e-07,
      "loss": 1.5833,
      "step": 3716
    },
    {
      "epoch": 0.5605912072996003,
      "grad_norm": 0.26064801863351983,
      "learning_rate": 2.3239943591133975e-07,
      "loss": 1.5298,
      "step": 3717
    },
    {
      "epoch": 0.5607420254882739,
      "grad_norm": 0.30004312330576116,
      "learning_rate": 2.3229475397558918e-07,
      "loss": 1.6146,
      "step": 3718
    },
    {
      "epoch": 0.5608928436769475,
      "grad_norm": 0.24133194064638744,
      "learning_rate": 2.3219008162841014e-07,
      "loss": 1.5237,
      "step": 3719
    },
    {
      "epoch": 0.561043661865621,
      "grad_norm": 0.2921675486036917,
      "learning_rate": 2.3208541889330446e-07,
      "loss": 1.5492,
      "step": 3720
    },
    {
      "epoch": 0.5611944800542945,
      "grad_norm": 0.2555644620420599,
      "learning_rate": 2.3198076579377208e-07,
      "loss": 1.4912,
      "step": 3721
    },
    {
      "epoch": 0.5613452982429681,
      "grad_norm": 0.24578747399944298,
      "learning_rate": 2.3187612235331067e-07,
      "loss": 1.6188,
      "step": 3722
    },
    {
      "epoch": 0.5614961164316417,
      "grad_norm": 0.26423511951666934,
      "learning_rate": 2.3177148859541563e-07,
      "loss": 1.5885,
      "step": 3723
    },
    {
      "epoch": 0.5616469346203152,
      "grad_norm": 0.2857468589383398,
      "learning_rate": 2.3166686454358042e-07,
      "loss": 1.5766,
      "step": 3724
    },
    {
      "epoch": 0.5617977528089888,
      "grad_norm": 0.25503409850321557,
      "learning_rate": 2.3156225022129604e-07,
      "loss": 1.6271,
      "step": 3725
    },
    {
      "epoch": 0.5619485709976623,
      "grad_norm": 0.24713449835532042,
      "learning_rate": 2.3145764565205162e-07,
      "loss": 1.5937,
      "step": 3726
    },
    {
      "epoch": 0.5620993891863358,
      "grad_norm": 0.24672820850266292,
      "learning_rate": 2.3135305085933376e-07,
      "loss": 1.5304,
      "step": 3727
    },
    {
      "epoch": 0.5622502073750094,
      "grad_norm": 0.2517702226825073,
      "learning_rate": 2.3124846586662717e-07,
      "loss": 1.5882,
      "step": 3728
    },
    {
      "epoch": 0.562401025563683,
      "grad_norm": 0.25707062152749144,
      "learning_rate": 2.3114389069741418e-07,
      "loss": 1.5541,
      "step": 3729
    },
    {
      "epoch": 0.5625518437523566,
      "grad_norm": 0.26060636056281195,
      "learning_rate": 2.3103932537517495e-07,
      "loss": 1.5432,
      "step": 3730
    },
    {
      "epoch": 0.5627026619410301,
      "grad_norm": 0.2389548776377199,
      "learning_rate": 2.3093476992338744e-07,
      "loss": 1.5368,
      "step": 3731
    },
    {
      "epoch": 0.5628534801297036,
      "grad_norm": 0.2944855833008349,
      "learning_rate": 2.308302243655274e-07,
      "loss": 1.4542,
      "step": 3732
    },
    {
      "epoch": 0.5630042983183772,
      "grad_norm": 0.31719225087494396,
      "learning_rate": 2.3072568872506836e-07,
      "loss": 1.5473,
      "step": 3733
    },
    {
      "epoch": 0.5631551165070507,
      "grad_norm": 0.24864431603286263,
      "learning_rate": 2.3062116302548168e-07,
      "loss": 1.5101,
      "step": 3734
    },
    {
      "epoch": 0.5633059346957243,
      "grad_norm": 0.27002909684434634,
      "learning_rate": 2.3051664729023628e-07,
      "loss": 1.5838,
      "step": 3735
    },
    {
      "epoch": 0.5634567528843979,
      "grad_norm": 0.25453113539837574,
      "learning_rate": 2.3041214154279904e-07,
      "loss": 1.5544,
      "step": 3736
    },
    {
      "epoch": 0.5636075710730715,
      "grad_norm": 0.25301132100861035,
      "learning_rate": 2.3030764580663465e-07,
      "loss": 1.5701,
      "step": 3737
    },
    {
      "epoch": 0.5637583892617449,
      "grad_norm": 0.35461958294716706,
      "learning_rate": 2.302031601052053e-07,
      "loss": 1.5366,
      "step": 3738
    },
    {
      "epoch": 0.5639092074504185,
      "grad_norm": 0.256245998210831,
      "learning_rate": 2.3009868446197118e-07,
      "loss": 1.552,
      "step": 3739
    },
    {
      "epoch": 0.5640600256390921,
      "grad_norm": 0.24497442119592555,
      "learning_rate": 2.2999421890039005e-07,
      "loss": 1.5586,
      "step": 3740
    },
    {
      "epoch": 0.5642108438277657,
      "grad_norm": 0.25352179266618674,
      "learning_rate": 2.298897634439174e-07,
      "loss": 1.5696,
      "step": 3741
    },
    {
      "epoch": 0.5643616620164392,
      "grad_norm": 0.24358541538783612,
      "learning_rate": 2.2978531811600678e-07,
      "loss": 1.5355,
      "step": 3742
    },
    {
      "epoch": 0.5645124802051127,
      "grad_norm": 1.5337264757037743,
      "learning_rate": 2.296808829401089e-07,
      "loss": 1.5272,
      "step": 3743
    },
    {
      "epoch": 0.5646632983937863,
      "grad_norm": 0.24206462688464736,
      "learning_rate": 2.2957645793967274e-07,
      "loss": 1.5689,
      "step": 3744
    },
    {
      "epoch": 0.5648141165824598,
      "grad_norm": 0.27781314679882974,
      "learning_rate": 2.2947204313814456e-07,
      "loss": 1.6009,
      "step": 3745
    },
    {
      "epoch": 0.5649649347711334,
      "grad_norm": 0.8659978798118473,
      "learning_rate": 2.293676385589687e-07,
      "loss": 1.5952,
      "step": 3746
    },
    {
      "epoch": 0.565115752959807,
      "grad_norm": 0.7918369023779016,
      "learning_rate": 2.2926324422558691e-07,
      "loss": 1.4788,
      "step": 3747
    },
    {
      "epoch": 0.5652665711484806,
      "grad_norm": 0.2993582645953468,
      "learning_rate": 2.2915886016143875e-07,
      "loss": 1.5096,
      "step": 3748
    },
    {
      "epoch": 0.565417389337154,
      "grad_norm": 0.5632692862640049,
      "learning_rate": 2.2905448638996155e-07,
      "loss": 1.5725,
      "step": 3749
    },
    {
      "epoch": 0.5655682075258276,
      "grad_norm": 0.2518484130311377,
      "learning_rate": 2.2895012293459025e-07,
      "loss": 1.5019,
      "step": 3750
    },
    {
      "epoch": 0.5657190257145012,
      "grad_norm": 0.26977467638308,
      "learning_rate": 2.2884576981875747e-07,
      "loss": 1.5284,
      "step": 3751
    },
    {
      "epoch": 0.5658698439031747,
      "grad_norm": 0.25014250802983745,
      "learning_rate": 2.2874142706589356e-07,
      "loss": 1.5237,
      "step": 3752
    },
    {
      "epoch": 0.5660206620918483,
      "grad_norm": 0.24500688644136545,
      "learning_rate": 2.2863709469942643e-07,
      "loss": 1.57,
      "step": 3753
    },
    {
      "epoch": 0.5661714802805218,
      "grad_norm": 0.3467656004210753,
      "learning_rate": 2.2853277274278177e-07,
      "loss": 1.5379,
      "step": 3754
    },
    {
      "epoch": 0.5663222984691954,
      "grad_norm": 0.33106795675581974,
      "learning_rate": 2.2842846121938293e-07,
      "loss": 1.6001,
      "step": 3755
    },
    {
      "epoch": 0.5664731166578689,
      "grad_norm": 0.5379001383327541,
      "learning_rate": 2.2832416015265087e-07,
      "loss": 1.5852,
      "step": 3756
    },
    {
      "epoch": 0.5666239348465425,
      "grad_norm": 0.5228243427759623,
      "learning_rate": 2.2821986956600419e-07,
      "loss": 1.5025,
      "step": 3757
    },
    {
      "epoch": 0.5667747530352161,
      "grad_norm": 0.2395550928183924,
      "learning_rate": 2.281155894828592e-07,
      "loss": 1.5079,
      "step": 3758
    },
    {
      "epoch": 0.5669255712238896,
      "grad_norm": 0.2571340962141452,
      "learning_rate": 2.280113199266297e-07,
      "loss": 1.5392,
      "step": 3759
    },
    {
      "epoch": 0.5670763894125631,
      "grad_norm": 0.27627938424007076,
      "learning_rate": 2.2790706092072742e-07,
      "loss": 1.577,
      "step": 3760
    },
    {
      "epoch": 0.5672272076012367,
      "grad_norm": 0.2735106100048377,
      "learning_rate": 2.2780281248856148e-07,
      "loss": 1.5253,
      "step": 3761
    },
    {
      "epoch": 0.5673780257899103,
      "grad_norm": 0.24565529996766425,
      "learning_rate": 2.2769857465353864e-07,
      "loss": 1.5414,
      "step": 3762
    },
    {
      "epoch": 0.5675288439785838,
      "grad_norm": 0.25911738100127885,
      "learning_rate": 2.2759434743906337e-07,
      "loss": 1.5168,
      "step": 3763
    },
    {
      "epoch": 0.5676796621672574,
      "grad_norm": 0.26083242181688693,
      "learning_rate": 2.274901308685377e-07,
      "loss": 1.5698,
      "step": 3764
    },
    {
      "epoch": 0.567830480355931,
      "grad_norm": 0.253637562025555,
      "learning_rate": 2.273859249653613e-07,
      "loss": 1.5868,
      "step": 3765
    },
    {
      "epoch": 0.5679812985446044,
      "grad_norm": 0.24777184255839774,
      "learning_rate": 2.2728172975293135e-07,
      "loss": 1.5206,
      "step": 3766
    },
    {
      "epoch": 0.568132116733278,
      "grad_norm": 0.30430013138718626,
      "learning_rate": 2.2717754525464284e-07,
      "loss": 1.5847,
      "step": 3767
    },
    {
      "epoch": 0.5682829349219516,
      "grad_norm": 0.2771173974071153,
      "learning_rate": 2.2707337149388822e-07,
      "loss": 1.5594,
      "step": 3768
    },
    {
      "epoch": 0.5684337531106252,
      "grad_norm": 0.24679284719922462,
      "learning_rate": 2.2696920849405742e-07,
      "loss": 1.5829,
      "step": 3769
    },
    {
      "epoch": 0.5685845712992987,
      "grad_norm": 0.2541246356975005,
      "learning_rate": 2.2686505627853813e-07,
      "loss": 1.5748,
      "step": 3770
    },
    {
      "epoch": 0.5687353894879722,
      "grad_norm": 0.29219160832881896,
      "learning_rate": 2.2676091487071552e-07,
      "loss": 1.5848,
      "step": 3771
    },
    {
      "epoch": 0.5688862076766458,
      "grad_norm": 0.38956183827483026,
      "learning_rate": 2.266567842939724e-07,
      "loss": 1.5496,
      "step": 3772
    },
    {
      "epoch": 0.5690370258653193,
      "grad_norm": 0.2506659621824879,
      "learning_rate": 2.265526645716892e-07,
      "loss": 1.5326,
      "step": 3773
    },
    {
      "epoch": 0.5691878440539929,
      "grad_norm": 0.2679515485132347,
      "learning_rate": 2.2644855572724363e-07,
      "loss": 1.5572,
      "step": 3774
    },
    {
      "epoch": 0.5693386622426665,
      "grad_norm": 0.2579569141417791,
      "learning_rate": 2.2634445778401134e-07,
      "loss": 1.5749,
      "step": 3775
    },
    {
      "epoch": 0.5694894804313401,
      "grad_norm": 0.250645661902425,
      "learning_rate": 2.262403707653653e-07,
      "loss": 1.5957,
      "step": 3776
    },
    {
      "epoch": 0.5696402986200135,
      "grad_norm": 0.6242568522969492,
      "learning_rate": 2.2613629469467598e-07,
      "loss": 1.6399,
      "step": 3777
    },
    {
      "epoch": 0.5697911168086871,
      "grad_norm": 0.24429435285297052,
      "learning_rate": 2.2603222959531165e-07,
      "loss": 1.5638,
      "step": 3778
    },
    {
      "epoch": 0.5699419349973607,
      "grad_norm": 0.26010481690712356,
      "learning_rate": 2.2592817549063786e-07,
      "loss": 1.5593,
      "step": 3779
    },
    {
      "epoch": 0.5700927531860343,
      "grad_norm": 0.24442274456846919,
      "learning_rate": 2.2582413240401778e-07,
      "loss": 1.5124,
      "step": 3780
    },
    {
      "epoch": 0.5702435713747078,
      "grad_norm": 0.2423865915093295,
      "learning_rate": 2.2572010035881216e-07,
      "loss": 1.5678,
      "step": 3781
    },
    {
      "epoch": 0.5703943895633814,
      "grad_norm": 0.25110373866973795,
      "learning_rate": 2.2561607937837917e-07,
      "loss": 1.5525,
      "step": 3782
    },
    {
      "epoch": 0.5705452077520549,
      "grad_norm": 0.25941477053110074,
      "learning_rate": 2.2551206948607465e-07,
      "loss": 1.6247,
      "step": 3783
    },
    {
      "epoch": 0.5706960259407284,
      "grad_norm": 0.2577626131281058,
      "learning_rate": 2.2540807070525164e-07,
      "loss": 1.5449,
      "step": 3784
    },
    {
      "epoch": 0.570846844129402,
      "grad_norm": 0.24301635040369168,
      "learning_rate": 2.2530408305926113e-07,
      "loss": 1.49,
      "step": 3785
    },
    {
      "epoch": 0.5709976623180756,
      "grad_norm": 0.2630750900131234,
      "learning_rate": 2.2520010657145127e-07,
      "loss": 1.5623,
      "step": 3786
    },
    {
      "epoch": 0.5711484805067492,
      "grad_norm": 0.24920024775254504,
      "learning_rate": 2.250961412651678e-07,
      "loss": 1.6354,
      "step": 3787
    },
    {
      "epoch": 0.5712992986954226,
      "grad_norm": 0.2787379757097677,
      "learning_rate": 2.2499218716375402e-07,
      "loss": 1.578,
      "step": 3788
    },
    {
      "epoch": 0.5714501168840962,
      "grad_norm": 0.2534437872582565,
      "learning_rate": 2.2488824429055055e-07,
      "loss": 1.5623,
      "step": 3789
    },
    {
      "epoch": 0.5716009350727698,
      "grad_norm": 0.26891581634427875,
      "learning_rate": 2.2478431266889565e-07,
      "loss": 1.5497,
      "step": 3790
    },
    {
      "epoch": 0.5717517532614433,
      "grad_norm": 0.24354950530761477,
      "learning_rate": 2.2468039232212506e-07,
      "loss": 1.5665,
      "step": 3791
    },
    {
      "epoch": 0.5719025714501169,
      "grad_norm": 0.26367058200099974,
      "learning_rate": 2.245764832735718e-07,
      "loss": 1.6466,
      "step": 3792
    },
    {
      "epoch": 0.5720533896387905,
      "grad_norm": 0.2873797075741446,
      "learning_rate": 2.2447258554656657e-07,
      "loss": 1.5831,
      "step": 3793
    },
    {
      "epoch": 0.572204207827464,
      "grad_norm": 0.25887675924205883,
      "learning_rate": 2.243686991644374e-07,
      "loss": 1.5302,
      "step": 3794
    },
    {
      "epoch": 0.5723550260161375,
      "grad_norm": 0.24775356955881006,
      "learning_rate": 2.2426482415050984e-07,
      "loss": 1.5206,
      "step": 3795
    },
    {
      "epoch": 0.5725058442048111,
      "grad_norm": 0.2609640185767476,
      "learning_rate": 2.2416096052810684e-07,
      "loss": 1.6058,
      "step": 3796
    },
    {
      "epoch": 0.5726566623934847,
      "grad_norm": 0.2541846246845802,
      "learning_rate": 2.2405710832054874e-07,
      "loss": 1.6014,
      "step": 3797
    },
    {
      "epoch": 0.5728074805821582,
      "grad_norm": 0.24060146349943168,
      "learning_rate": 2.2395326755115347e-07,
      "loss": 1.5616,
      "step": 3798
    },
    {
      "epoch": 0.5729582987708317,
      "grad_norm": 0.2699389201639666,
      "learning_rate": 2.2384943824323632e-07,
      "loss": 1.4967,
      "step": 3799
    },
    {
      "epoch": 0.5731091169595053,
      "grad_norm": 0.2590233871468359,
      "learning_rate": 2.2374562042010986e-07,
      "loss": 1.5868,
      "step": 3800
    },
    {
      "epoch": 0.5732599351481789,
      "grad_norm": 0.2500973098596572,
      "learning_rate": 2.2364181410508441e-07,
      "loss": 1.5975,
      "step": 3801
    },
    {
      "epoch": 0.5734107533368524,
      "grad_norm": 0.24656651461348164,
      "learning_rate": 2.235380193214673e-07,
      "loss": 1.5572,
      "step": 3802
    },
    {
      "epoch": 0.573561571525526,
      "grad_norm": 0.29060161703836196,
      "learning_rate": 2.2343423609256368e-07,
      "loss": 1.5095,
      "step": 3803
    },
    {
      "epoch": 0.5737123897141996,
      "grad_norm": 0.25440387679712745,
      "learning_rate": 2.233304644416758e-07,
      "loss": 1.5419,
      "step": 3804
    },
    {
      "epoch": 0.573863207902873,
      "grad_norm": 0.25823801268964536,
      "learning_rate": 2.2322670439210339e-07,
      "loss": 1.5465,
      "step": 3805
    },
    {
      "epoch": 0.5740140260915466,
      "grad_norm": 0.28298262211840103,
      "learning_rate": 2.231229559671437e-07,
      "loss": 1.479,
      "step": 3806
    },
    {
      "epoch": 0.5741648442802202,
      "grad_norm": 0.2474312053351583,
      "learning_rate": 2.230192191900912e-07,
      "loss": 1.577,
      "step": 3807
    },
    {
      "epoch": 0.5743156624688938,
      "grad_norm": 0.2577236463799376,
      "learning_rate": 2.2291549408423788e-07,
      "loss": 1.5948,
      "step": 3808
    },
    {
      "epoch": 0.5744664806575673,
      "grad_norm": 0.2405755948683268,
      "learning_rate": 2.22811780672873e-07,
      "loss": 1.5225,
      "step": 3809
    },
    {
      "epoch": 0.5746172988462409,
      "grad_norm": 0.27403598796818784,
      "learning_rate": 2.2270807897928317e-07,
      "loss": 1.6322,
      "step": 3810
    },
    {
      "epoch": 0.5747681170349144,
      "grad_norm": 0.24616655581097505,
      "learning_rate": 2.2260438902675254e-07,
      "loss": 1.5112,
      "step": 3811
    },
    {
      "epoch": 0.574918935223588,
      "grad_norm": 0.26025260418138463,
      "learning_rate": 2.2250071083856259e-07,
      "loss": 1.5589,
      "step": 3812
    },
    {
      "epoch": 0.5750697534122615,
      "grad_norm": 0.2533320997307876,
      "learning_rate": 2.223970444379919e-07,
      "loss": 1.578,
      "step": 3813
    },
    {
      "epoch": 0.5752205716009351,
      "grad_norm": 0.6182973675590029,
      "learning_rate": 2.222933898483168e-07,
      "loss": 1.5124,
      "step": 3814
    },
    {
      "epoch": 0.5753713897896087,
      "grad_norm": 0.2607767696742802,
      "learning_rate": 2.2218974709281058e-07,
      "loss": 1.5282,
      "step": 3815
    },
    {
      "epoch": 0.5755222079782821,
      "grad_norm": 1.746288228687721,
      "learning_rate": 2.2208611619474414e-07,
      "loss": 1.5935,
      "step": 3816
    },
    {
      "epoch": 0.5756730261669557,
      "grad_norm": 0.2781940675906223,
      "learning_rate": 2.219824971773857e-07,
      "loss": 1.5586,
      "step": 3817
    },
    {
      "epoch": 0.5758238443556293,
      "grad_norm": 0.2666272791271762,
      "learning_rate": 2.2187889006400067e-07,
      "loss": 1.5286,
      "step": 3818
    },
    {
      "epoch": 0.5759746625443029,
      "grad_norm": 0.24416966759777037,
      "learning_rate": 2.2177529487785184e-07,
      "loss": 1.5812,
      "step": 3819
    },
    {
      "epoch": 0.5761254807329764,
      "grad_norm": 0.4255803719795962,
      "learning_rate": 2.2167171164219945e-07,
      "loss": 1.5936,
      "step": 3820
    },
    {
      "epoch": 0.57627629892165,
      "grad_norm": 0.28413577943475554,
      "learning_rate": 2.215681403803008e-07,
      "loss": 1.5585,
      "step": 3821
    },
    {
      "epoch": 0.5764271171103235,
      "grad_norm": 0.45129086175053656,
      "learning_rate": 2.2146458111541083e-07,
      "loss": 1.5197,
      "step": 3822
    },
    {
      "epoch": 0.576577935298997,
      "grad_norm": 0.2698861269924893,
      "learning_rate": 2.2136103387078148e-07,
      "loss": 1.5499,
      "step": 3823
    },
    {
      "epoch": 0.5767287534876706,
      "grad_norm": 0.24600013904480686,
      "learning_rate": 2.2125749866966214e-07,
      "loss": 1.5971,
      "step": 3824
    },
    {
      "epoch": 0.5768795716763442,
      "grad_norm": 0.24129779490920517,
      "learning_rate": 2.2115397553529958e-07,
      "loss": 1.5806,
      "step": 3825
    },
    {
      "epoch": 0.5770303898650178,
      "grad_norm": 0.29701513447079625,
      "learning_rate": 2.2105046449093762e-07,
      "loss": 1.6232,
      "step": 3826
    },
    {
      "epoch": 0.5771812080536913,
      "grad_norm": 0.7615362675290088,
      "learning_rate": 2.2094696555981763e-07,
      "loss": 1.5411,
      "step": 3827
    },
    {
      "epoch": 0.5773320262423648,
      "grad_norm": 0.2413044971417661,
      "learning_rate": 2.2084347876517796e-07,
      "loss": 1.4892,
      "step": 3828
    },
    {
      "epoch": 0.5774828444310384,
      "grad_norm": 0.24703204099069684,
      "learning_rate": 2.2074000413025457e-07,
      "loss": 1.5733,
      "step": 3829
    },
    {
      "epoch": 0.5776336626197119,
      "grad_norm": 0.25368532396740934,
      "learning_rate": 2.206365416782805e-07,
      "loss": 1.6596,
      "step": 3830
    },
    {
      "epoch": 0.5777844808083855,
      "grad_norm": 0.2737552950500867,
      "learning_rate": 2.20533091432486e-07,
      "loss": 1.6007,
      "step": 3831
    },
    {
      "epoch": 0.5779352989970591,
      "grad_norm": 0.24841815713314006,
      "learning_rate": 2.204296534160988e-07,
      "loss": 1.5218,
      "step": 3832
    },
    {
      "epoch": 0.5780861171857326,
      "grad_norm": 0.2555444899340187,
      "learning_rate": 2.2032622765234355e-07,
      "loss": 1.5312,
      "step": 3833
    },
    {
      "epoch": 0.5782369353744061,
      "grad_norm": 0.25121951884051724,
      "learning_rate": 2.202228141644426e-07,
      "loss": 1.5525,
      "step": 3834
    },
    {
      "epoch": 0.5783877535630797,
      "grad_norm": 0.2826540264763573,
      "learning_rate": 2.2011941297561514e-07,
      "loss": 1.561,
      "step": 3835
    },
    {
      "epoch": 0.5785385717517533,
      "grad_norm": 0.24319345438007933,
      "learning_rate": 2.200160241090777e-07,
      "loss": 1.5542,
      "step": 3836
    },
    {
      "epoch": 0.5786893899404268,
      "grad_norm": 0.2542961847052196,
      "learning_rate": 2.1991264758804424e-07,
      "loss": 1.5275,
      "step": 3837
    },
    {
      "epoch": 0.5788402081291004,
      "grad_norm": 0.2626142301729653,
      "learning_rate": 2.1980928343572569e-07,
      "loss": 1.5552,
      "step": 3838
    },
    {
      "epoch": 0.5789910263177739,
      "grad_norm": 0.2716722650906108,
      "learning_rate": 2.1970593167533035e-07,
      "loss": 1.5288,
      "step": 3839
    },
    {
      "epoch": 0.5791418445064475,
      "grad_norm": 0.24899949433714882,
      "learning_rate": 2.196025923300636e-07,
      "loss": 1.5404,
      "step": 3840
    },
    {
      "epoch": 0.579292662695121,
      "grad_norm": 0.25354168178863573,
      "learning_rate": 2.1949926542312835e-07,
      "loss": 1.5423,
      "step": 3841
    },
    {
      "epoch": 0.5794434808837946,
      "grad_norm": 0.23854730518745276,
      "learning_rate": 2.1939595097772435e-07,
      "loss": 1.5197,
      "step": 3842
    },
    {
      "epoch": 0.5795942990724682,
      "grad_norm": 0.2430693716168873,
      "learning_rate": 2.1929264901704863e-07,
      "loss": 1.4972,
      "step": 3843
    },
    {
      "epoch": 0.5797451172611416,
      "grad_norm": 0.24301961919786527,
      "learning_rate": 2.1918935956429568e-07,
      "loss": 1.5387,
      "step": 3844
    },
    {
      "epoch": 0.5798959354498152,
      "grad_norm": 0.2524188769060236,
      "learning_rate": 2.1908608264265688e-07,
      "loss": 1.4952,
      "step": 3845
    },
    {
      "epoch": 0.5800467536384888,
      "grad_norm": 0.263113455281279,
      "learning_rate": 2.1898281827532088e-07,
      "loss": 1.5065,
      "step": 3846
    },
    {
      "epoch": 0.5801975718271624,
      "grad_norm": 0.2705355413432033,
      "learning_rate": 2.1887956648547362e-07,
      "loss": 1.6136,
      "step": 3847
    },
    {
      "epoch": 0.5803483900158359,
      "grad_norm": 0.2970875459648603,
      "learning_rate": 2.1877632729629802e-07,
      "loss": 1.5707,
      "step": 3848
    },
    {
      "epoch": 0.5804992082045095,
      "grad_norm": 0.2593001241846887,
      "learning_rate": 2.1867310073097438e-07,
      "loss": 1.5253,
      "step": 3849
    },
    {
      "epoch": 0.580650026393183,
      "grad_norm": 0.2569704845840797,
      "learning_rate": 2.1856988681268014e-07,
      "loss": 1.5755,
      "step": 3850
    },
    {
      "epoch": 0.5808008445818565,
      "grad_norm": 0.2555578317437223,
      "learning_rate": 2.1846668556458964e-07,
      "loss": 1.5297,
      "step": 3851
    },
    {
      "epoch": 0.5809516627705301,
      "grad_norm": 0.24976164652961383,
      "learning_rate": 2.1836349700987472e-07,
      "loss": 1.5918,
      "step": 3852
    },
    {
      "epoch": 0.5811024809592037,
      "grad_norm": 0.25412501981978447,
      "learning_rate": 2.182603211717041e-07,
      "loss": 1.5667,
      "step": 3853
    },
    {
      "epoch": 0.5812532991478773,
      "grad_norm": 0.24643386448414026,
      "learning_rate": 2.1815715807324387e-07,
      "loss": 1.5016,
      "step": 3854
    },
    {
      "epoch": 0.5814041173365508,
      "grad_norm": 0.291967919185025,
      "learning_rate": 2.180540077376571e-07,
      "loss": 1.5208,
      "step": 3855
    },
    {
      "epoch": 0.5815549355252243,
      "grad_norm": 0.2607898661757403,
      "learning_rate": 2.17950870188104e-07,
      "loss": 1.6324,
      "step": 3856
    },
    {
      "epoch": 0.5817057537138979,
      "grad_norm": 0.3237350230718295,
      "learning_rate": 2.1784774544774205e-07,
      "loss": 1.5811,
      "step": 3857
    },
    {
      "epoch": 0.5818565719025715,
      "grad_norm": 0.25465789432054664,
      "learning_rate": 2.177446335397256e-07,
      "loss": 1.6522,
      "step": 3858
    },
    {
      "epoch": 0.582007390091245,
      "grad_norm": 0.25650077735120685,
      "learning_rate": 2.1764153448720646e-07,
      "loss": 1.5092,
      "step": 3859
    },
    {
      "epoch": 0.5821582082799186,
      "grad_norm": 0.2556223473650482,
      "learning_rate": 2.1753844831333335e-07,
      "loss": 1.4928,
      "step": 3860
    },
    {
      "epoch": 0.5823090264685921,
      "grad_norm": 0.2647938026212078,
      "learning_rate": 2.1743537504125199e-07,
      "loss": 1.6117,
      "step": 3861
    },
    {
      "epoch": 0.5824598446572656,
      "grad_norm": 0.28386285783857657,
      "learning_rate": 2.173323146941054e-07,
      "loss": 1.5646,
      "step": 3862
    },
    {
      "epoch": 0.5826106628459392,
      "grad_norm": 0.25605121004622367,
      "learning_rate": 2.1722926729503365e-07,
      "loss": 1.655,
      "step": 3863
    },
    {
      "epoch": 0.5827614810346128,
      "grad_norm": 0.347275229460256,
      "learning_rate": 2.171262328671738e-07,
      "loss": 1.6725,
      "step": 3864
    },
    {
      "epoch": 0.5829122992232864,
      "grad_norm": 0.25644950265792565,
      "learning_rate": 2.170232114336602e-07,
      "loss": 1.563,
      "step": 3865
    },
    {
      "epoch": 0.5830631174119599,
      "grad_norm": 0.2514527888535762,
      "learning_rate": 2.169202030176241e-07,
      "loss": 1.6298,
      "step": 3866
    },
    {
      "epoch": 0.5832139356006334,
      "grad_norm": 0.25577460819356496,
      "learning_rate": 2.1681720764219384e-07,
      "loss": 1.5562,
      "step": 3867
    },
    {
      "epoch": 0.583364753789307,
      "grad_norm": 0.26046606047108944,
      "learning_rate": 2.1671422533049504e-07,
      "loss": 1.4611,
      "step": 3868
    },
    {
      "epoch": 0.5835155719779805,
      "grad_norm": 0.23726670710428088,
      "learning_rate": 2.1661125610565006e-07,
      "loss": 1.5789,
      "step": 3869
    },
    {
      "epoch": 0.5836663901666541,
      "grad_norm": 0.2669902319815929,
      "learning_rate": 2.1650829999077856e-07,
      "loss": 1.6146,
      "step": 3870
    },
    {
      "epoch": 0.5838172083553277,
      "grad_norm": 0.24610296922448516,
      "learning_rate": 2.1640535700899714e-07,
      "loss": 1.5298,
      "step": 3871
    },
    {
      "epoch": 0.5839680265440013,
      "grad_norm": 0.2590328572776037,
      "learning_rate": 2.1630242718341957e-07,
      "loss": 1.5387,
      "step": 3872
    },
    {
      "epoch": 0.5841188447326747,
      "grad_norm": 0.2628650412755619,
      "learning_rate": 2.1619951053715657e-07,
      "loss": 1.6035,
      "step": 3873
    },
    {
      "epoch": 0.5842696629213483,
      "grad_norm": 0.3192328346924656,
      "learning_rate": 2.160966070933159e-07,
      "loss": 1.5955,
      "step": 3874
    },
    {
      "epoch": 0.5844204811100219,
      "grad_norm": 0.2526909035008379,
      "learning_rate": 2.1599371687500245e-07,
      "loss": 1.5603,
      "step": 3875
    },
    {
      "epoch": 0.5845712992986954,
      "grad_norm": 0.26297610180789677,
      "learning_rate": 2.1589083990531793e-07,
      "loss": 1.5714,
      "step": 3876
    },
    {
      "epoch": 0.584722117487369,
      "grad_norm": 0.3217290084400424,
      "learning_rate": 2.1578797620736133e-07,
      "loss": 1.5672,
      "step": 3877
    },
    {
      "epoch": 0.5848729356760425,
      "grad_norm": 0.3162169538147139,
      "learning_rate": 2.1568512580422855e-07,
      "loss": 1.4855,
      "step": 3878
    },
    {
      "epoch": 0.5850237538647161,
      "grad_norm": 0.2516293552966027,
      "learning_rate": 2.155822887190124e-07,
      "loss": 1.5633,
      "step": 3879
    },
    {
      "epoch": 0.5851745720533896,
      "grad_norm": 0.29894170183508184,
      "learning_rate": 2.154794649748029e-07,
      "loss": 1.5917,
      "step": 3880
    },
    {
      "epoch": 0.5853253902420632,
      "grad_norm": 0.2682481575960108,
      "learning_rate": 2.1537665459468696e-07,
      "loss": 1.5779,
      "step": 3881
    },
    {
      "epoch": 0.5854762084307368,
      "grad_norm": 0.25103199650586117,
      "learning_rate": 2.152738576017485e-07,
      "loss": 1.5169,
      "step": 3882
    },
    {
      "epoch": 0.5856270266194104,
      "grad_norm": 0.25201957555043975,
      "learning_rate": 2.1517107401906842e-07,
      "loss": 1.5541,
      "step": 3883
    },
    {
      "epoch": 0.5857778448080838,
      "grad_norm": 0.8384191903433962,
      "learning_rate": 2.1506830386972458e-07,
      "loss": 1.587,
      "step": 3884
    },
    {
      "epoch": 0.5859286629967574,
      "grad_norm": 0.2488293562239644,
      "learning_rate": 2.1496554717679194e-07,
      "loss": 1.5721,
      "step": 3885
    },
    {
      "epoch": 0.586079481185431,
      "grad_norm": 0.2368679296727655,
      "learning_rate": 2.1486280396334239e-07,
      "loss": 1.5324,
      "step": 3886
    },
    {
      "epoch": 0.5862302993741045,
      "grad_norm": 0.24897976851256423,
      "learning_rate": 2.147600742524447e-07,
      "loss": 1.54,
      "step": 3887
    },
    {
      "epoch": 0.5863811175627781,
      "grad_norm": 0.3002451271596729,
      "learning_rate": 2.1465735806716478e-07,
      "loss": 1.5172,
      "step": 3888
    },
    {
      "epoch": 0.5865319357514516,
      "grad_norm": 0.24850770735669614,
      "learning_rate": 2.1455465543056527e-07,
      "loss": 1.5526,
      "step": 3889
    },
    {
      "epoch": 0.5866827539401251,
      "grad_norm": 0.4619670490548418,
      "learning_rate": 2.14451966365706e-07,
      "loss": 1.5955,
      "step": 3890
    },
    {
      "epoch": 0.5868335721287987,
      "grad_norm": 0.43523556266765423,
      "learning_rate": 2.1434929089564375e-07,
      "loss": 1.6064,
      "step": 3891
    },
    {
      "epoch": 0.5869843903174723,
      "grad_norm": 0.23674067638948412,
      "learning_rate": 2.1424662904343188e-07,
      "loss": 1.5527,
      "step": 3892
    },
    {
      "epoch": 0.5871352085061459,
      "grad_norm": 0.25585996624482177,
      "learning_rate": 2.1414398083212114e-07,
      "loss": 1.6381,
      "step": 3893
    },
    {
      "epoch": 0.5872860266948194,
      "grad_norm": 0.3467997914699444,
      "learning_rate": 2.1404134628475912e-07,
      "loss": 1.4996,
      "step": 3894
    },
    {
      "epoch": 0.5874368448834929,
      "grad_norm": 0.27717865650431744,
      "learning_rate": 2.139387254243901e-07,
      "loss": 1.542,
      "step": 3895
    },
    {
      "epoch": 0.5875876630721665,
      "grad_norm": 0.24240472352347092,
      "learning_rate": 2.1383611827405557e-07,
      "loss": 1.5739,
      "step": 3896
    },
    {
      "epoch": 0.58773848126084,
      "grad_norm": 0.2550797345859492,
      "learning_rate": 2.1373352485679375e-07,
      "loss": 1.5743,
      "step": 3897
    },
    {
      "epoch": 0.5878892994495136,
      "grad_norm": 0.25890179909780026,
      "learning_rate": 2.1363094519563984e-07,
      "loss": 1.6189,
      "step": 3898
    },
    {
      "epoch": 0.5880401176381872,
      "grad_norm": 0.3744524860252214,
      "learning_rate": 2.1352837931362605e-07,
      "loss": 1.615,
      "step": 3899
    },
    {
      "epoch": 0.5881909358268608,
      "grad_norm": 0.27406405524412275,
      "learning_rate": 2.134258272337814e-07,
      "loss": 1.5882,
      "step": 3900
    },
    {
      "epoch": 0.5883417540155342,
      "grad_norm": 0.25618129568741593,
      "learning_rate": 2.133232889791317e-07,
      "loss": 1.5719,
      "step": 3901
    },
    {
      "epoch": 0.5884925722042078,
      "grad_norm": 0.5344404015159859,
      "learning_rate": 2.1322076457269983e-07,
      "loss": 1.5345,
      "step": 3902
    },
    {
      "epoch": 0.5886433903928814,
      "grad_norm": 0.27394922769442587,
      "learning_rate": 2.1311825403750554e-07,
      "loss": 1.5363,
      "step": 3903
    },
    {
      "epoch": 0.588794208581555,
      "grad_norm": 0.2516339285494544,
      "learning_rate": 2.1301575739656547e-07,
      "loss": 1.5915,
      "step": 3904
    },
    {
      "epoch": 0.5889450267702285,
      "grad_norm": 0.25531568449077874,
      "learning_rate": 2.12913274672893e-07,
      "loss": 1.6105,
      "step": 3905
    },
    {
      "epoch": 0.589095844958902,
      "grad_norm": 0.25052078610351264,
      "learning_rate": 2.1281080588949851e-07,
      "loss": 1.6057,
      "step": 3906
    },
    {
      "epoch": 0.5892466631475756,
      "grad_norm": 0.2875496049149756,
      "learning_rate": 2.127083510693893e-07,
      "loss": 1.4833,
      "step": 3907
    },
    {
      "epoch": 0.5893974813362491,
      "grad_norm": 0.26066187862838874,
      "learning_rate": 2.126059102355694e-07,
      "loss": 1.5166,
      "step": 3908
    },
    {
      "epoch": 0.5895482995249227,
      "grad_norm": 0.24066311208593907,
      "learning_rate": 2.1250348341103974e-07,
      "loss": 1.5614,
      "step": 3909
    },
    {
      "epoch": 0.5896991177135963,
      "grad_norm": 0.2527211765328934,
      "learning_rate": 2.1240107061879813e-07,
      "loss": 1.5479,
      "step": 3910
    },
    {
      "epoch": 0.5898499359022699,
      "grad_norm": 0.24507118283919913,
      "learning_rate": 2.1229867188183932e-07,
      "loss": 1.5131,
      "step": 3911
    },
    {
      "epoch": 0.5900007540909433,
      "grad_norm": 0.24660986393765302,
      "learning_rate": 2.1219628722315468e-07,
      "loss": 1.5353,
      "step": 3912
    },
    {
      "epoch": 0.5901515722796169,
      "grad_norm": 0.27213896453072617,
      "learning_rate": 2.1209391666573262e-07,
      "loss": 1.5589,
      "step": 3913
    },
    {
      "epoch": 0.5903023904682905,
      "grad_norm": 0.26620695172163944,
      "learning_rate": 2.1199156023255838e-07,
      "loss": 1.5554,
      "step": 3914
    },
    {
      "epoch": 0.590453208656964,
      "grad_norm": 0.25728273695387793,
      "learning_rate": 2.1188921794661373e-07,
      "loss": 1.535,
      "step": 3915
    },
    {
      "epoch": 0.5906040268456376,
      "grad_norm": 0.2489799042047875,
      "learning_rate": 2.1178688983087774e-07,
      "loss": 1.5579,
      "step": 3916
    },
    {
      "epoch": 0.5907548450343112,
      "grad_norm": 0.25706920572749437,
      "learning_rate": 2.1168457590832594e-07,
      "loss": 1.6016,
      "step": 3917
    },
    {
      "epoch": 0.5909056632229847,
      "grad_norm": 0.25777099994890706,
      "learning_rate": 2.1158227620193077e-07,
      "loss": 1.5538,
      "step": 3918
    },
    {
      "epoch": 0.5910564814116582,
      "grad_norm": 0.8221472064317823,
      "learning_rate": 2.1147999073466156e-07,
      "loss": 1.5954,
      "step": 3919
    },
    {
      "epoch": 0.5912072996003318,
      "grad_norm": 0.2636602003466519,
      "learning_rate": 2.1137771952948425e-07,
      "loss": 1.5922,
      "step": 3920
    },
    {
      "epoch": 0.5913581177890054,
      "grad_norm": 0.2476386765784861,
      "learning_rate": 2.1127546260936185e-07,
      "loss": 1.5343,
      "step": 3921
    },
    {
      "epoch": 0.591508935977679,
      "grad_norm": 0.32515634295585055,
      "learning_rate": 2.1117321999725396e-07,
      "loss": 1.5778,
      "step": 3922
    },
    {
      "epoch": 0.5916597541663524,
      "grad_norm": 0.26955179218666003,
      "learning_rate": 2.1107099171611704e-07,
      "loss": 1.604,
      "step": 3923
    },
    {
      "epoch": 0.591810572355026,
      "grad_norm": 0.26142776224103237,
      "learning_rate": 2.109687777889042e-07,
      "loss": 1.5378,
      "step": 3924
    },
    {
      "epoch": 0.5919613905436996,
      "grad_norm": 0.25235168002590846,
      "learning_rate": 2.1086657823856567e-07,
      "loss": 1.5882,
      "step": 3925
    },
    {
      "epoch": 0.5921122087323731,
      "grad_norm": 0.3268180416647588,
      "learning_rate": 2.1076439308804806e-07,
      "loss": 1.5245,
      "step": 3926
    },
    {
      "epoch": 0.5922630269210467,
      "grad_norm": 0.25122857572439317,
      "learning_rate": 2.1066222236029496e-07,
      "loss": 1.53,
      "step": 3927
    },
    {
      "epoch": 0.5924138451097203,
      "grad_norm": 0.2518402680010454,
      "learning_rate": 2.105600660782466e-07,
      "loss": 1.5644,
      "step": 3928
    },
    {
      "epoch": 0.5925646632983937,
      "grad_norm": 0.6317106903102638,
      "learning_rate": 2.1045792426484016e-07,
      "loss": 1.5885,
      "step": 3929
    },
    {
      "epoch": 0.5927154814870673,
      "grad_norm": 0.23885732382701932,
      "learning_rate": 2.1035579694300936e-07,
      "loss": 1.5454,
      "step": 3930
    },
    {
      "epoch": 0.5928662996757409,
      "grad_norm": 0.24169605774079195,
      "learning_rate": 2.102536841356848e-07,
      "loss": 1.5475,
      "step": 3931
    },
    {
      "epoch": 0.5930171178644145,
      "grad_norm": 0.2496554472407174,
      "learning_rate": 2.1015158586579385e-07,
      "loss": 1.5379,
      "step": 3932
    },
    {
      "epoch": 0.593167936053088,
      "grad_norm": 0.23619446956023743,
      "learning_rate": 2.1004950215626032e-07,
      "loss": 1.5518,
      "step": 3933
    },
    {
      "epoch": 0.5933187542417615,
      "grad_norm": 0.24512528344650525,
      "learning_rate": 2.0994743303000522e-07,
      "loss": 1.5636,
      "step": 3934
    },
    {
      "epoch": 0.5934695724304351,
      "grad_norm": 0.29386974334811894,
      "learning_rate": 2.0984537850994595e-07,
      "loss": 1.5043,
      "step": 3935
    },
    {
      "epoch": 0.5936203906191087,
      "grad_norm": 0.24607778184964832,
      "learning_rate": 2.0974333861899668e-07,
      "loss": 1.6379,
      "step": 3936
    },
    {
      "epoch": 0.5937712088077822,
      "grad_norm": 0.2537985300545943,
      "learning_rate": 2.0964131338006836e-07,
      "loss": 1.5556,
      "step": 3937
    },
    {
      "epoch": 0.5939220269964558,
      "grad_norm": 0.2938624465908146,
      "learning_rate": 2.0953930281606875e-07,
      "loss": 1.6214,
      "step": 3938
    },
    {
      "epoch": 0.5940728451851294,
      "grad_norm": 0.2575265487300153,
      "learning_rate": 2.0943730694990197e-07,
      "loss": 1.5301,
      "step": 3939
    },
    {
      "epoch": 0.5942236633738028,
      "grad_norm": 0.25235912105681335,
      "learning_rate": 2.0933532580446923e-07,
      "loss": 1.5302,
      "step": 3940
    },
    {
      "epoch": 0.5943744815624764,
      "grad_norm": 0.26834261526057135,
      "learning_rate": 2.0923335940266817e-07,
      "loss": 1.4762,
      "step": 3941
    },
    {
      "epoch": 0.59452529975115,
      "grad_norm": 0.274112552646199,
      "learning_rate": 2.091314077673933e-07,
      "loss": 1.5406,
      "step": 3942
    },
    {
      "epoch": 0.5946761179398236,
      "grad_norm": 0.25979791178173584,
      "learning_rate": 2.0902947092153566e-07,
      "loss": 1.549,
      "step": 3943
    },
    {
      "epoch": 0.5948269361284971,
      "grad_norm": 0.24123835415858874,
      "learning_rate": 2.0892754888798308e-07,
      "loss": 1.5285,
      "step": 3944
    },
    {
      "epoch": 0.5949777543171707,
      "grad_norm": 0.2833528756405838,
      "learning_rate": 2.0882564168962007e-07,
      "loss": 1.5937,
      "step": 3945
    },
    {
      "epoch": 0.5951285725058442,
      "grad_norm": 0.2558238217311259,
      "learning_rate": 2.0872374934932764e-07,
      "loss": 1.5854,
      "step": 3946
    },
    {
      "epoch": 0.5952793906945177,
      "grad_norm": 0.28536413947594585,
      "learning_rate": 2.086218718899837e-07,
      "loss": 1.5424,
      "step": 3947
    },
    {
      "epoch": 0.5954302088831913,
      "grad_norm": 0.28409176854031953,
      "learning_rate": 2.0852000933446268e-07,
      "loss": 1.6091,
      "step": 3948
    },
    {
      "epoch": 0.5955810270718649,
      "grad_norm": 0.2638455821143265,
      "learning_rate": 2.0841816170563558e-07,
      "loss": 1.4921,
      "step": 3949
    },
    {
      "epoch": 0.5957318452605385,
      "grad_norm": 0.2544125404436739,
      "learning_rate": 2.0831632902637036e-07,
      "loss": 1.5437,
      "step": 3950
    },
    {
      "epoch": 0.5958826634492119,
      "grad_norm": 0.3329482205826952,
      "learning_rate": 2.0821451131953128e-07,
      "loss": 1.5533,
      "step": 3951
    },
    {
      "epoch": 0.5960334816378855,
      "grad_norm": 0.24549932350258638,
      "learning_rate": 2.0811270860797942e-07,
      "loss": 1.5775,
      "step": 3952
    },
    {
      "epoch": 0.5961842998265591,
      "grad_norm": 0.24280489344835257,
      "learning_rate": 2.0801092091457252e-07,
      "loss": 1.532,
      "step": 3953
    },
    {
      "epoch": 0.5963351180152326,
      "grad_norm": 0.2662286734750068,
      "learning_rate": 2.0790914826216466e-07,
      "loss": 1.506,
      "step": 3954
    },
    {
      "epoch": 0.5964859362039062,
      "grad_norm": 0.23685295151761798,
      "learning_rate": 2.0780739067360696e-07,
      "loss": 1.602,
      "step": 3955
    },
    {
      "epoch": 0.5966367543925798,
      "grad_norm": 0.24488927836775568,
      "learning_rate": 2.07705648171747e-07,
      "loss": 1.6221,
      "step": 3956
    },
    {
      "epoch": 0.5967875725812533,
      "grad_norm": 0.30208706647893807,
      "learning_rate": 2.0760392077942885e-07,
      "loss": 1.5745,
      "step": 3957
    },
    {
      "epoch": 0.5969383907699268,
      "grad_norm": 0.27614397463665463,
      "learning_rate": 2.0750220851949324e-07,
      "loss": 1.5399,
      "step": 3958
    },
    {
      "epoch": 0.5970892089586004,
      "grad_norm": 0.24605552384885424,
      "learning_rate": 2.074005114147775e-07,
      "loss": 1.4549,
      "step": 3959
    },
    {
      "epoch": 0.597240027147274,
      "grad_norm": 0.2891713160108519,
      "learning_rate": 2.0729882948811573e-07,
      "loss": 1.5384,
      "step": 3960
    },
    {
      "epoch": 0.5973908453359476,
      "grad_norm": 0.2635344708459758,
      "learning_rate": 2.0719716276233844e-07,
      "loss": 1.5661,
      "step": 3961
    },
    {
      "epoch": 0.597541663524621,
      "grad_norm": 0.24959908716759605,
      "learning_rate": 2.070955112602727e-07,
      "loss": 1.5404,
      "step": 3962
    },
    {
      "epoch": 0.5976924817132946,
      "grad_norm": 0.2530298085425925,
      "learning_rate": 2.069938750047423e-07,
      "loss": 1.5452,
      "step": 3963
    },
    {
      "epoch": 0.5978432999019682,
      "grad_norm": 0.26999816586828845,
      "learning_rate": 2.0689225401856754e-07,
      "loss": 1.4886,
      "step": 3964
    },
    {
      "epoch": 0.5979941180906417,
      "grad_norm": 0.24534664683760726,
      "learning_rate": 2.0679064832456522e-07,
      "loss": 1.5164,
      "step": 3965
    },
    {
      "epoch": 0.5981449362793153,
      "grad_norm": 0.25437673218079726,
      "learning_rate": 2.0668905794554886e-07,
      "loss": 1.5683,
      "step": 3966
    },
    {
      "epoch": 0.5982957544679889,
      "grad_norm": 0.24109383361353498,
      "learning_rate": 2.0658748290432842e-07,
      "loss": 1.5497,
      "step": 3967
    },
    {
      "epoch": 0.5984465726566623,
      "grad_norm": 0.2510432340228446,
      "learning_rate": 2.0648592322371045e-07,
      "loss": 1.5323,
      "step": 3968
    },
    {
      "epoch": 0.5985973908453359,
      "grad_norm": 0.2606103881986699,
      "learning_rate": 2.0638437892649814e-07,
      "loss": 1.6385,
      "step": 3969
    },
    {
      "epoch": 0.5987482090340095,
      "grad_norm": 0.2555271402806662,
      "learning_rate": 2.0628285003549095e-07,
      "loss": 1.6314,
      "step": 3970
    },
    {
      "epoch": 0.5988990272226831,
      "grad_norm": 0.24917813790419496,
      "learning_rate": 2.061813365734853e-07,
      "loss": 1.5815,
      "step": 3971
    },
    {
      "epoch": 0.5990498454113566,
      "grad_norm": 0.2490699589494564,
      "learning_rate": 2.0607983856327376e-07,
      "loss": 1.5371,
      "step": 3972
    },
    {
      "epoch": 0.5992006636000302,
      "grad_norm": 0.25179033656276567,
      "learning_rate": 2.0597835602764558e-07,
      "loss": 1.632,
      "step": 3973
    },
    {
      "epoch": 0.5993514817887037,
      "grad_norm": 0.3379907290009692,
      "learning_rate": 2.058768889893867e-07,
      "loss": 1.6252,
      "step": 3974
    },
    {
      "epoch": 0.5995022999773773,
      "grad_norm": 0.2509877314699308,
      "learning_rate": 2.0577543747127928e-07,
      "loss": 1.6167,
      "step": 3975
    },
    {
      "epoch": 0.5996531181660508,
      "grad_norm": 0.325693924804953,
      "learning_rate": 2.056740014961022e-07,
      "loss": 1.5144,
      "step": 3976
    },
    {
      "epoch": 0.5998039363547244,
      "grad_norm": 0.2966623552278896,
      "learning_rate": 2.0557258108663077e-07,
      "loss": 1.603,
      "step": 3977
    },
    {
      "epoch": 0.599954754543398,
      "grad_norm": 0.2412681146127633,
      "learning_rate": 2.0547117626563687e-07,
      "loss": 1.5428,
      "step": 3978
    },
    {
      "epoch": 0.6001055727320714,
      "grad_norm": 0.2453501162309635,
      "learning_rate": 2.0536978705588876e-07,
      "loss": 1.6435,
      "step": 3979
    },
    {
      "epoch": 0.600256390920745,
      "grad_norm": 0.2757165902222816,
      "learning_rate": 2.0526841348015134e-07,
      "loss": 1.5337,
      "step": 3980
    },
    {
      "epoch": 0.6004072091094186,
      "grad_norm": 0.6674557603985731,
      "learning_rate": 2.0516705556118586e-07,
      "loss": 1.4952,
      "step": 3981
    },
    {
      "epoch": 0.6005580272980922,
      "grad_norm": 0.2597450747834464,
      "learning_rate": 2.0506571332175027e-07,
      "loss": 1.5062,
      "step": 3982
    },
    {
      "epoch": 0.6007088454867657,
      "grad_norm": 0.26841242776289004,
      "learning_rate": 2.0496438678459866e-07,
      "loss": 1.598,
      "step": 3983
    },
    {
      "epoch": 0.6008596636754393,
      "grad_norm": 0.2836316492514382,
      "learning_rate": 2.0486307597248192e-07,
      "loss": 1.5803,
      "step": 3984
    },
    {
      "epoch": 0.6010104818641128,
      "grad_norm": 0.24917175257954272,
      "learning_rate": 2.0476178090814716e-07,
      "loss": 1.4978,
      "step": 3985
    },
    {
      "epoch": 0.6011613000527863,
      "grad_norm": 0.3142681695486749,
      "learning_rate": 2.0466050161433822e-07,
      "loss": 1.594,
      "step": 3986
    },
    {
      "epoch": 0.6013121182414599,
      "grad_norm": 0.24425064910154232,
      "learning_rate": 2.0455923811379516e-07,
      "loss": 1.5431,
      "step": 3987
    },
    {
      "epoch": 0.6014629364301335,
      "grad_norm": 0.26105924672835484,
      "learning_rate": 2.044579904292546e-07,
      "loss": 1.5695,
      "step": 3988
    },
    {
      "epoch": 0.6016137546188071,
      "grad_norm": 0.3693476273425113,
      "learning_rate": 2.0435675858344964e-07,
      "loss": 1.5336,
      "step": 3989
    },
    {
      "epoch": 0.6017645728074806,
      "grad_norm": 0.2862201779770262,
      "learning_rate": 2.0425554259910972e-07,
      "loss": 1.5806,
      "step": 3990
    },
    {
      "epoch": 0.6019153909961541,
      "grad_norm": 0.26953985052228474,
      "learning_rate": 2.0415434249896073e-07,
      "loss": 1.5872,
      "step": 3991
    },
    {
      "epoch": 0.6020662091848277,
      "grad_norm": 0.2670217912865157,
      "learning_rate": 2.0405315830572524e-07,
      "loss": 1.5602,
      "step": 3992
    },
    {
      "epoch": 0.6022170273735012,
      "grad_norm": 0.2799420163304301,
      "learning_rate": 2.0395199004212177e-07,
      "loss": 1.5834,
      "step": 3993
    },
    {
      "epoch": 0.6023678455621748,
      "grad_norm": 0.2622691870431353,
      "learning_rate": 2.038508377308657e-07,
      "loss": 1.5618,
      "step": 3994
    },
    {
      "epoch": 0.6025186637508484,
      "grad_norm": 0.24447491424701062,
      "learning_rate": 2.0374970139466874e-07,
      "loss": 1.5483,
      "step": 3995
    },
    {
      "epoch": 0.6026694819395219,
      "grad_norm": 0.25278398998840484,
      "learning_rate": 2.0364858105623877e-07,
      "loss": 1.5617,
      "step": 3996
    },
    {
      "epoch": 0.6028203001281954,
      "grad_norm": 0.3259578925650183,
      "learning_rate": 2.0354747673828037e-07,
      "loss": 1.647,
      "step": 3997
    },
    {
      "epoch": 0.602971118316869,
      "grad_norm": 0.2537031308112455,
      "learning_rate": 2.0344638846349438e-07,
      "loss": 1.5762,
      "step": 3998
    },
    {
      "epoch": 0.6031219365055426,
      "grad_norm": 0.3072984206056757,
      "learning_rate": 2.03345316254578e-07,
      "loss": 1.5335,
      "step": 3999
    },
    {
      "epoch": 0.6032727546942162,
      "grad_norm": 0.23685479682711305,
      "learning_rate": 2.0324426013422503e-07,
      "loss": 1.5777,
      "step": 4000
    },
    {
      "epoch": 0.6034235728828897,
      "grad_norm": 0.24977825998686742,
      "learning_rate": 2.0314322012512536e-07,
      "loss": 1.5163,
      "step": 4001
    },
    {
      "epoch": 0.6035743910715632,
      "grad_norm": 0.24791281456150227,
      "learning_rate": 2.0304219624996548e-07,
      "loss": 1.5596,
      "step": 4002
    },
    {
      "epoch": 0.6037252092602368,
      "grad_norm": 0.2446735385952689,
      "learning_rate": 2.0294118853142816e-07,
      "loss": 1.6651,
      "step": 4003
    },
    {
      "epoch": 0.6038760274489103,
      "grad_norm": 0.25592372644780303,
      "learning_rate": 2.0284019699219261e-07,
      "loss": 1.5529,
      "step": 4004
    },
    {
      "epoch": 0.6040268456375839,
      "grad_norm": 0.24918388591171547,
      "learning_rate": 2.027392216549344e-07,
      "loss": 1.5621,
      "step": 4005
    },
    {
      "epoch": 0.6041776638262575,
      "grad_norm": 0.24863410118812368,
      "learning_rate": 2.0263826254232535e-07,
      "loss": 1.6028,
      "step": 4006
    },
    {
      "epoch": 0.604328482014931,
      "grad_norm": 0.2579812114037648,
      "learning_rate": 2.025373196770338e-07,
      "loss": 1.6146,
      "step": 4007
    },
    {
      "epoch": 0.6044793002036045,
      "grad_norm": 0.25664846405886427,
      "learning_rate": 2.0243639308172433e-07,
      "loss": 1.6093,
      "step": 4008
    },
    {
      "epoch": 0.6046301183922781,
      "grad_norm": 0.2468779556543227,
      "learning_rate": 2.0233548277905788e-07,
      "loss": 1.5615,
      "step": 4009
    },
    {
      "epoch": 0.6047809365809517,
      "grad_norm": 0.24596144695452257,
      "learning_rate": 2.0223458879169176e-07,
      "loss": 1.5465,
      "step": 4010
    },
    {
      "epoch": 0.6049317547696252,
      "grad_norm": 1.3899818192881177,
      "learning_rate": 2.021337111422796e-07,
      "loss": 1.5285,
      "step": 4011
    },
    {
      "epoch": 0.6050825729582988,
      "grad_norm": 0.297674427115539,
      "learning_rate": 2.020328498534714e-07,
      "loss": 1.5641,
      "step": 4012
    },
    {
      "epoch": 0.6052333911469723,
      "grad_norm": 0.25306946958463306,
      "learning_rate": 2.019320049479135e-07,
      "loss": 1.5092,
      "step": 4013
    },
    {
      "epoch": 0.6053842093356459,
      "grad_norm": 0.2622726337402047,
      "learning_rate": 2.0183117644824844e-07,
      "loss": 1.5436,
      "step": 4014
    },
    {
      "epoch": 0.6055350275243194,
      "grad_norm": 0.25230067375888293,
      "learning_rate": 2.017303643771152e-07,
      "loss": 1.5619,
      "step": 4015
    },
    {
      "epoch": 0.605685845712993,
      "grad_norm": 0.24811629245500297,
      "learning_rate": 2.0162956875714892e-07,
      "loss": 1.5575,
      "step": 4016
    },
    {
      "epoch": 0.6058366639016666,
      "grad_norm": 0.25084692060505426,
      "learning_rate": 2.015287896109813e-07,
      "loss": 1.5078,
      "step": 4017
    },
    {
      "epoch": 0.6059874820903401,
      "grad_norm": 0.27316312077471316,
      "learning_rate": 2.014280269612401e-07,
      "loss": 1.6608,
      "step": 4018
    },
    {
      "epoch": 0.6061383002790136,
      "grad_norm": 0.25027999660426603,
      "learning_rate": 2.013272808305495e-07,
      "loss": 1.5155,
      "step": 4019
    },
    {
      "epoch": 0.6062891184676872,
      "grad_norm": 0.27436443112357595,
      "learning_rate": 2.0122655124153003e-07,
      "loss": 1.5238,
      "step": 4020
    },
    {
      "epoch": 0.6064399366563608,
      "grad_norm": 0.29271848922103766,
      "learning_rate": 2.0112583821679823e-07,
      "loss": 1.607,
      "step": 4021
    },
    {
      "epoch": 0.6065907548450343,
      "grad_norm": 0.27458871893177195,
      "learning_rate": 2.0102514177896723e-07,
      "loss": 1.5824,
      "step": 4022
    },
    {
      "epoch": 0.6067415730337079,
      "grad_norm": 0.2871429383848267,
      "learning_rate": 2.0092446195064632e-07,
      "loss": 1.595,
      "step": 4023
    },
    {
      "epoch": 0.6068923912223814,
      "grad_norm": 0.24987310951695893,
      "learning_rate": 2.0082379875444095e-07,
      "loss": 1.5381,
      "step": 4024
    },
    {
      "epoch": 0.6070432094110549,
      "grad_norm": 0.288569270895411,
      "learning_rate": 2.0072315221295304e-07,
      "loss": 1.6683,
      "step": 4025
    },
    {
      "epoch": 0.6071940275997285,
      "grad_norm": 0.2602470848370709,
      "learning_rate": 2.006225223487807e-07,
      "loss": 1.5364,
      "step": 4026
    },
    {
      "epoch": 0.6073448457884021,
      "grad_norm": 0.292666506670437,
      "learning_rate": 2.0052190918451812e-07,
      "loss": 1.538,
      "step": 4027
    },
    {
      "epoch": 0.6074956639770757,
      "grad_norm": 0.25021384441705,
      "learning_rate": 2.0042131274275603e-07,
      "loss": 1.5249,
      "step": 4028
    },
    {
      "epoch": 0.6076464821657492,
      "grad_norm": 0.25042684593440523,
      "learning_rate": 2.0032073304608116e-07,
      "loss": 1.5947,
      "step": 4029
    },
    {
      "epoch": 0.6077973003544227,
      "grad_norm": 0.260428866319,
      "learning_rate": 2.0022017011707659e-07,
      "loss": 1.5524,
      "step": 4030
    },
    {
      "epoch": 0.6079481185430963,
      "grad_norm": 0.25240729324746547,
      "learning_rate": 2.0011962397832176e-07,
      "loss": 1.5652,
      "step": 4031
    },
    {
      "epoch": 0.6080989367317698,
      "grad_norm": 0.26500416833054574,
      "learning_rate": 2.0001909465239204e-07,
      "loss": 1.5401,
      "step": 4032
    },
    {
      "epoch": 0.6082497549204434,
      "grad_norm": 0.24371524278783477,
      "learning_rate": 1.9991858216185926e-07,
      "loss": 1.4963,
      "step": 4033
    },
    {
      "epoch": 0.608400573109117,
      "grad_norm": 0.24616987432015056,
      "learning_rate": 1.998180865292914e-07,
      "loss": 1.5697,
      "step": 4034
    },
    {
      "epoch": 0.6085513912977906,
      "grad_norm": 0.24365652566052104,
      "learning_rate": 1.9971760777725267e-07,
      "loss": 1.5275,
      "step": 4035
    },
    {
      "epoch": 0.608702209486464,
      "grad_norm": 0.3418343694455725,
      "learning_rate": 1.9961714592830343e-07,
      "loss": 1.5607,
      "step": 4036
    },
    {
      "epoch": 0.6088530276751376,
      "grad_norm": 0.24263105443052232,
      "learning_rate": 1.995167010050003e-07,
      "loss": 1.6039,
      "step": 4037
    },
    {
      "epoch": 0.6090038458638112,
      "grad_norm": 0.24789525388839767,
      "learning_rate": 1.994162730298961e-07,
      "loss": 1.5246,
      "step": 4038
    },
    {
      "epoch": 0.6091546640524848,
      "grad_norm": 0.2519990536823345,
      "learning_rate": 1.9931586202553996e-07,
      "loss": 1.5572,
      "step": 4039
    },
    {
      "epoch": 0.6093054822411583,
      "grad_norm": 0.24938178893953983,
      "learning_rate": 1.9921546801447686e-07,
      "loss": 1.5402,
      "step": 4040
    },
    {
      "epoch": 0.6094563004298318,
      "grad_norm": 0.2519735953711437,
      "learning_rate": 1.9911509101924822e-07,
      "loss": 1.5569,
      "step": 4041
    },
    {
      "epoch": 0.6096071186185054,
      "grad_norm": 0.25966022656929916,
      "learning_rate": 1.9901473106239175e-07,
      "loss": 1.5073,
      "step": 4042
    },
    {
      "epoch": 0.6097579368071789,
      "grad_norm": 0.25092742548521996,
      "learning_rate": 1.9891438816644114e-07,
      "loss": 1.5618,
      "step": 4043
    },
    {
      "epoch": 0.6099087549958525,
      "grad_norm": 0.24640923677698962,
      "learning_rate": 1.9881406235392616e-07,
      "loss": 1.5218,
      "step": 4044
    },
    {
      "epoch": 0.6100595731845261,
      "grad_norm": 0.24504171106026,
      "learning_rate": 1.9871375364737303e-07,
      "loss": 1.6179,
      "step": 4045
    },
    {
      "epoch": 0.6102103913731997,
      "grad_norm": 0.2515031678751933,
      "learning_rate": 1.986134620693039e-07,
      "loss": 1.5529,
      "step": 4046
    },
    {
      "epoch": 0.6103612095618731,
      "grad_norm": 0.24473526970393245,
      "learning_rate": 1.9851318764223707e-07,
      "loss": 1.4985,
      "step": 4047
    },
    {
      "epoch": 0.6105120277505467,
      "grad_norm": 0.278452783492205,
      "learning_rate": 1.9841293038868728e-07,
      "loss": 1.6168,
      "step": 4048
    },
    {
      "epoch": 0.6106628459392203,
      "grad_norm": 0.27096654217276134,
      "learning_rate": 1.98312690331165e-07,
      "loss": 1.5433,
      "step": 4049
    },
    {
      "epoch": 0.6108136641278938,
      "grad_norm": 0.2585151663611356,
      "learning_rate": 1.9821246749217718e-07,
      "loss": 1.5119,
      "step": 4050
    },
    {
      "epoch": 0.6109644823165674,
      "grad_norm": 0.24595928472114498,
      "learning_rate": 1.9811226189422665e-07,
      "loss": 1.5151,
      "step": 4051
    },
    {
      "epoch": 0.6111153005052409,
      "grad_norm": 0.25664183667744295,
      "learning_rate": 1.9801207355981253e-07,
      "loss": 1.4884,
      "step": 4052
    },
    {
      "epoch": 0.6112661186939145,
      "grad_norm": 0.2819628569318754,
      "learning_rate": 1.979119025114301e-07,
      "loss": 1.5883,
      "step": 4053
    },
    {
      "epoch": 0.611416936882588,
      "grad_norm": 0.26122868180042536,
      "learning_rate": 1.9781174877157058e-07,
      "loss": 1.5647,
      "step": 4054
    },
    {
      "epoch": 0.6115677550712616,
      "grad_norm": 0.25351052367174465,
      "learning_rate": 1.9771161236272138e-07,
      "loss": 1.5438,
      "step": 4055
    },
    {
      "epoch": 0.6117185732599352,
      "grad_norm": 0.25176624911153417,
      "learning_rate": 1.9761149330736615e-07,
      "loss": 1.5529,
      "step": 4056
    },
    {
      "epoch": 0.6118693914486087,
      "grad_norm": 0.2658656205789021,
      "learning_rate": 1.9751139162798442e-07,
      "loss": 1.5326,
      "step": 4057
    },
    {
      "epoch": 0.6120202096372822,
      "grad_norm": 0.2441990909999017,
      "learning_rate": 1.9741130734705208e-07,
      "loss": 1.5739,
      "step": 4058
    },
    {
      "epoch": 0.6121710278259558,
      "grad_norm": 0.2585026208719849,
      "learning_rate": 1.9731124048704083e-07,
      "loss": 1.5382,
      "step": 4059
    },
    {
      "epoch": 0.6123218460146294,
      "grad_norm": 0.27435498431153166,
      "learning_rate": 1.9721119107041862e-07,
      "loss": 1.5223,
      "step": 4060
    },
    {
      "epoch": 0.6124726642033029,
      "grad_norm": 0.2544450887008572,
      "learning_rate": 1.9711115911964948e-07,
      "loss": 1.5327,
      "step": 4061
    },
    {
      "epoch": 0.6126234823919765,
      "grad_norm": 0.2613375063886286,
      "learning_rate": 1.970111446571935e-07,
      "loss": 1.5335,
      "step": 4062
    },
    {
      "epoch": 0.6127743005806501,
      "grad_norm": 0.2616005863266627,
      "learning_rate": 1.9691114770550683e-07,
      "loss": 1.5132,
      "step": 4063
    },
    {
      "epoch": 0.6129251187693235,
      "grad_norm": 0.24819046142554235,
      "learning_rate": 1.968111682870417e-07,
      "loss": 1.4915,
      "step": 4064
    },
    {
      "epoch": 0.6130759369579971,
      "grad_norm": 0.2406507972592332,
      "learning_rate": 1.9671120642424638e-07,
      "loss": 1.5525,
      "step": 4065
    },
    {
      "epoch": 0.6132267551466707,
      "grad_norm": 0.3982133511343078,
      "learning_rate": 1.9661126213956526e-07,
      "loss": 1.5404,
      "step": 4066
    },
    {
      "epoch": 0.6133775733353443,
      "grad_norm": 0.2461650672475944,
      "learning_rate": 1.9651133545543874e-07,
      "loss": 1.6025,
      "step": 4067
    },
    {
      "epoch": 0.6135283915240178,
      "grad_norm": 0.23898555321982162,
      "learning_rate": 1.964114263943032e-07,
      "loss": 1.5512,
      "step": 4068
    },
    {
      "epoch": 0.6136792097126913,
      "grad_norm": 0.27853965969912664,
      "learning_rate": 1.9631153497859121e-07,
      "loss": 1.5427,
      "step": 4069
    },
    {
      "epoch": 0.6138300279013649,
      "grad_norm": 0.42909251202956133,
      "learning_rate": 1.962116612307313e-07,
      "loss": 1.6012,
      "step": 4070
    },
    {
      "epoch": 0.6139808460900384,
      "grad_norm": 0.3696437855995362,
      "learning_rate": 1.9611180517314802e-07,
      "loss": 1.5802,
      "step": 4071
    },
    {
      "epoch": 0.614131664278712,
      "grad_norm": 0.2549663908066695,
      "learning_rate": 1.9601196682826193e-07,
      "loss": 1.6248,
      "step": 4072
    },
    {
      "epoch": 0.6142824824673856,
      "grad_norm": 0.3301196031328827,
      "learning_rate": 1.9591214621848962e-07,
      "loss": 1.5649,
      "step": 4073
    },
    {
      "epoch": 0.6144333006560592,
      "grad_norm": 0.2621416162410892,
      "learning_rate": 1.9581234336624392e-07,
      "loss": 1.5402,
      "step": 4074
    },
    {
      "epoch": 0.6145841188447326,
      "grad_norm": 0.25361749740991746,
      "learning_rate": 1.9571255829393318e-07,
      "loss": 1.5929,
      "step": 4075
    },
    {
      "epoch": 0.6147349370334062,
      "grad_norm": 0.2747034766804121,
      "learning_rate": 1.9561279102396233e-07,
      "loss": 1.5866,
      "step": 4076
    },
    {
      "epoch": 0.6148857552220798,
      "grad_norm": 0.25233153353526366,
      "learning_rate": 1.9551304157873182e-07,
      "loss": 1.6599,
      "step": 4077
    },
    {
      "epoch": 0.6150365734107534,
      "grad_norm": 0.25539884183733497,
      "learning_rate": 1.9541330998063843e-07,
      "loss": 1.5836,
      "step": 4078
    },
    {
      "epoch": 0.6151873915994269,
      "grad_norm": 0.2507081858447005,
      "learning_rate": 1.9531359625207477e-07,
      "loss": 1.5236,
      "step": 4079
    },
    {
      "epoch": 0.6153382097881005,
      "grad_norm": 0.2558971095723662,
      "learning_rate": 1.9521390041542945e-07,
      "loss": 1.5469,
      "step": 4080
    },
    {
      "epoch": 0.615489027976774,
      "grad_norm": 0.24583239296880544,
      "learning_rate": 1.9511422249308717e-07,
      "loss": 1.5651,
      "step": 4081
    },
    {
      "epoch": 0.6156398461654475,
      "grad_norm": 0.2707883188380288,
      "learning_rate": 1.9501456250742843e-07,
      "loss": 1.5377,
      "step": 4082
    },
    {
      "epoch": 0.6157906643541211,
      "grad_norm": 0.3134379839846514,
      "learning_rate": 1.9491492048082982e-07,
      "loss": 1.5443,
      "step": 4083
    },
    {
      "epoch": 0.6159414825427947,
      "grad_norm": 0.2681812828397381,
      "learning_rate": 1.9481529643566396e-07,
      "loss": 1.6049,
      "step": 4084
    },
    {
      "epoch": 0.6160923007314683,
      "grad_norm": 0.24977055392234673,
      "learning_rate": 1.9471569039429926e-07,
      "loss": 1.5978,
      "step": 4085
    },
    {
      "epoch": 0.6162431189201417,
      "grad_norm": 0.43051972962166635,
      "learning_rate": 1.9461610237910016e-07,
      "loss": 1.5634,
      "step": 4086
    },
    {
      "epoch": 0.6163939371088153,
      "grad_norm": 0.2428402746597223,
      "learning_rate": 1.9451653241242716e-07,
      "loss": 1.5482,
      "step": 4087
    },
    {
      "epoch": 0.6165447552974889,
      "grad_norm": 0.3218770038957548,
      "learning_rate": 1.9441698051663657e-07,
      "loss": 1.5731,
      "step": 4088
    },
    {
      "epoch": 0.6166955734861624,
      "grad_norm": 0.2800037101511921,
      "learning_rate": 1.9431744671408068e-07,
      "loss": 1.639,
      "step": 4089
    },
    {
      "epoch": 0.616846391674836,
      "grad_norm": 0.2654757134978187,
      "learning_rate": 1.9421793102710777e-07,
      "loss": 1.5251,
      "step": 4090
    },
    {
      "epoch": 0.6169972098635096,
      "grad_norm": 0.24389321581221143,
      "learning_rate": 1.941184334780619e-07,
      "loss": 1.5301,
      "step": 4091
    },
    {
      "epoch": 0.617148028052183,
      "grad_norm": 0.24763840870790535,
      "learning_rate": 1.940189540892833e-07,
      "loss": 1.5054,
      "step": 4092
    },
    {
      "epoch": 0.6172988462408566,
      "grad_norm": 0.2496724949164925,
      "learning_rate": 1.9391949288310794e-07,
      "loss": 1.5838,
      "step": 4093
    },
    {
      "epoch": 0.6174496644295302,
      "grad_norm": 0.24786919043003802,
      "learning_rate": 1.9382004988186778e-07,
      "loss": 1.527,
      "step": 4094
    },
    {
      "epoch": 0.6176004826182038,
      "grad_norm": 0.28112801249262676,
      "learning_rate": 1.9372062510789062e-07,
      "loss": 1.5374,
      "step": 4095
    },
    {
      "epoch": 0.6177513008068773,
      "grad_norm": 0.2515386545134396,
      "learning_rate": 1.9362121858350017e-07,
      "loss": 1.6068,
      "step": 4096
    },
    {
      "epoch": 0.6179021189955508,
      "grad_norm": 0.24586907780483525,
      "learning_rate": 1.9352183033101626e-07,
      "loss": 1.5445,
      "step": 4097
    },
    {
      "epoch": 0.6180529371842244,
      "grad_norm": 0.25770929010992843,
      "learning_rate": 1.934224603727543e-07,
      "loss": 1.607,
      "step": 4098
    },
    {
      "epoch": 0.618203755372898,
      "grad_norm": 0.28897863823031006,
      "learning_rate": 1.9332310873102576e-07,
      "loss": 1.544,
      "step": 4099
    },
    {
      "epoch": 0.6183545735615715,
      "grad_norm": 0.24925170298732696,
      "learning_rate": 1.9322377542813806e-07,
      "loss": 1.5623,
      "step": 4100
    },
    {
      "epoch": 0.6185053917502451,
      "grad_norm": 0.2685205241567941,
      "learning_rate": 1.931244604863943e-07,
      "loss": 1.5672,
      "step": 4101
    },
    {
      "epoch": 0.6186562099389187,
      "grad_norm": 0.2731422561837393,
      "learning_rate": 1.9302516392809377e-07,
      "loss": 1.525,
      "step": 4102
    },
    {
      "epoch": 0.6188070281275921,
      "grad_norm": 0.24643168544194732,
      "learning_rate": 1.9292588577553116e-07,
      "loss": 1.5496,
      "step": 4103
    },
    {
      "epoch": 0.6189578463162657,
      "grad_norm": 0.5197430988428763,
      "learning_rate": 1.9282662605099747e-07,
      "loss": 1.5146,
      "step": 4104
    },
    {
      "epoch": 0.6191086645049393,
      "grad_norm": 0.3065588217115942,
      "learning_rate": 1.9272738477677947e-07,
      "loss": 1.5909,
      "step": 4105
    },
    {
      "epoch": 0.6192594826936129,
      "grad_norm": 0.25905693240224775,
      "learning_rate": 1.9262816197515957e-07,
      "loss": 1.5024,
      "step": 4106
    },
    {
      "epoch": 0.6194103008822864,
      "grad_norm": 0.2770648457699646,
      "learning_rate": 1.9252895766841632e-07,
      "loss": 1.5263,
      "step": 4107
    },
    {
      "epoch": 0.61956111907096,
      "grad_norm": 0.26308592325563007,
      "learning_rate": 1.9242977187882381e-07,
      "loss": 1.5136,
      "step": 4108
    },
    {
      "epoch": 0.6197119372596335,
      "grad_norm": 0.24897013505706644,
      "learning_rate": 1.9233060462865226e-07,
      "loss": 1.5643,
      "step": 4109
    },
    {
      "epoch": 0.619862755448307,
      "grad_norm": 0.2551089624531133,
      "learning_rate": 1.9223145594016753e-07,
      "loss": 1.5402,
      "step": 4110
    },
    {
      "epoch": 0.6200135736369806,
      "grad_norm": 0.254746090937665,
      "learning_rate": 1.9213232583563144e-07,
      "loss": 1.553,
      "step": 4111
    },
    {
      "epoch": 0.6201643918256542,
      "grad_norm": 0.25297452968194123,
      "learning_rate": 1.920332143373016e-07,
      "loss": 1.6361,
      "step": 4112
    },
    {
      "epoch": 0.6203152100143278,
      "grad_norm": 0.2642876174112166,
      "learning_rate": 1.9193412146743137e-07,
      "loss": 1.5536,
      "step": 4113
    },
    {
      "epoch": 0.6204660282030012,
      "grad_norm": 0.25329441823211885,
      "learning_rate": 1.9183504724827005e-07,
      "loss": 1.6001,
      "step": 4114
    },
    {
      "epoch": 0.6206168463916748,
      "grad_norm": 0.2529701432326738,
      "learning_rate": 1.9173599170206266e-07,
      "loss": 1.6273,
      "step": 4115
    },
    {
      "epoch": 0.6207676645803484,
      "grad_norm": 0.27508609656579774,
      "learning_rate": 1.9163695485105003e-07,
      "loss": 1.529,
      "step": 4116
    },
    {
      "epoch": 0.620918482769022,
      "grad_norm": 0.2468866471606738,
      "learning_rate": 1.915379367174688e-07,
      "loss": 1.583,
      "step": 4117
    },
    {
      "epoch": 0.6210693009576955,
      "grad_norm": 0.24609628520209548,
      "learning_rate": 1.9143893732355153e-07,
      "loss": 1.6009,
      "step": 4118
    },
    {
      "epoch": 0.6212201191463691,
      "grad_norm": 0.2514840187389571,
      "learning_rate": 1.9133995669152632e-07,
      "loss": 1.5842,
      "step": 4119
    },
    {
      "epoch": 0.6213709373350426,
      "grad_norm": 0.243764288526915,
      "learning_rate": 1.9124099484361738e-07,
      "loss": 1.4551,
      "step": 4120
    },
    {
      "epoch": 0.6215217555237161,
      "grad_norm": 0.24310431183917738,
      "learning_rate": 1.9114205180204436e-07,
      "loss": 1.5369,
      "step": 4121
    },
    {
      "epoch": 0.6216725737123897,
      "grad_norm": 0.24773421290054362,
      "learning_rate": 1.910431275890229e-07,
      "loss": 1.5442,
      "step": 4122
    },
    {
      "epoch": 0.6218233919010633,
      "grad_norm": 0.24121504981746866,
      "learning_rate": 1.909442222267645e-07,
      "loss": 1.5791,
      "step": 4123
    },
    {
      "epoch": 0.6219742100897369,
      "grad_norm": 0.9265375389096069,
      "learning_rate": 1.9084533573747607e-07,
      "loss": 1.5504,
      "step": 4124
    },
    {
      "epoch": 0.6221250282784104,
      "grad_norm": 0.24661482815201358,
      "learning_rate": 1.907464681433607e-07,
      "loss": 1.5427,
      "step": 4125
    },
    {
      "epoch": 0.6222758464670839,
      "grad_norm": 0.25449076998397746,
      "learning_rate": 1.906476194666169e-07,
      "loss": 1.5549,
      "step": 4126
    },
    {
      "epoch": 0.6224266646557575,
      "grad_norm": 0.24234561801686147,
      "learning_rate": 1.9054878972943917e-07,
      "loss": 1.6136,
      "step": 4127
    },
    {
      "epoch": 0.622577482844431,
      "grad_norm": 0.26050689682241945,
      "learning_rate": 1.9044997895401764e-07,
      "loss": 1.6193,
      "step": 4128
    },
    {
      "epoch": 0.6227283010331046,
      "grad_norm": 0.2978038318937061,
      "learning_rate": 1.9035118716253824e-07,
      "loss": 1.5952,
      "step": 4129
    },
    {
      "epoch": 0.6228791192217782,
      "grad_norm": 0.31268001378147847,
      "learning_rate": 1.9025241437718247e-07,
      "loss": 1.6071,
      "step": 4130
    },
    {
      "epoch": 0.6230299374104517,
      "grad_norm": 0.2466418509727293,
      "learning_rate": 1.901536606201279e-07,
      "loss": 1.5854,
      "step": 4131
    },
    {
      "epoch": 0.6231807555991252,
      "grad_norm": 0.2617793842289526,
      "learning_rate": 1.900549259135475e-07,
      "loss": 1.5385,
      "step": 4132
    },
    {
      "epoch": 0.6233315737877988,
      "grad_norm": 0.24814544058042998,
      "learning_rate": 1.8995621027961005e-07,
      "loss": 1.5733,
      "step": 4133
    },
    {
      "epoch": 0.6234823919764724,
      "grad_norm": 0.2599446957413963,
      "learning_rate": 1.898575137404802e-07,
      "loss": 1.52,
      "step": 4134
    },
    {
      "epoch": 0.623633210165146,
      "grad_norm": 0.2858657339108542,
      "learning_rate": 1.8975883631831812e-07,
      "loss": 1.6453,
      "step": 4135
    },
    {
      "epoch": 0.6237840283538195,
      "grad_norm": 0.33921126820227165,
      "learning_rate": 1.8966017803527977e-07,
      "loss": 1.5529,
      "step": 4136
    },
    {
      "epoch": 0.623934846542493,
      "grad_norm": 0.5693270764008028,
      "learning_rate": 1.8956153891351683e-07,
      "loss": 1.5517,
      "step": 4137
    },
    {
      "epoch": 0.6240856647311666,
      "grad_norm": 0.24994044412573277,
      "learning_rate": 1.894629189751767e-07,
      "loss": 1.5789,
      "step": 4138
    },
    {
      "epoch": 0.6242364829198401,
      "grad_norm": 0.24992411867556763,
      "learning_rate": 1.893643182424023e-07,
      "loss": 1.5707,
      "step": 4139
    },
    {
      "epoch": 0.6243873011085137,
      "grad_norm": 0.36508795915231895,
      "learning_rate": 1.8926573673733243e-07,
      "loss": 1.5648,
      "step": 4140
    },
    {
      "epoch": 0.6245381192971873,
      "grad_norm": 0.27187533818843984,
      "learning_rate": 1.8916717448210157e-07,
      "loss": 1.5407,
      "step": 4141
    },
    {
      "epoch": 0.6246889374858607,
      "grad_norm": 0.24719777663478043,
      "learning_rate": 1.8906863149883972e-07,
      "loss": 1.5622,
      "step": 4142
    },
    {
      "epoch": 0.6248397556745343,
      "grad_norm": 0.3215843836818221,
      "learning_rate": 1.8897010780967275e-07,
      "loss": 1.5543,
      "step": 4143
    },
    {
      "epoch": 0.6249905738632079,
      "grad_norm": 0.25889412312839055,
      "learning_rate": 1.8887160343672205e-07,
      "loss": 1.5937,
      "step": 4144
    },
    {
      "epoch": 0.6251413920518815,
      "grad_norm": 0.2486312762659672,
      "learning_rate": 1.8877311840210464e-07,
      "loss": 1.5016,
      "step": 4145
    },
    {
      "epoch": 0.625292210240555,
      "grad_norm": 0.25252796679476264,
      "learning_rate": 1.8867465272793341e-07,
      "loss": 1.5647,
      "step": 4146
    },
    {
      "epoch": 0.6254430284292286,
      "grad_norm": 0.26325625302217504,
      "learning_rate": 1.8857620643631672e-07,
      "loss": 1.5759,
      "step": 4147
    },
    {
      "epoch": 0.6255938466179021,
      "grad_norm": 0.2794617366637707,
      "learning_rate": 1.8847777954935856e-07,
      "loss": 1.6256,
      "step": 4148
    },
    {
      "epoch": 0.6257446648065756,
      "grad_norm": 0.25124112050953085,
      "learning_rate": 1.8837937208915876e-07,
      "loss": 1.6167,
      "step": 4149
    },
    {
      "epoch": 0.6258954829952492,
      "grad_norm": 0.2403652724633617,
      "learning_rate": 1.8828098407781256e-07,
      "loss": 1.5463,
      "step": 4150
    },
    {
      "epoch": 0.6260463011839228,
      "grad_norm": 0.24589197110547903,
      "learning_rate": 1.8818261553741106e-07,
      "loss": 1.522,
      "step": 4151
    },
    {
      "epoch": 0.6261971193725964,
      "grad_norm": 0.2414084395099176,
      "learning_rate": 1.8808426649004066e-07,
      "loss": 1.576,
      "step": 4152
    },
    {
      "epoch": 0.6263479375612699,
      "grad_norm": 0.2621010820780024,
      "learning_rate": 1.8798593695778376e-07,
      "loss": 1.4603,
      "step": 4153
    },
    {
      "epoch": 0.6264987557499434,
      "grad_norm": 0.2530761046314033,
      "learning_rate": 1.8788762696271815e-07,
      "loss": 1.5216,
      "step": 4154
    },
    {
      "epoch": 0.626649573938617,
      "grad_norm": 0.2630464111986687,
      "learning_rate": 1.8778933652691724e-07,
      "loss": 1.537,
      "step": 4155
    },
    {
      "epoch": 0.6268003921272906,
      "grad_norm": 0.24619072070004963,
      "learning_rate": 1.8769106567245012e-07,
      "loss": 1.5224,
      "step": 4156
    },
    {
      "epoch": 0.6269512103159641,
      "grad_norm": 0.2701082424355122,
      "learning_rate": 1.8759281442138152e-07,
      "loss": 1.6291,
      "step": 4157
    },
    {
      "epoch": 0.6271020285046377,
      "grad_norm": 0.23715739982926645,
      "learning_rate": 1.8749458279577162e-07,
      "loss": 1.5534,
      "step": 4158
    },
    {
      "epoch": 0.6272528466933112,
      "grad_norm": 0.2503082038843583,
      "learning_rate": 1.8739637081767638e-07,
      "loss": 1.5226,
      "step": 4159
    },
    {
      "epoch": 0.6274036648819847,
      "grad_norm": 0.25121712160535314,
      "learning_rate": 1.8729817850914713e-07,
      "loss": 1.5512,
      "step": 4160
    },
    {
      "epoch": 0.6275544830706583,
      "grad_norm": 0.2490038140958148,
      "learning_rate": 1.87200005892231e-07,
      "loss": 1.5235,
      "step": 4161
    },
    {
      "epoch": 0.6277053012593319,
      "grad_norm": 0.26388132410554616,
      "learning_rate": 1.8710185298897057e-07,
      "loss": 1.5502,
      "step": 4162
    },
    {
      "epoch": 0.6278561194480055,
      "grad_norm": 0.24615806276388288,
      "learning_rate": 1.8700371982140407e-07,
      "loss": 1.5226,
      "step": 4163
    },
    {
      "epoch": 0.628006937636679,
      "grad_norm": 0.6160809143870036,
      "learning_rate": 1.869056064115651e-07,
      "loss": 1.5449,
      "step": 4164
    },
    {
      "epoch": 0.6281577558253525,
      "grad_norm": 0.3681537783717758,
      "learning_rate": 1.8680751278148315e-07,
      "loss": 1.5616,
      "step": 4165
    },
    {
      "epoch": 0.6283085740140261,
      "grad_norm": 0.2539030267347039,
      "learning_rate": 1.8670943895318297e-07,
      "loss": 1.5492,
      "step": 4166
    },
    {
      "epoch": 0.6284593922026996,
      "grad_norm": 0.24935666844681972,
      "learning_rate": 1.8661138494868513e-07,
      "loss": 1.5975,
      "step": 4167
    },
    {
      "epoch": 0.6286102103913732,
      "grad_norm": 0.2388405326692446,
      "learning_rate": 1.8651335079000547e-07,
      "loss": 1.5921,
      "step": 4168
    },
    {
      "epoch": 0.6287610285800468,
      "grad_norm": 0.24942525861218032,
      "learning_rate": 1.8641533649915558e-07,
      "loss": 1.5575,
      "step": 4169
    },
    {
      "epoch": 0.6289118467687204,
      "grad_norm": 0.281992690958799,
      "learning_rate": 1.8631734209814254e-07,
      "loss": 1.5531,
      "step": 4170
    },
    {
      "epoch": 0.6290626649573938,
      "grad_norm": 0.4165446083848909,
      "learning_rate": 1.862193676089689e-07,
      "loss": 1.523,
      "step": 4171
    },
    {
      "epoch": 0.6292134831460674,
      "grad_norm": 0.270594878229444,
      "learning_rate": 1.8612141305363282e-07,
      "loss": 1.4721,
      "step": 4172
    },
    {
      "epoch": 0.629364301334741,
      "grad_norm": 0.2636488038229057,
      "learning_rate": 1.8602347845412796e-07,
      "loss": 1.5209,
      "step": 4173
    },
    {
      "epoch": 0.6295151195234145,
      "grad_norm": 0.26570010948929257,
      "learning_rate": 1.859255638324434e-07,
      "loss": 1.5453,
      "step": 4174
    },
    {
      "epoch": 0.6296659377120881,
      "grad_norm": 0.30017922416708503,
      "learning_rate": 1.8582766921056403e-07,
      "loss": 1.5877,
      "step": 4175
    },
    {
      "epoch": 0.6298167559007616,
      "grad_norm": 0.24676668094362056,
      "learning_rate": 1.8572979461046978e-07,
      "loss": 1.5554,
      "step": 4176
    },
    {
      "epoch": 0.6299675740894352,
      "grad_norm": 0.23806938270752506,
      "learning_rate": 1.8563194005413652e-07,
      "loss": 1.4845,
      "step": 4177
    },
    {
      "epoch": 0.6301183922781087,
      "grad_norm": 0.24248553109147253,
      "learning_rate": 1.8553410556353537e-07,
      "loss": 1.5381,
      "step": 4178
    },
    {
      "epoch": 0.6302692104667823,
      "grad_norm": 0.2440376958567314,
      "learning_rate": 1.854362911606331e-07,
      "loss": 1.5795,
      "step": 4179
    },
    {
      "epoch": 0.6304200286554559,
      "grad_norm": 0.24707297107098733,
      "learning_rate": 1.853384968673919e-07,
      "loss": 1.5483,
      "step": 4180
    },
    {
      "epoch": 0.6305708468441295,
      "grad_norm": 0.2701950141920131,
      "learning_rate": 1.852407227057693e-07,
      "loss": 1.5881,
      "step": 4181
    },
    {
      "epoch": 0.6307216650328029,
      "grad_norm": 0.25399214732994313,
      "learning_rate": 1.8514296869771856e-07,
      "loss": 1.6224,
      "step": 4182
    },
    {
      "epoch": 0.6308724832214765,
      "grad_norm": 0.25343116507938845,
      "learning_rate": 1.8504523486518825e-07,
      "loss": 1.5684,
      "step": 4183
    },
    {
      "epoch": 0.6310233014101501,
      "grad_norm": 0.25533834657361526,
      "learning_rate": 1.8494752123012247e-07,
      "loss": 1.5733,
      "step": 4184
    },
    {
      "epoch": 0.6311741195988236,
      "grad_norm": 0.41570541171889785,
      "learning_rate": 1.8484982781446085e-07,
      "loss": 1.5857,
      "step": 4185
    },
    {
      "epoch": 0.6313249377874972,
      "grad_norm": 0.25628694395084733,
      "learning_rate": 1.8475215464013826e-07,
      "loss": 1.5359,
      "step": 4186
    },
    {
      "epoch": 0.6314757559761707,
      "grad_norm": 1.398194438395537,
      "learning_rate": 1.8465450172908535e-07,
      "loss": 1.547,
      "step": 4187
    },
    {
      "epoch": 0.6316265741648442,
      "grad_norm": 0.2669251497151142,
      "learning_rate": 1.8455686910322794e-07,
      "loss": 1.5542,
      "step": 4188
    },
    {
      "epoch": 0.6317773923535178,
      "grad_norm": 0.2680685107882268,
      "learning_rate": 1.8445925678448737e-07,
      "loss": 1.5726,
      "step": 4189
    },
    {
      "epoch": 0.6319282105421914,
      "grad_norm": 0.2598035544709078,
      "learning_rate": 1.8436166479478054e-07,
      "loss": 1.5905,
      "step": 4190
    },
    {
      "epoch": 0.632079028730865,
      "grad_norm": 0.3312860252153446,
      "learning_rate": 1.8426409315601964e-07,
      "loss": 1.5328,
      "step": 4191
    },
    {
      "epoch": 0.6322298469195385,
      "grad_norm": 0.24867557124552955,
      "learning_rate": 1.8416654189011231e-07,
      "loss": 1.5652,
      "step": 4192
    },
    {
      "epoch": 0.632380665108212,
      "grad_norm": 0.2539702365191936,
      "learning_rate": 1.8406901101896172e-07,
      "loss": 1.5334,
      "step": 4193
    },
    {
      "epoch": 0.6325314832968856,
      "grad_norm": 0.25319783760576814,
      "learning_rate": 1.8397150056446637e-07,
      "loss": 1.5103,
      "step": 4194
    },
    {
      "epoch": 0.6326823014855592,
      "grad_norm": 0.2725195248103415,
      "learning_rate": 1.8387401054852015e-07,
      "loss": 1.5734,
      "step": 4195
    },
    {
      "epoch": 0.6328331196742327,
      "grad_norm": 0.2389547759149391,
      "learning_rate": 1.8377654099301242e-07,
      "loss": 1.5676,
      "step": 4196
    },
    {
      "epoch": 0.6329839378629063,
      "grad_norm": 0.24308762508937254,
      "learning_rate": 1.83679091919828e-07,
      "loss": 1.531,
      "step": 4197
    },
    {
      "epoch": 0.6331347560515799,
      "grad_norm": 0.23951624981640487,
      "learning_rate": 1.83581663350847e-07,
      "loss": 1.4717,
      "step": 4198
    },
    {
      "epoch": 0.6332855742402533,
      "grad_norm": 0.24921256044384987,
      "learning_rate": 1.8348425530794497e-07,
      "loss": 1.4919,
      "step": 4199
    },
    {
      "epoch": 0.6334363924289269,
      "grad_norm": 0.24498417312046017,
      "learning_rate": 1.8338686781299283e-07,
      "loss": 1.5877,
      "step": 4200
    },
    {
      "epoch": 0.6335872106176005,
      "grad_norm": 0.25287524883896284,
      "learning_rate": 1.83289500887857e-07,
      "loss": 1.5059,
      "step": 4201
    },
    {
      "epoch": 0.6337380288062741,
      "grad_norm": 0.28046838176968375,
      "learning_rate": 1.8319215455439907e-07,
      "loss": 1.5184,
      "step": 4202
    },
    {
      "epoch": 0.6338888469949476,
      "grad_norm": 0.24926218979066564,
      "learning_rate": 1.8309482883447624e-07,
      "loss": 1.5998,
      "step": 4203
    },
    {
      "epoch": 0.6340396651836211,
      "grad_norm": 0.26904621255126415,
      "learning_rate": 1.8299752374994087e-07,
      "loss": 1.4524,
      "step": 4204
    },
    {
      "epoch": 0.6341904833722947,
      "grad_norm": 0.2516115836972916,
      "learning_rate": 1.829002393226408e-07,
      "loss": 1.5444,
      "step": 4205
    },
    {
      "epoch": 0.6343413015609682,
      "grad_norm": 0.25427584168397666,
      "learning_rate": 1.8280297557441933e-07,
      "loss": 1.5456,
      "step": 4206
    },
    {
      "epoch": 0.6344921197496418,
      "grad_norm": 0.2607497963600299,
      "learning_rate": 1.827057325271148e-07,
      "loss": 1.5775,
      "step": 4207
    },
    {
      "epoch": 0.6346429379383154,
      "grad_norm": 0.2713007671030094,
      "learning_rate": 1.8260851020256135e-07,
      "loss": 1.5586,
      "step": 4208
    },
    {
      "epoch": 0.634793756126989,
      "grad_norm": 0.24027781792831884,
      "learning_rate": 1.82511308622588e-07,
      "loss": 1.5769,
      "step": 4209
    },
    {
      "epoch": 0.6349445743156624,
      "grad_norm": 0.24382415193812368,
      "learning_rate": 1.8241412780901942e-07,
      "loss": 1.5387,
      "step": 4210
    },
    {
      "epoch": 0.635095392504336,
      "grad_norm": 0.2487821871282781,
      "learning_rate": 1.8231696778367561e-07,
      "loss": 1.6078,
      "step": 4211
    },
    {
      "epoch": 0.6352462106930096,
      "grad_norm": 0.26504042617639684,
      "learning_rate": 1.8221982856837177e-07,
      "loss": 1.6453,
      "step": 4212
    },
    {
      "epoch": 0.6353970288816831,
      "grad_norm": 0.2900287076542723,
      "learning_rate": 1.8212271018491836e-07,
      "loss": 1.5267,
      "step": 4213
    },
    {
      "epoch": 0.6355478470703567,
      "grad_norm": 0.2729493662176146,
      "learning_rate": 1.8202561265512155e-07,
      "loss": 1.5371,
      "step": 4214
    },
    {
      "epoch": 0.6356986652590303,
      "grad_norm": 0.5427101588791572,
      "learning_rate": 1.8192853600078234e-07,
      "loss": 1.5208,
      "step": 4215
    },
    {
      "epoch": 0.6358494834477038,
      "grad_norm": 0.2438775766503901,
      "learning_rate": 1.8183148024369736e-07,
      "loss": 1.5903,
      "step": 4216
    },
    {
      "epoch": 0.6360003016363773,
      "grad_norm": 0.25030235583666205,
      "learning_rate": 1.817344454056584e-07,
      "loss": 1.6066,
      "step": 4217
    },
    {
      "epoch": 0.6361511198250509,
      "grad_norm": 0.24859572445405265,
      "learning_rate": 1.8163743150845263e-07,
      "loss": 1.5843,
      "step": 4218
    },
    {
      "epoch": 0.6363019380137245,
      "grad_norm": 0.2847060495976388,
      "learning_rate": 1.815404385738626e-07,
      "loss": 1.5628,
      "step": 4219
    },
    {
      "epoch": 0.636452756202398,
      "grad_norm": 0.2545438298708443,
      "learning_rate": 1.8144346662366582e-07,
      "loss": 1.4878,
      "step": 4220
    },
    {
      "epoch": 0.6366035743910715,
      "grad_norm": 0.3434970471598928,
      "learning_rate": 1.8134651567963555e-07,
      "loss": 1.6083,
      "step": 4221
    },
    {
      "epoch": 0.6367543925797451,
      "grad_norm": 0.24761430835041962,
      "learning_rate": 1.812495857635399e-07,
      "loss": 1.5511,
      "step": 4222
    },
    {
      "epoch": 0.6369052107684187,
      "grad_norm": 0.25743559630787793,
      "learning_rate": 1.8115267689714265e-07,
      "loss": 1.5761,
      "step": 4223
    },
    {
      "epoch": 0.6370560289570922,
      "grad_norm": 0.24586976247005674,
      "learning_rate": 1.810557891022026e-07,
      "loss": 1.5507,
      "step": 4224
    },
    {
      "epoch": 0.6372068471457658,
      "grad_norm": 0.24560702102994042,
      "learning_rate": 1.8095892240047382e-07,
      "loss": 1.5253,
      "step": 4225
    },
    {
      "epoch": 0.6373576653344394,
      "grad_norm": 0.35455214235373617,
      "learning_rate": 1.8086207681370562e-07,
      "loss": 1.5711,
      "step": 4226
    },
    {
      "epoch": 0.6375084835231128,
      "grad_norm": 0.2694202111370069,
      "learning_rate": 1.8076525236364292e-07,
      "loss": 1.4951,
      "step": 4227
    },
    {
      "epoch": 0.6376593017117864,
      "grad_norm": 0.2935064832001404,
      "learning_rate": 1.806684490720255e-07,
      "loss": 1.58,
      "step": 4228
    },
    {
      "epoch": 0.63781011990046,
      "grad_norm": 0.25526045148889653,
      "learning_rate": 1.8057166696058834e-07,
      "loss": 1.4672,
      "step": 4229
    },
    {
      "epoch": 0.6379609380891336,
      "grad_norm": 0.2488160762655106,
      "learning_rate": 1.804749060510622e-07,
      "loss": 1.536,
      "step": 4230
    },
    {
      "epoch": 0.6381117562778071,
      "grad_norm": 0.2489581523780075,
      "learning_rate": 1.803781663651724e-07,
      "loss": 1.6017,
      "step": 4231
    },
    {
      "epoch": 0.6382625744664806,
      "grad_norm": 0.3321037514342219,
      "learning_rate": 1.8028144792463996e-07,
      "loss": 1.6216,
      "step": 4232
    },
    {
      "epoch": 0.6384133926551542,
      "grad_norm": 0.24786754870364802,
      "learning_rate": 1.80184750751181e-07,
      "loss": 1.6027,
      "step": 4233
    },
    {
      "epoch": 0.6385642108438278,
      "grad_norm": 0.25282574867184987,
      "learning_rate": 1.8008807486650679e-07,
      "loss": 1.5709,
      "step": 4234
    },
    {
      "epoch": 0.6387150290325013,
      "grad_norm": 0.25331586615858,
      "learning_rate": 1.7999142029232388e-07,
      "loss": 1.5655,
      "step": 4235
    },
    {
      "epoch": 0.6388658472211749,
      "grad_norm": 0.2552052135322833,
      "learning_rate": 1.7989478705033417e-07,
      "loss": 1.6102,
      "step": 4236
    },
    {
      "epoch": 0.6390166654098485,
      "grad_norm": 0.2521213219961141,
      "learning_rate": 1.7979817516223456e-07,
      "loss": 1.5544,
      "step": 4237
    },
    {
      "epoch": 0.6391674835985219,
      "grad_norm": 0.24411593375663035,
      "learning_rate": 1.7970158464971713e-07,
      "loss": 1.538,
      "step": 4238
    },
    {
      "epoch": 0.6393183017871955,
      "grad_norm": 0.25975553327168827,
      "learning_rate": 1.7960501553446927e-07,
      "loss": 1.5823,
      "step": 4239
    },
    {
      "epoch": 0.6394691199758691,
      "grad_norm": 0.41432267503092424,
      "learning_rate": 1.7950846783817375e-07,
      "loss": 1.5641,
      "step": 4240
    },
    {
      "epoch": 0.6396199381645427,
      "grad_norm": 0.2521228127440944,
      "learning_rate": 1.7941194158250823e-07,
      "loss": 1.586,
      "step": 4241
    },
    {
      "epoch": 0.6397707563532162,
      "grad_norm": 0.2808400435101359,
      "learning_rate": 1.793154367891455e-07,
      "loss": 1.5899,
      "step": 4242
    },
    {
      "epoch": 0.6399215745418898,
      "grad_norm": 0.24672901445972867,
      "learning_rate": 1.79218953479754e-07,
      "loss": 1.5811,
      "step": 4243
    },
    {
      "epoch": 0.6400723927305633,
      "grad_norm": 0.2664023244654792,
      "learning_rate": 1.7912249167599692e-07,
      "loss": 1.596,
      "step": 4244
    },
    {
      "epoch": 0.6402232109192368,
      "grad_norm": 0.24914972989068035,
      "learning_rate": 1.7902605139953258e-07,
      "loss": 1.5428,
      "step": 4245
    },
    {
      "epoch": 0.6403740291079104,
      "grad_norm": 0.30448809907106317,
      "learning_rate": 1.789296326720149e-07,
      "loss": 1.5993,
      "step": 4246
    },
    {
      "epoch": 0.640524847296584,
      "grad_norm": 0.2400644326933849,
      "learning_rate": 1.788332355150925e-07,
      "loss": 1.5132,
      "step": 4247
    },
    {
      "epoch": 0.6406756654852576,
      "grad_norm": 0.29287295103545796,
      "learning_rate": 1.7873685995040939e-07,
      "loss": 1.5137,
      "step": 4248
    },
    {
      "epoch": 0.640826483673931,
      "grad_norm": 0.24555063234062374,
      "learning_rate": 1.7864050599960467e-07,
      "loss": 1.5487,
      "step": 4249
    },
    {
      "epoch": 0.6409773018626046,
      "grad_norm": 0.3538453607491625,
      "learning_rate": 1.785441736843128e-07,
      "loss": 1.5447,
      "step": 4250
    },
    {
      "epoch": 0.6411281200512782,
      "grad_norm": 0.26030436621563,
      "learning_rate": 1.7844786302616299e-07,
      "loss": 1.501,
      "step": 4251
    },
    {
      "epoch": 0.6412789382399517,
      "grad_norm": 0.26606952015233887,
      "learning_rate": 1.7835157404677975e-07,
      "loss": 1.531,
      "step": 4252
    },
    {
      "epoch": 0.6414297564286253,
      "grad_norm": 0.2442741412802352,
      "learning_rate": 1.7825530676778298e-07,
      "loss": 1.538,
      "step": 4253
    },
    {
      "epoch": 0.6415805746172989,
      "grad_norm": 0.2766729345972109,
      "learning_rate": 1.7815906121078734e-07,
      "loss": 1.5777,
      "step": 4254
    },
    {
      "epoch": 0.6417313928059724,
      "grad_norm": 0.2633113523056404,
      "learning_rate": 1.7806283739740275e-07,
      "loss": 1.6583,
      "step": 4255
    },
    {
      "epoch": 0.6418822109946459,
      "grad_norm": 0.26819538082082267,
      "learning_rate": 1.7796663534923417e-07,
      "loss": 1.4589,
      "step": 4256
    },
    {
      "epoch": 0.6420330291833195,
      "grad_norm": 0.24879651699590694,
      "learning_rate": 1.7787045508788208e-07,
      "loss": 1.4937,
      "step": 4257
    },
    {
      "epoch": 0.6421838473719931,
      "grad_norm": 0.3620468432092983,
      "learning_rate": 1.7777429663494149e-07,
      "loss": 1.5409,
      "step": 4258
    },
    {
      "epoch": 0.6423346655606667,
      "grad_norm": 0.24575768302139991,
      "learning_rate": 1.7767816001200265e-07,
      "loss": 1.5894,
      "step": 4259
    },
    {
      "epoch": 0.6424854837493402,
      "grad_norm": 0.2455419739713735,
      "learning_rate": 1.775820452406514e-07,
      "loss": 1.5973,
      "step": 4260
    },
    {
      "epoch": 0.6426363019380137,
      "grad_norm": 0.44783851720614026,
      "learning_rate": 1.7748595234246812e-07,
      "loss": 1.5683,
      "step": 4261
    },
    {
      "epoch": 0.6427871201266873,
      "grad_norm": 0.27759236136718557,
      "learning_rate": 1.7738988133902832e-07,
      "loss": 1.5398,
      "step": 4262
    },
    {
      "epoch": 0.6429379383153608,
      "grad_norm": 0.264656048146391,
      "learning_rate": 1.7729383225190302e-07,
      "loss": 1.5681,
      "step": 4263
    },
    {
      "epoch": 0.6430887565040344,
      "grad_norm": 0.39913174201778384,
      "learning_rate": 1.771978051026579e-07,
      "loss": 1.5774,
      "step": 4264
    },
    {
      "epoch": 0.643239574692708,
      "grad_norm": 0.2412047861650146,
      "learning_rate": 1.7710179991285384e-07,
      "loss": 1.6106,
      "step": 4265
    },
    {
      "epoch": 0.6433903928813814,
      "grad_norm": 0.27653888375962127,
      "learning_rate": 1.7700581670404668e-07,
      "loss": 1.5251,
      "step": 4266
    },
    {
      "epoch": 0.643541211070055,
      "grad_norm": 0.2482221340611367,
      "learning_rate": 1.769098554977877e-07,
      "loss": 1.571,
      "step": 4267
    },
    {
      "epoch": 0.6436920292587286,
      "grad_norm": 0.2536941588689048,
      "learning_rate": 1.7681391631562287e-07,
      "loss": 1.6051,
      "step": 4268
    },
    {
      "epoch": 0.6438428474474022,
      "grad_norm": 0.2612095300387124,
      "learning_rate": 1.7671799917909324e-07,
      "loss": 1.5267,
      "step": 4269
    },
    {
      "epoch": 0.6439936656360757,
      "grad_norm": 0.544029058097784,
      "learning_rate": 1.7662210410973512e-07,
      "loss": 1.6163,
      "step": 4270
    },
    {
      "epoch": 0.6441444838247493,
      "grad_norm": 0.30183016903165544,
      "learning_rate": 1.7652623112907971e-07,
      "loss": 1.5697,
      "step": 4271
    },
    {
      "epoch": 0.6442953020134228,
      "grad_norm": 0.26179017106545627,
      "learning_rate": 1.7643038025865326e-07,
      "loss": 1.6243,
      "step": 4272
    },
    {
      "epoch": 0.6444461202020964,
      "grad_norm": 0.2512366290362913,
      "learning_rate": 1.7633455151997718e-07,
      "loss": 1.5723,
      "step": 4273
    },
    {
      "epoch": 0.6445969383907699,
      "grad_norm": 0.2833266668155499,
      "learning_rate": 1.762387449345677e-07,
      "loss": 1.6155,
      "step": 4274
    },
    {
      "epoch": 0.6447477565794435,
      "grad_norm": 0.25350106304489006,
      "learning_rate": 1.7614296052393618e-07,
      "loss": 1.5511,
      "step": 4275
    },
    {
      "epoch": 0.6448985747681171,
      "grad_norm": 0.25067693443556277,
      "learning_rate": 1.7604719830958913e-07,
      "loss": 1.566,
      "step": 4276
    },
    {
      "epoch": 0.6450493929567905,
      "grad_norm": 0.44205510669995635,
      "learning_rate": 1.759514583130279e-07,
      "loss": 1.5355,
      "step": 4277
    },
    {
      "epoch": 0.6452002111454641,
      "grad_norm": 0.25698267380120476,
      "learning_rate": 1.7585574055574882e-07,
      "loss": 1.5815,
      "step": 4278
    },
    {
      "epoch": 0.6453510293341377,
      "grad_norm": 0.2577964193893113,
      "learning_rate": 1.7576004505924335e-07,
      "loss": 1.5543,
      "step": 4279
    },
    {
      "epoch": 0.6455018475228113,
      "grad_norm": 0.693098392046214,
      "learning_rate": 1.7566437184499806e-07,
      "loss": 1.6158,
      "step": 4280
    },
    {
      "epoch": 0.6456526657114848,
      "grad_norm": 0.2592870202788428,
      "learning_rate": 1.755687209344942e-07,
      "loss": 1.5987,
      "step": 4281
    },
    {
      "epoch": 0.6458034839001584,
      "grad_norm": 0.23694233652541807,
      "learning_rate": 1.7547309234920827e-07,
      "loss": 1.5027,
      "step": 4282
    },
    {
      "epoch": 0.6459543020888319,
      "grad_norm": 0.39543519488811646,
      "learning_rate": 1.7537748611061166e-07,
      "loss": 1.5788,
      "step": 4283
    },
    {
      "epoch": 0.6461051202775054,
      "grad_norm": 0.24855769967495914,
      "learning_rate": 1.752819022401707e-07,
      "loss": 1.4575,
      "step": 4284
    },
    {
      "epoch": 0.646255938466179,
      "grad_norm": 0.243320883100979,
      "learning_rate": 1.751863407593468e-07,
      "loss": 1.5473,
      "step": 4285
    },
    {
      "epoch": 0.6464067566548526,
      "grad_norm": 0.2542812940612368,
      "learning_rate": 1.7509080168959634e-07,
      "loss": 1.5126,
      "step": 4286
    },
    {
      "epoch": 0.6465575748435262,
      "grad_norm": 0.2728851797285914,
      "learning_rate": 1.7499528505237058e-07,
      "loss": 1.5611,
      "step": 4287
    },
    {
      "epoch": 0.6467083930321997,
      "grad_norm": 0.24571581123573824,
      "learning_rate": 1.7489979086911572e-07,
      "loss": 1.5596,
      "step": 4288
    },
    {
      "epoch": 0.6468592112208732,
      "grad_norm": 0.25357135141351167,
      "learning_rate": 1.7480431916127314e-07,
      "loss": 1.5041,
      "step": 4289
    },
    {
      "epoch": 0.6470100294095468,
      "grad_norm": 0.25122414364095647,
      "learning_rate": 1.7470886995027889e-07,
      "loss": 1.5632,
      "step": 4290
    },
    {
      "epoch": 0.6471608475982203,
      "grad_norm": 0.2504427056630961,
      "learning_rate": 1.7461344325756416e-07,
      "loss": 1.5337,
      "step": 4291
    },
    {
      "epoch": 0.6473116657868939,
      "grad_norm": 0.2536601641407676,
      "learning_rate": 1.7451803910455493e-07,
      "loss": 1.4983,
      "step": 4292
    },
    {
      "epoch": 0.6474624839755675,
      "grad_norm": 0.3148369931095937,
      "learning_rate": 1.7442265751267233e-07,
      "loss": 1.504,
      "step": 4293
    },
    {
      "epoch": 0.647613302164241,
      "grad_norm": 0.2528144702491195,
      "learning_rate": 1.7432729850333227e-07,
      "loss": 1.5981,
      "step": 4294
    },
    {
      "epoch": 0.6477641203529145,
      "grad_norm": 0.24553333308016317,
      "learning_rate": 1.7423196209794553e-07,
      "loss": 1.6014,
      "step": 4295
    },
    {
      "epoch": 0.6479149385415881,
      "grad_norm": 0.2522726324420554,
      "learning_rate": 1.7413664831791808e-07,
      "loss": 1.5271,
      "step": 4296
    },
    {
      "epoch": 0.6480657567302617,
      "grad_norm": 0.24628252707932746,
      "learning_rate": 1.7404135718465052e-07,
      "loss": 1.5046,
      "step": 4297
    },
    {
      "epoch": 0.6482165749189353,
      "grad_norm": 0.25665776928582656,
      "learning_rate": 1.7394608871953846e-07,
      "loss": 1.5659,
      "step": 4298
    },
    {
      "epoch": 0.6483673931076088,
      "grad_norm": 0.2546847179612797,
      "learning_rate": 1.7385084294397255e-07,
      "loss": 1.5515,
      "step": 4299
    },
    {
      "epoch": 0.6485182112962823,
      "grad_norm": 0.39990758357897654,
      "learning_rate": 1.7375561987933823e-07,
      "loss": 1.551,
      "step": 4300
    },
    {
      "epoch": 0.6486690294849559,
      "grad_norm": 0.2391955745190497,
      "learning_rate": 1.7366041954701572e-07,
      "loss": 1.5303,
      "step": 4301
    },
    {
      "epoch": 0.6488198476736294,
      "grad_norm": 0.25178143314721985,
      "learning_rate": 1.7356524196838038e-07,
      "loss": 1.5213,
      "step": 4302
    },
    {
      "epoch": 0.648970665862303,
      "grad_norm": 0.3185469911409178,
      "learning_rate": 1.7347008716480239e-07,
      "loss": 1.5558,
      "step": 4303
    },
    {
      "epoch": 0.6491214840509766,
      "grad_norm": 0.26510807044531254,
      "learning_rate": 1.7337495515764667e-07,
      "loss": 1.6366,
      "step": 4304
    },
    {
      "epoch": 0.6492723022396502,
      "grad_norm": 0.24514148823357573,
      "learning_rate": 1.732798459682731e-07,
      "loss": 1.4934,
      "step": 4305
    },
    {
      "epoch": 0.6494231204283236,
      "grad_norm": 0.2644675615376364,
      "learning_rate": 1.7318475961803663e-07,
      "loss": 1.5726,
      "step": 4306
    },
    {
      "epoch": 0.6495739386169972,
      "grad_norm": 0.27290084390136327,
      "learning_rate": 1.7308969612828677e-07,
      "loss": 1.5,
      "step": 4307
    },
    {
      "epoch": 0.6497247568056708,
      "grad_norm": 0.40749283237853495,
      "learning_rate": 1.7299465552036803e-07,
      "loss": 1.5564,
      "step": 4308
    },
    {
      "epoch": 0.6498755749943443,
      "grad_norm": 0.24428773072085164,
      "learning_rate": 1.728996378156199e-07,
      "loss": 1.5298,
      "step": 4309
    },
    {
      "epoch": 0.6500263931830179,
      "grad_norm": 0.2644942551318615,
      "learning_rate": 1.7280464303537662e-07,
      "loss": 1.5577,
      "step": 4310
    },
    {
      "epoch": 0.6501772113716914,
      "grad_norm": 0.24525025933050998,
      "learning_rate": 1.7270967120096713e-07,
      "loss": 1.5385,
      "step": 4311
    },
    {
      "epoch": 0.650328029560365,
      "grad_norm": 0.23709008992363353,
      "learning_rate": 1.7261472233371553e-07,
      "loss": 1.5163,
      "step": 4312
    },
    {
      "epoch": 0.6504788477490385,
      "grad_norm": 1.4112663261911442,
      "learning_rate": 1.7251979645494064e-07,
      "loss": 1.5637,
      "step": 4313
    },
    {
      "epoch": 0.6506296659377121,
      "grad_norm": 0.256267568489337,
      "learning_rate": 1.7242489358595587e-07,
      "loss": 1.6153,
      "step": 4314
    },
    {
      "epoch": 0.6507804841263857,
      "grad_norm": 0.2660453656221649,
      "learning_rate": 1.7233001374806988e-07,
      "loss": 1.6332,
      "step": 4315
    },
    {
      "epoch": 0.6509313023150592,
      "grad_norm": 0.46434640586819476,
      "learning_rate": 1.722351569625859e-07,
      "loss": 1.6546,
      "step": 4316
    },
    {
      "epoch": 0.6510821205037327,
      "grad_norm": 0.24542522675788647,
      "learning_rate": 1.7214032325080202e-07,
      "loss": 1.5376,
      "step": 4317
    },
    {
      "epoch": 0.6512329386924063,
      "grad_norm": 0.3163164998179516,
      "learning_rate": 1.7204551263401108e-07,
      "loss": 1.5534,
      "step": 4318
    },
    {
      "epoch": 0.6513837568810799,
      "grad_norm": 0.24978001107360961,
      "learning_rate": 1.71950725133501e-07,
      "loss": 1.5299,
      "step": 4319
    },
    {
      "epoch": 0.6515345750697534,
      "grad_norm": 0.5574009388465192,
      "learning_rate": 1.7185596077055425e-07,
      "loss": 1.6008,
      "step": 4320
    },
    {
      "epoch": 0.651685393258427,
      "grad_norm": 0.25102482710261503,
      "learning_rate": 1.7176121956644812e-07,
      "loss": 1.525,
      "step": 4321
    },
    {
      "epoch": 0.6518362114471005,
      "grad_norm": 0.2556993576030097,
      "learning_rate": 1.7166650154245492e-07,
      "loss": 1.5854,
      "step": 4322
    },
    {
      "epoch": 0.651987029635774,
      "grad_norm": 0.7906151359618411,
      "learning_rate": 1.7157180671984156e-07,
      "loss": 1.5772,
      "step": 4323
    },
    {
      "epoch": 0.6521378478244476,
      "grad_norm": 0.25693223933049275,
      "learning_rate": 1.7147713511986967e-07,
      "loss": 1.5045,
      "step": 4324
    },
    {
      "epoch": 0.6522886660131212,
      "grad_norm": 0.24547260528732132,
      "learning_rate": 1.713824867637959e-07,
      "loss": 1.6148,
      "step": 4325
    },
    {
      "epoch": 0.6524394842017948,
      "grad_norm": 0.24976059592113195,
      "learning_rate": 1.7128786167287158e-07,
      "loss": 1.5302,
      "step": 4326
    },
    {
      "epoch": 0.6525903023904683,
      "grad_norm": 0.256866786888526,
      "learning_rate": 1.711932598683426e-07,
      "loss": 1.5909,
      "step": 4327
    },
    {
      "epoch": 0.6527411205791418,
      "grad_norm": 0.2536906065453723,
      "learning_rate": 1.710986813714501e-07,
      "loss": 1.5116,
      "step": 4328
    },
    {
      "epoch": 0.6528919387678154,
      "grad_norm": 0.2674631416210052,
      "learning_rate": 1.7100412620342957e-07,
      "loss": 1.5351,
      "step": 4329
    },
    {
      "epoch": 0.653042756956489,
      "grad_norm": 0.25525115253926994,
      "learning_rate": 1.709095943855114e-07,
      "loss": 1.6358,
      "step": 4330
    },
    {
      "epoch": 0.6531935751451625,
      "grad_norm": 1.4586056729239687,
      "learning_rate": 1.7081508593892064e-07,
      "loss": 1.4906,
      "step": 4331
    },
    {
      "epoch": 0.6533443933338361,
      "grad_norm": 0.2703883011500383,
      "learning_rate": 1.7072060088487734e-07,
      "loss": 1.5704,
      "step": 4332
    },
    {
      "epoch": 0.6534952115225097,
      "grad_norm": 0.29758567824060134,
      "learning_rate": 1.7062613924459613e-07,
      "loss": 1.6118,
      "step": 4333
    },
    {
      "epoch": 0.6536460297111831,
      "grad_norm": 0.3897839427934797,
      "learning_rate": 1.7053170103928627e-07,
      "loss": 1.6127,
      "step": 4334
    },
    {
      "epoch": 0.6537968478998567,
      "grad_norm": 0.25338403351835703,
      "learning_rate": 1.7043728629015207e-07,
      "loss": 1.6034,
      "step": 4335
    },
    {
      "epoch": 0.6539476660885303,
      "grad_norm": 0.24361921837387265,
      "learning_rate": 1.7034289501839234e-07,
      "loss": 1.5631,
      "step": 4336
    },
    {
      "epoch": 0.6540984842772039,
      "grad_norm": 0.23947647600801733,
      "learning_rate": 1.702485272452005e-07,
      "loss": 1.4588,
      "step": 4337
    },
    {
      "epoch": 0.6542493024658774,
      "grad_norm": 0.25680951317289713,
      "learning_rate": 1.7015418299176507e-07,
      "loss": 1.5945,
      "step": 4338
    },
    {
      "epoch": 0.6544001206545509,
      "grad_norm": 0.25021590941728317,
      "learning_rate": 1.7005986227926902e-07,
      "loss": 1.5184,
      "step": 4339
    },
    {
      "epoch": 0.6545509388432245,
      "grad_norm": 0.2856791616026467,
      "learning_rate": 1.6996556512889e-07,
      "loss": 1.4966,
      "step": 4340
    },
    {
      "epoch": 0.654701757031898,
      "grad_norm": 0.27081040180416743,
      "learning_rate": 1.6987129156180063e-07,
      "loss": 1.4834,
      "step": 4341
    },
    {
      "epoch": 0.6548525752205716,
      "grad_norm": 0.25935752153647074,
      "learning_rate": 1.6977704159916801e-07,
      "loss": 1.5971,
      "step": 4342
    },
    {
      "epoch": 0.6550033934092452,
      "grad_norm": 0.24553600502404302,
      "learning_rate": 1.6968281526215393e-07,
      "loss": 1.5911,
      "step": 4343
    },
    {
      "epoch": 0.6551542115979188,
      "grad_norm": 0.2558518932840195,
      "learning_rate": 1.6958861257191497e-07,
      "loss": 1.5687,
      "step": 4344
    },
    {
      "epoch": 0.6553050297865922,
      "grad_norm": 0.2589811311294846,
      "learning_rate": 1.6949443354960247e-07,
      "loss": 1.5707,
      "step": 4345
    },
    {
      "epoch": 0.6554558479752658,
      "grad_norm": 0.24884607307303297,
      "learning_rate": 1.6940027821636228e-07,
      "loss": 1.5615,
      "step": 4346
    },
    {
      "epoch": 0.6556066661639394,
      "grad_norm": 0.2647618922190364,
      "learning_rate": 1.6930614659333498e-07,
      "loss": 1.5405,
      "step": 4347
    },
    {
      "epoch": 0.6557574843526129,
      "grad_norm": 0.2623678103868431,
      "learning_rate": 1.69212038701656e-07,
      "loss": 1.539,
      "step": 4348
    },
    {
      "epoch": 0.6559083025412865,
      "grad_norm": 0.25316236256720226,
      "learning_rate": 1.691179545624552e-07,
      "loss": 1.5687,
      "step": 4349
    },
    {
      "epoch": 0.6560591207299601,
      "grad_norm": 0.25834994094591196,
      "learning_rate": 1.6902389419685716e-07,
      "loss": 1.5376,
      "step": 4350
    },
    {
      "epoch": 0.6562099389186336,
      "grad_norm": 0.2547110661971648,
      "learning_rate": 1.6892985762598133e-07,
      "loss": 1.5755,
      "step": 4351
    },
    {
      "epoch": 0.6563607571073071,
      "grad_norm": 0.3476854271744261,
      "learning_rate": 1.6883584487094154e-07,
      "loss": 1.5526,
      "step": 4352
    },
    {
      "epoch": 0.6565115752959807,
      "grad_norm": 0.2509408636464211,
      "learning_rate": 1.687418559528463e-07,
      "loss": 1.4926,
      "step": 4353
    },
    {
      "epoch": 0.6566623934846543,
      "grad_norm": 0.24533886050655487,
      "learning_rate": 1.6864789089279913e-07,
      "loss": 1.5253,
      "step": 4354
    },
    {
      "epoch": 0.6568132116733278,
      "grad_norm": 0.2696381455151316,
      "learning_rate": 1.6855394971189778e-07,
      "loss": 1.5332,
      "step": 4355
    },
    {
      "epoch": 0.6569640298620013,
      "grad_norm": 0.24429394764520096,
      "learning_rate": 1.6846003243123472e-07,
      "loss": 1.5204,
      "step": 4356
    },
    {
      "epoch": 0.6571148480506749,
      "grad_norm": 0.2594783788516782,
      "learning_rate": 1.6836613907189712e-07,
      "loss": 1.557,
      "step": 4357
    },
    {
      "epoch": 0.6572656662393485,
      "grad_norm": 0.2717059267374578,
      "learning_rate": 1.6827226965496686e-07,
      "loss": 1.5728,
      "step": 4358
    },
    {
      "epoch": 0.657416484428022,
      "grad_norm": 0.24778293229119935,
      "learning_rate": 1.6817842420152035e-07,
      "loss": 1.5528,
      "step": 4359
    },
    {
      "epoch": 0.6575673026166956,
      "grad_norm": 0.2553800911389374,
      "learning_rate": 1.6808460273262847e-07,
      "loss": 1.5626,
      "step": 4360
    },
    {
      "epoch": 0.6577181208053692,
      "grad_norm": 0.27268535161135665,
      "learning_rate": 1.6799080526935713e-07,
      "loss": 1.5631,
      "step": 4361
    },
    {
      "epoch": 0.6578689389940426,
      "grad_norm": 0.2548385837394829,
      "learning_rate": 1.6789703183276647e-07,
      "loss": 1.5719,
      "step": 4362
    },
    {
      "epoch": 0.6580197571827162,
      "grad_norm": 0.26768870655081467,
      "learning_rate": 1.6780328244391123e-07,
      "loss": 1.544,
      "step": 4363
    },
    {
      "epoch": 0.6581705753713898,
      "grad_norm": 0.23771140979603766,
      "learning_rate": 1.6770955712384105e-07,
      "loss": 1.5666,
      "step": 4364
    },
    {
      "epoch": 0.6583213935600634,
      "grad_norm": 0.2412574229422365,
      "learning_rate": 1.6761585589359997e-07,
      "loss": 1.5838,
      "step": 4365
    },
    {
      "epoch": 0.6584722117487369,
      "grad_norm": 0.26441662184861997,
      "learning_rate": 1.6752217877422665e-07,
      "loss": 1.537,
      "step": 4366
    },
    {
      "epoch": 0.6586230299374104,
      "grad_norm": 0.24615839066654882,
      "learning_rate": 1.6742852578675415e-07,
      "loss": 1.5528,
      "step": 4367
    },
    {
      "epoch": 0.658773848126084,
      "grad_norm": 0.24312429921517675,
      "learning_rate": 1.6733489695221054e-07,
      "loss": 1.4885,
      "step": 4368
    },
    {
      "epoch": 0.6589246663147575,
      "grad_norm": 0.25986900550707687,
      "learning_rate": 1.6724129229161815e-07,
      "loss": 1.5759,
      "step": 4369
    },
    {
      "epoch": 0.6590754845034311,
      "grad_norm": 0.2617579223876186,
      "learning_rate": 1.671477118259938e-07,
      "loss": 1.5496,
      "step": 4370
    },
    {
      "epoch": 0.6592263026921047,
      "grad_norm": 0.2621628735602423,
      "learning_rate": 1.6705415557634926e-07,
      "loss": 1.4728,
      "step": 4371
    },
    {
      "epoch": 0.6593771208807783,
      "grad_norm": 0.25976189558002166,
      "learning_rate": 1.6696062356369052e-07,
      "loss": 1.6121,
      "step": 4372
    },
    {
      "epoch": 0.6595279390694517,
      "grad_norm": 0.25191569383887047,
      "learning_rate": 1.6686711580901816e-07,
      "loss": 1.5294,
      "step": 4373
    },
    {
      "epoch": 0.6596787572581253,
      "grad_norm": 1.6830369004072032,
      "learning_rate": 1.6677363233332753e-07,
      "loss": 1.5476,
      "step": 4374
    },
    {
      "epoch": 0.6598295754467989,
      "grad_norm": 0.24972879055668815,
      "learning_rate": 1.6668017315760837e-07,
      "loss": 1.5387,
      "step": 4375
    },
    {
      "epoch": 0.6599803936354725,
      "grad_norm": 0.2529044244935057,
      "learning_rate": 1.665867383028449e-07,
      "loss": 1.5556,
      "step": 4376
    },
    {
      "epoch": 0.660131211824146,
      "grad_norm": 0.24237828487076207,
      "learning_rate": 1.6649332779001606e-07,
      "loss": 1.5094,
      "step": 4377
    },
    {
      "epoch": 0.6602820300128196,
      "grad_norm": 0.2516651522330598,
      "learning_rate": 1.663999416400952e-07,
      "loss": 1.5412,
      "step": 4378
    },
    {
      "epoch": 0.6604328482014931,
      "grad_norm": 0.2472022069889934,
      "learning_rate": 1.6630657987405027e-07,
      "loss": 1.5465,
      "step": 4379
    },
    {
      "epoch": 0.6605836663901666,
      "grad_norm": 0.42366911674571833,
      "learning_rate": 1.6621324251284355e-07,
      "loss": 1.5475,
      "step": 4380
    },
    {
      "epoch": 0.6607344845788402,
      "grad_norm": 0.24492696411215548,
      "learning_rate": 1.6611992957743214e-07,
      "loss": 1.5285,
      "step": 4381
    },
    {
      "epoch": 0.6608853027675138,
      "grad_norm": 0.2567745155121796,
      "learning_rate": 1.6602664108876757e-07,
      "loss": 1.4723,
      "step": 4382
    },
    {
      "epoch": 0.6610361209561874,
      "grad_norm": 0.2482074703553842,
      "learning_rate": 1.659333770677956e-07,
      "loss": 1.6355,
      "step": 4383
    },
    {
      "epoch": 0.6611869391448608,
      "grad_norm": 0.24325576127420598,
      "learning_rate": 1.658401375354569e-07,
      "loss": 1.5417,
      "step": 4384
    },
    {
      "epoch": 0.6613377573335344,
      "grad_norm": 0.24726061523357257,
      "learning_rate": 1.6574692251268645e-07,
      "loss": 1.5517,
      "step": 4385
    },
    {
      "epoch": 0.661488575522208,
      "grad_norm": 0.24787383673380384,
      "learning_rate": 1.6565373202041364e-07,
      "loss": 1.5424,
      "step": 4386
    },
    {
      "epoch": 0.6616393937108815,
      "grad_norm": 0.2546839322501766,
      "learning_rate": 1.6556056607956254e-07,
      "loss": 1.5875,
      "step": 4387
    },
    {
      "epoch": 0.6617902118995551,
      "grad_norm": 0.2530227398838417,
      "learning_rate": 1.654674247110516e-07,
      "loss": 1.5844,
      "step": 4388
    },
    {
      "epoch": 0.6619410300882287,
      "grad_norm": 0.2472382108263328,
      "learning_rate": 1.6537430793579367e-07,
      "loss": 1.558,
      "step": 4389
    },
    {
      "epoch": 0.6620918482769022,
      "grad_norm": 0.2467916743913005,
      "learning_rate": 1.6528121577469633e-07,
      "loss": 1.6053,
      "step": 4390
    },
    {
      "epoch": 0.6622426664655757,
      "grad_norm": 0.2806261435847341,
      "learning_rate": 1.6518814824866144e-07,
      "loss": 1.5899,
      "step": 4391
    },
    {
      "epoch": 0.6623934846542493,
      "grad_norm": 0.2902261400426562,
      "learning_rate": 1.6509510537858537e-07,
      "loss": 1.5363,
      "step": 4392
    },
    {
      "epoch": 0.6625443028429229,
      "grad_norm": 0.2708674758772053,
      "learning_rate": 1.650020871853588e-07,
      "loss": 1.5422,
      "step": 4393
    },
    {
      "epoch": 0.6626951210315964,
      "grad_norm": 0.24276755358204255,
      "learning_rate": 1.649090936898672e-07,
      "loss": 1.5363,
      "step": 4394
    },
    {
      "epoch": 0.66284593922027,
      "grad_norm": 0.24804347418554512,
      "learning_rate": 1.6481612491299036e-07,
      "loss": 1.5663,
      "step": 4395
    },
    {
      "epoch": 0.6629967574089435,
      "grad_norm": 0.25096976238548924,
      "learning_rate": 1.647231808756022e-07,
      "loss": 1.5343,
      "step": 4396
    },
    {
      "epoch": 0.6631475755976171,
      "grad_norm": 0.24496843643929367,
      "learning_rate": 1.6463026159857172e-07,
      "loss": 1.5362,
      "step": 4397
    },
    {
      "epoch": 0.6632983937862906,
      "grad_norm": 0.25238739350256006,
      "learning_rate": 1.645373671027618e-07,
      "loss": 1.6324,
      "step": 4398
    },
    {
      "epoch": 0.6634492119749642,
      "grad_norm": 0.25179695590369966,
      "learning_rate": 1.6444449740902987e-07,
      "loss": 1.5586,
      "step": 4399
    },
    {
      "epoch": 0.6636000301636378,
      "grad_norm": 0.24998595205338636,
      "learning_rate": 1.6435165253822813e-07,
      "loss": 1.5956,
      "step": 4400
    },
    {
      "epoch": 0.6637508483523112,
      "grad_norm": 0.2984166708920905,
      "learning_rate": 1.6425883251120285e-07,
      "loss": 1.5744,
      "step": 4401
    },
    {
      "epoch": 0.6639016665409848,
      "grad_norm": 0.24872797484434014,
      "learning_rate": 1.6416603734879469e-07,
      "loss": 1.5456,
      "step": 4402
    },
    {
      "epoch": 0.6640524847296584,
      "grad_norm": 0.31371587804843853,
      "learning_rate": 1.6407326707183907e-07,
      "loss": 1.5451,
      "step": 4403
    },
    {
      "epoch": 0.664203302918332,
      "grad_norm": 0.25024145507022,
      "learning_rate": 1.6398052170116556e-07,
      "loss": 1.5041,
      "step": 4404
    },
    {
      "epoch": 0.6643541211070055,
      "grad_norm": 0.252617873762467,
      "learning_rate": 1.6388780125759816e-07,
      "loss": 1.5515,
      "step": 4405
    },
    {
      "epoch": 0.6645049392956791,
      "grad_norm": 0.2579091788972358,
      "learning_rate": 1.6379510576195524e-07,
      "loss": 1.549,
      "step": 4406
    },
    {
      "epoch": 0.6646557574843526,
      "grad_norm": 0.2406197129149719,
      "learning_rate": 1.6370243523504977e-07,
      "loss": 1.5341,
      "step": 4407
    },
    {
      "epoch": 0.6648065756730261,
      "grad_norm": 0.2543884142132946,
      "learning_rate": 1.6360978969768902e-07,
      "loss": 1.5269,
      "step": 4408
    },
    {
      "epoch": 0.6649573938616997,
      "grad_norm": 0.2531155660170818,
      "learning_rate": 1.6351716917067438e-07,
      "loss": 1.604,
      "step": 4409
    },
    {
      "epoch": 0.6651082120503733,
      "grad_norm": 0.25200196374912637,
      "learning_rate": 1.6342457367480207e-07,
      "loss": 1.5982,
      "step": 4410
    },
    {
      "epoch": 0.6652590302390469,
      "grad_norm": 0.2594260257387867,
      "learning_rate": 1.6333200323086244e-07,
      "loss": 1.5579,
      "step": 4411
    },
    {
      "epoch": 0.6654098484277203,
      "grad_norm": 0.2555082936930443,
      "learning_rate": 1.632394578596401e-07,
      "loss": 1.4944,
      "step": 4412
    },
    {
      "epoch": 0.6655606666163939,
      "grad_norm": 0.26174160023642556,
      "learning_rate": 1.6314693758191442e-07,
      "loss": 1.5242,
      "step": 4413
    },
    {
      "epoch": 0.6657114848050675,
      "grad_norm": 0.24320823578852807,
      "learning_rate": 1.6305444241845875e-07,
      "loss": 1.6164,
      "step": 4414
    },
    {
      "epoch": 0.665862302993741,
      "grad_norm": 0.23972350200655893,
      "learning_rate": 1.629619723900409e-07,
      "loss": 1.5389,
      "step": 4415
    },
    {
      "epoch": 0.6660131211824146,
      "grad_norm": 0.24819399239616022,
      "learning_rate": 1.628695275174232e-07,
      "loss": 1.539,
      "step": 4416
    },
    {
      "epoch": 0.6661639393710882,
      "grad_norm": 0.2738680504735075,
      "learning_rate": 1.6277710782136223e-07,
      "loss": 1.5313,
      "step": 4417
    },
    {
      "epoch": 0.6663147575597617,
      "grad_norm": 0.29736483013703435,
      "learning_rate": 1.6268471332260882e-07,
      "loss": 1.5923,
      "step": 4418
    },
    {
      "epoch": 0.6664655757484352,
      "grad_norm": 0.2691164164594476,
      "learning_rate": 1.6259234404190815e-07,
      "loss": 1.5661,
      "step": 4419
    },
    {
      "epoch": 0.6666163939371088,
      "grad_norm": 0.24407951769351988,
      "learning_rate": 1.6250000000000006e-07,
      "loss": 1.5182,
      "step": 4420
    },
    {
      "epoch": 0.6667672121257824,
      "grad_norm": 0.3036247360046378,
      "learning_rate": 1.6240768121761826e-07,
      "loss": 1.5337,
      "step": 4421
    },
    {
      "epoch": 0.666918030314456,
      "grad_norm": 0.2681579943835327,
      "learning_rate": 1.6231538771549104e-07,
      "loss": 1.5705,
      "step": 4422
    },
    {
      "epoch": 0.6670688485031295,
      "grad_norm": 0.25816631047547445,
      "learning_rate": 1.6222311951434108e-07,
      "loss": 1.563,
      "step": 4423
    },
    {
      "epoch": 0.667219666691803,
      "grad_norm": 0.25013242748167597,
      "learning_rate": 1.6213087663488522e-07,
      "loss": 1.5577,
      "step": 4424
    },
    {
      "epoch": 0.6673704848804766,
      "grad_norm": 0.2455927102819475,
      "learning_rate": 1.620386590978346e-07,
      "loss": 1.536,
      "step": 4425
    },
    {
      "epoch": 0.6675213030691501,
      "grad_norm": 0.24801865500769413,
      "learning_rate": 1.6194646692389484e-07,
      "loss": 1.4987,
      "step": 4426
    },
    {
      "epoch": 0.6676721212578237,
      "grad_norm": 0.24911440013001487,
      "learning_rate": 1.6185430013376582e-07,
      "loss": 1.5705,
      "step": 4427
    },
    {
      "epoch": 0.6678229394464973,
      "grad_norm": 0.2628882747468611,
      "learning_rate": 1.6176215874814146e-07,
      "loss": 1.5372,
      "step": 4428
    },
    {
      "epoch": 0.6679737576351708,
      "grad_norm": 0.2595634659919045,
      "learning_rate": 1.6167004278771037e-07,
      "loss": 1.5658,
      "step": 4429
    },
    {
      "epoch": 0.6681245758238443,
      "grad_norm": 0.29775413439795473,
      "learning_rate": 1.6157795227315524e-07,
      "loss": 1.5422,
      "step": 4430
    },
    {
      "epoch": 0.6682753940125179,
      "grad_norm": 0.2690934522371737,
      "learning_rate": 1.6148588722515304e-07,
      "loss": 1.5789,
      "step": 4431
    },
    {
      "epoch": 0.6684262122011915,
      "grad_norm": 0.25994584944910876,
      "learning_rate": 1.6139384766437497e-07,
      "loss": 1.5574,
      "step": 4432
    },
    {
      "epoch": 0.668577030389865,
      "grad_norm": 0.49385878510095477,
      "learning_rate": 1.6130183361148672e-07,
      "loss": 1.6128,
      "step": 4433
    },
    {
      "epoch": 0.6687278485785386,
      "grad_norm": 0.26612645215918884,
      "learning_rate": 1.6120984508714814e-07,
      "loss": 1.588,
      "step": 4434
    },
    {
      "epoch": 0.6688786667672121,
      "grad_norm": 0.2466698877353191,
      "learning_rate": 1.6111788211201317e-07,
      "loss": 1.5513,
      "step": 4435
    },
    {
      "epoch": 0.6690294849558857,
      "grad_norm": 0.3837798363225113,
      "learning_rate": 1.6102594470673031e-07,
      "loss": 1.5866,
      "step": 4436
    },
    {
      "epoch": 0.6691803031445592,
      "grad_norm": 0.24367217671826055,
      "learning_rate": 1.6093403289194225e-07,
      "loss": 1.5122,
      "step": 4437
    },
    {
      "epoch": 0.6693311213332328,
      "grad_norm": 0.24662484557225653,
      "learning_rate": 1.608421466882856e-07,
      "loss": 1.5657,
      "step": 4438
    },
    {
      "epoch": 0.6694819395219064,
      "grad_norm": 0.24653542481933985,
      "learning_rate": 1.607502861163918e-07,
      "loss": 1.55,
      "step": 4439
    },
    {
      "epoch": 0.66963275771058,
      "grad_norm": 0.26100371071233147,
      "learning_rate": 1.606584511968861e-07,
      "loss": 1.5384,
      "step": 4440
    },
    {
      "epoch": 0.6697835758992534,
      "grad_norm": 0.2510751090311021,
      "learning_rate": 1.6056664195038797e-07,
      "loss": 1.5858,
      "step": 4441
    },
    {
      "epoch": 0.669934394087927,
      "grad_norm": 0.25432632198219307,
      "learning_rate": 1.6047485839751148e-07,
      "loss": 1.5679,
      "step": 4442
    },
    {
      "epoch": 0.6700852122766006,
      "grad_norm": 0.2613752320293682,
      "learning_rate": 1.6038310055886467e-07,
      "loss": 1.6188,
      "step": 4443
    },
    {
      "epoch": 0.6702360304652741,
      "grad_norm": 0.25383575216431087,
      "learning_rate": 1.6029136845504974e-07,
      "loss": 1.5303,
      "step": 4444
    },
    {
      "epoch": 0.6703868486539477,
      "grad_norm": 0.28151295313508107,
      "learning_rate": 1.601996621066632e-07,
      "loss": 1.5927,
      "step": 4445
    },
    {
      "epoch": 0.6705376668426212,
      "grad_norm": 0.26884315365474487,
      "learning_rate": 1.6010798153429597e-07,
      "loss": 1.5429,
      "step": 4446
    },
    {
      "epoch": 0.6706884850312947,
      "grad_norm": 0.25801191861820594,
      "learning_rate": 1.6001632675853295e-07,
      "loss": 1.627,
      "step": 4447
    },
    {
      "epoch": 0.6708393032199683,
      "grad_norm": 0.2493698982634603,
      "learning_rate": 1.5992469779995312e-07,
      "loss": 1.5584,
      "step": 4448
    },
    {
      "epoch": 0.6709901214086419,
      "grad_norm": 0.2488284571200544,
      "learning_rate": 1.5983309467913008e-07,
      "loss": 1.5873,
      "step": 4449
    },
    {
      "epoch": 0.6711409395973155,
      "grad_norm": 0.25185762106130766,
      "learning_rate": 1.597415174166314e-07,
      "loss": 1.6337,
      "step": 4450
    },
    {
      "epoch": 0.671291757785989,
      "grad_norm": 0.24107801326174988,
      "learning_rate": 1.5964996603301857e-07,
      "loss": 1.5663,
      "step": 4451
    },
    {
      "epoch": 0.6714425759746625,
      "grad_norm": 0.23335337442589293,
      "learning_rate": 1.5955844054884788e-07,
      "loss": 1.6478,
      "step": 4452
    },
    {
      "epoch": 0.6715933941633361,
      "grad_norm": 0.24953066525510237,
      "learning_rate": 1.594669409846693e-07,
      "loss": 1.6057,
      "step": 4453
    },
    {
      "epoch": 0.6717442123520097,
      "grad_norm": 0.2537367589132101,
      "learning_rate": 1.5937546736102708e-07,
      "loss": 1.6144,
      "step": 4454
    },
    {
      "epoch": 0.6718950305406832,
      "grad_norm": 0.26427627828933026,
      "learning_rate": 1.5928401969845977e-07,
      "loss": 1.593,
      "step": 4455
    },
    {
      "epoch": 0.6720458487293568,
      "grad_norm": 0.252361012839834,
      "learning_rate": 1.591925980175002e-07,
      "loss": 1.5249,
      "step": 4456
    },
    {
      "epoch": 0.6721966669180303,
      "grad_norm": 0.24555805311657805,
      "learning_rate": 1.5910120233867501e-07,
      "loss": 1.5458,
      "step": 4457
    },
    {
      "epoch": 0.6723474851067038,
      "grad_norm": 0.2606785058051602,
      "learning_rate": 1.5900983268250518e-07,
      "loss": 1.6235,
      "step": 4458
    },
    {
      "epoch": 0.6724983032953774,
      "grad_norm": 0.24333960694058768,
      "learning_rate": 1.58918489069506e-07,
      "loss": 1.5034,
      "step": 4459
    },
    {
      "epoch": 0.672649121484051,
      "grad_norm": 0.305832013139834,
      "learning_rate": 1.5882717152018666e-07,
      "loss": 1.4752,
      "step": 4460
    },
    {
      "epoch": 0.6727999396727246,
      "grad_norm": 0.25273299194813803,
      "learning_rate": 1.5873588005505067e-07,
      "loss": 1.6172,
      "step": 4461
    },
    {
      "epoch": 0.6729507578613981,
      "grad_norm": 0.24424501947561303,
      "learning_rate": 1.586446146945955e-07,
      "loss": 1.6006,
      "step": 4462
    },
    {
      "epoch": 0.6731015760500716,
      "grad_norm": 0.24893274250145397,
      "learning_rate": 1.5855337545931308e-07,
      "loss": 1.5378,
      "step": 4463
    },
    {
      "epoch": 0.6732523942387452,
      "grad_norm": 2.34854981138626,
      "learning_rate": 1.5846216236968913e-07,
      "loss": 1.5606,
      "step": 4464
    },
    {
      "epoch": 0.6734032124274187,
      "grad_norm": 0.25952343083126733,
      "learning_rate": 1.5837097544620364e-07,
      "loss": 1.5542,
      "step": 4465
    },
    {
      "epoch": 0.6735540306160923,
      "grad_norm": 0.24166214284987536,
      "learning_rate": 1.5827981470933083e-07,
      "loss": 1.4965,
      "step": 4466
    },
    {
      "epoch": 0.6737048488047659,
      "grad_norm": 0.2623262912298193,
      "learning_rate": 1.581886801795389e-07,
      "loss": 1.5544,
      "step": 4467
    },
    {
      "epoch": 0.6738556669934395,
      "grad_norm": 0.2812617083911704,
      "learning_rate": 1.5809757187729013e-07,
      "loss": 1.574,
      "step": 4468
    },
    {
      "epoch": 0.6740064851821129,
      "grad_norm": 0.5291270399283328,
      "learning_rate": 1.580064898230411e-07,
      "loss": 1.5888,
      "step": 4469
    },
    {
      "epoch": 0.6741573033707865,
      "grad_norm": 0.40802473392185296,
      "learning_rate": 1.5791543403724233e-07,
      "loss": 1.6316,
      "step": 4470
    },
    {
      "epoch": 0.6743081215594601,
      "grad_norm": 0.2843350286941599,
      "learning_rate": 1.5782440454033847e-07,
      "loss": 1.5192,
      "step": 4471
    },
    {
      "epoch": 0.6744589397481336,
      "grad_norm": 0.41370144878011966,
      "learning_rate": 1.5773340135276836e-07,
      "loss": 1.5708,
      "step": 4472
    },
    {
      "epoch": 0.6746097579368072,
      "grad_norm": 0.2621786014549776,
      "learning_rate": 1.576424244949648e-07,
      "loss": 1.6857,
      "step": 4473
    },
    {
      "epoch": 0.6747605761254807,
      "grad_norm": 0.2539712941682984,
      "learning_rate": 1.575514739873548e-07,
      "loss": 1.5274,
      "step": 4474
    },
    {
      "epoch": 0.6749113943141543,
      "grad_norm": 0.262397966494765,
      "learning_rate": 1.5746054985035929e-07,
      "loss": 1.5127,
      "step": 4475
    },
    {
      "epoch": 0.6750622125028278,
      "grad_norm": 0.2988519798684835,
      "learning_rate": 1.5736965210439355e-07,
      "loss": 1.5159,
      "step": 4476
    },
    {
      "epoch": 0.6752130306915014,
      "grad_norm": 0.2625430928739007,
      "learning_rate": 1.5727878076986667e-07,
      "loss": 1.5121,
      "step": 4477
    },
    {
      "epoch": 0.675363848880175,
      "grad_norm": 0.2693026704994772,
      "learning_rate": 1.5718793586718187e-07,
      "loss": 1.5429,
      "step": 4478
    },
    {
      "epoch": 0.6755146670688486,
      "grad_norm": 0.24741617489781206,
      "learning_rate": 1.5709711741673658e-07,
      "loss": 1.6317,
      "step": 4479
    },
    {
      "epoch": 0.675665485257522,
      "grad_norm": 0.24554929918868298,
      "learning_rate": 1.5700632543892216e-07,
      "loss": 1.5224,
      "step": 4480
    },
    {
      "epoch": 0.6758163034461956,
      "grad_norm": 0.2531089314030115,
      "learning_rate": 1.5691555995412397e-07,
      "loss": 1.5566,
      "step": 4481
    },
    {
      "epoch": 0.6759671216348692,
      "grad_norm": 0.270813179835107,
      "learning_rate": 1.5682482098272166e-07,
      "loss": 1.5137,
      "step": 4482
    },
    {
      "epoch": 0.6761179398235427,
      "grad_norm": 0.2531790176228573,
      "learning_rate": 1.5673410854508862e-07,
      "loss": 1.5078,
      "step": 4483
    },
    {
      "epoch": 0.6762687580122163,
      "grad_norm": 0.24704562757704143,
      "learning_rate": 1.5664342266159253e-07,
      "loss": 1.5327,
      "step": 4484
    },
    {
      "epoch": 0.6764195762008899,
      "grad_norm": 0.31259419416536566,
      "learning_rate": 1.5655276335259493e-07,
      "loss": 1.5217,
      "step": 4485
    },
    {
      "epoch": 0.6765703943895633,
      "grad_norm": 0.29975032453330397,
      "learning_rate": 1.5646213063845157e-07,
      "loss": 1.5935,
      "step": 4486
    },
    {
      "epoch": 0.6767212125782369,
      "grad_norm": 0.25402666044063815,
      "learning_rate": 1.563715245395121e-07,
      "loss": 1.4853,
      "step": 4487
    },
    {
      "epoch": 0.6768720307669105,
      "grad_norm": 0.8082627996462705,
      "learning_rate": 1.562809450761201e-07,
      "loss": 1.5272,
      "step": 4488
    },
    {
      "epoch": 0.6770228489555841,
      "grad_norm": 0.2522359601460765,
      "learning_rate": 1.5619039226861351e-07,
      "loss": 1.5272,
      "step": 4489
    },
    {
      "epoch": 0.6771736671442576,
      "grad_norm": 0.2432005732759633,
      "learning_rate": 1.5609986613732402e-07,
      "loss": 1.5519,
      "step": 4490
    },
    {
      "epoch": 0.6773244853329311,
      "grad_norm": 0.2530184344932088,
      "learning_rate": 1.5600936670257718e-07,
      "loss": 1.6198,
      "step": 4491
    },
    {
      "epoch": 0.6774753035216047,
      "grad_norm": 0.2382296864813892,
      "learning_rate": 1.5591889398469302e-07,
      "loss": 1.5021,
      "step": 4492
    },
    {
      "epoch": 0.6776261217102783,
      "grad_norm": 0.2503873698254462,
      "learning_rate": 1.558284480039852e-07,
      "loss": 1.5769,
      "step": 4493
    },
    {
      "epoch": 0.6777769398989518,
      "grad_norm": 0.6478262856588054,
      "learning_rate": 1.5573802878076136e-07,
      "loss": 1.5224,
      "step": 4494
    },
    {
      "epoch": 0.6779277580876254,
      "grad_norm": 0.25990690300475844,
      "learning_rate": 1.556476363353234e-07,
      "loss": 1.5723,
      "step": 4495
    },
    {
      "epoch": 0.678078576276299,
      "grad_norm": 0.2762708223326884,
      "learning_rate": 1.555572706879671e-07,
      "loss": 1.462,
      "step": 4496
    },
    {
      "epoch": 0.6782293944649724,
      "grad_norm": 0.2563697519669219,
      "learning_rate": 1.5546693185898203e-07,
      "loss": 1.5847,
      "step": 4497
    },
    {
      "epoch": 0.678380212653646,
      "grad_norm": 0.28862133225427156,
      "learning_rate": 1.5537661986865187e-07,
      "loss": 1.5384,
      "step": 4498
    },
    {
      "epoch": 0.6785310308423196,
      "grad_norm": 0.24185984802208632,
      "learning_rate": 1.5528633473725443e-07,
      "loss": 1.6157,
      "step": 4499
    },
    {
      "epoch": 0.6786818490309932,
      "grad_norm": 0.24390163793301547,
      "learning_rate": 1.5519607648506133e-07,
      "loss": 1.5652,
      "step": 4500
    },
    {
      "epoch": 0.6788326672196667,
      "grad_norm": 0.25433487014020545,
      "learning_rate": 1.55105845132338e-07,
      "loss": 1.5813,
      "step": 4501
    },
    {
      "epoch": 0.6789834854083402,
      "grad_norm": 0.24768666349044535,
      "learning_rate": 1.550156406993443e-07,
      "loss": 1.5257,
      "step": 4502
    },
    {
      "epoch": 0.6791343035970138,
      "grad_norm": 0.25434925804032016,
      "learning_rate": 1.5492546320633355e-07,
      "loss": 1.5431,
      "step": 4503
    },
    {
      "epoch": 0.6792851217856873,
      "grad_norm": 0.25432855324045334,
      "learning_rate": 1.5483531267355315e-07,
      "loss": 1.5667,
      "step": 4504
    },
    {
      "epoch": 0.6794359399743609,
      "grad_norm": 0.29078320294039717,
      "learning_rate": 1.5474518912124474e-07,
      "loss": 1.5261,
      "step": 4505
    },
    {
      "epoch": 0.6795867581630345,
      "grad_norm": 0.2534399553146919,
      "learning_rate": 1.546550925696436e-07,
      "loss": 1.5477,
      "step": 4506
    },
    {
      "epoch": 0.6797375763517081,
      "grad_norm": 0.26475011017631184,
      "learning_rate": 1.5456502303897893e-07,
      "loss": 1.4716,
      "step": 4507
    },
    {
      "epoch": 0.6798883945403815,
      "grad_norm": 0.25023873514497286,
      "learning_rate": 1.544749805494741e-07,
      "loss": 1.5263,
      "step": 4508
    },
    {
      "epoch": 0.6800392127290551,
      "grad_norm": 0.4316287456279205,
      "learning_rate": 1.5438496512134625e-07,
      "loss": 1.5983,
      "step": 4509
    },
    {
      "epoch": 0.6801900309177287,
      "grad_norm": 0.2489370624450571,
      "learning_rate": 1.5429497677480638e-07,
      "loss": 1.4934,
      "step": 4510
    },
    {
      "epoch": 0.6803408491064022,
      "grad_norm": 0.3669846777729833,
      "learning_rate": 1.5420501553005948e-07,
      "loss": 1.5831,
      "step": 4511
    },
    {
      "epoch": 0.6804916672950758,
      "grad_norm": 0.2648456804985748,
      "learning_rate": 1.541150814073046e-07,
      "loss": 1.6421,
      "step": 4512
    },
    {
      "epoch": 0.6806424854837494,
      "grad_norm": 0.26796561522886525,
      "learning_rate": 1.5402517442673454e-07,
      "loss": 1.5622,
      "step": 4513
    },
    {
      "epoch": 0.6807933036724229,
      "grad_norm": 0.2525029359985612,
      "learning_rate": 1.539352946085359e-07,
      "loss": 1.5529,
      "step": 4514
    },
    {
      "epoch": 0.6809441218610964,
      "grad_norm": 0.3438247994299546,
      "learning_rate": 1.538454419728895e-07,
      "loss": 1.5502,
      "step": 4515
    },
    {
      "epoch": 0.68109494004977,
      "grad_norm": 0.3565079090528708,
      "learning_rate": 1.5375561653996977e-07,
      "loss": 1.5665,
      "step": 4516
    },
    {
      "epoch": 0.6812457582384436,
      "grad_norm": 0.2502136419414264,
      "learning_rate": 1.5366581832994508e-07,
      "loss": 1.5328,
      "step": 4517
    },
    {
      "epoch": 0.6813965764271172,
      "grad_norm": 0.25637007238519177,
      "learning_rate": 1.535760473629779e-07,
      "loss": 1.5678,
      "step": 4518
    },
    {
      "epoch": 0.6815473946157906,
      "grad_norm": 0.2542868537875946,
      "learning_rate": 1.5348630365922432e-07,
      "loss": 1.5661,
      "step": 4519
    },
    {
      "epoch": 0.6816982128044642,
      "grad_norm": 0.24491684976074926,
      "learning_rate": 1.5339658723883443e-07,
      "loss": 1.5301,
      "step": 4520
    },
    {
      "epoch": 0.6818490309931378,
      "grad_norm": 0.2532871996773783,
      "learning_rate": 1.533068981219522e-07,
      "loss": 1.5442,
      "step": 4521
    },
    {
      "epoch": 0.6819998491818113,
      "grad_norm": 0.2969754695815701,
      "learning_rate": 1.5321723632871546e-07,
      "loss": 1.5914,
      "step": 4522
    },
    {
      "epoch": 0.6821506673704849,
      "grad_norm": 0.23776153267951033,
      "learning_rate": 1.5312760187925586e-07,
      "loss": 1.5446,
      "step": 4523
    },
    {
      "epoch": 0.6823014855591585,
      "grad_norm": 0.2567751502907507,
      "learning_rate": 1.5303799479369888e-07,
      "loss": 1.633,
      "step": 4524
    },
    {
      "epoch": 0.682452303747832,
      "grad_norm": 0.24803655120611262,
      "learning_rate": 1.5294841509216412e-07,
      "loss": 1.606,
      "step": 4525
    },
    {
      "epoch": 0.6826031219365055,
      "grad_norm": 0.2502764130264785,
      "learning_rate": 1.5285886279476462e-07,
      "loss": 1.5575,
      "step": 4526
    },
    {
      "epoch": 0.6827539401251791,
      "grad_norm": 0.2552962726477028,
      "learning_rate": 1.5276933792160752e-07,
      "loss": 1.5266,
      "step": 4527
    },
    {
      "epoch": 0.6829047583138527,
      "grad_norm": 0.2618369174454743,
      "learning_rate": 1.5267984049279389e-07,
      "loss": 1.5161,
      "step": 4528
    },
    {
      "epoch": 0.6830555765025262,
      "grad_norm": 0.2469464741877154,
      "learning_rate": 1.5259037052841844e-07,
      "loss": 1.5408,
      "step": 4529
    },
    {
      "epoch": 0.6832063946911997,
      "grad_norm": 0.252624797216419,
      "learning_rate": 1.525009280485697e-07,
      "loss": 1.5406,
      "step": 4530
    },
    {
      "epoch": 0.6833572128798733,
      "grad_norm": 0.25393176829144687,
      "learning_rate": 1.5241151307333022e-07,
      "loss": 1.5946,
      "step": 4531
    },
    {
      "epoch": 0.6835080310685469,
      "grad_norm": 0.26189789836181204,
      "learning_rate": 1.5232212562277625e-07,
      "loss": 1.6088,
      "step": 4532
    },
    {
      "epoch": 0.6836588492572204,
      "grad_norm": 0.2542125862503045,
      "learning_rate": 1.522327657169778e-07,
      "loss": 1.546,
      "step": 4533
    },
    {
      "epoch": 0.683809667445894,
      "grad_norm": 0.29066485291620314,
      "learning_rate": 1.5214343337599885e-07,
      "loss": 1.4975,
      "step": 4534
    },
    {
      "epoch": 0.6839604856345676,
      "grad_norm": 0.24255544535163215,
      "learning_rate": 1.5205412861989713e-07,
      "loss": 1.558,
      "step": 4535
    },
    {
      "epoch": 0.684111303823241,
      "grad_norm": 0.4483495544033751,
      "learning_rate": 1.5196485146872406e-07,
      "loss": 1.6128,
      "step": 4536
    },
    {
      "epoch": 0.6842621220119146,
      "grad_norm": 0.30525871600786114,
      "learning_rate": 1.5187560194252496e-07,
      "loss": 1.5494,
      "step": 4537
    },
    {
      "epoch": 0.6844129402005882,
      "grad_norm": 0.23836452613561476,
      "learning_rate": 1.517863800613391e-07,
      "loss": 1.5943,
      "step": 4538
    },
    {
      "epoch": 0.6845637583892618,
      "grad_norm": 0.2501342378997237,
      "learning_rate": 1.516971858451993e-07,
      "loss": 1.5817,
      "step": 4539
    },
    {
      "epoch": 0.6847145765779353,
      "grad_norm": 0.3276380893394134,
      "learning_rate": 1.5160801931413215e-07,
      "loss": 1.5141,
      "step": 4540
    },
    {
      "epoch": 0.6848653947666089,
      "grad_norm": 0.2615657297936482,
      "learning_rate": 1.5151888048815833e-07,
      "loss": 1.5306,
      "step": 4541
    },
    {
      "epoch": 0.6850162129552824,
      "grad_norm": 0.26231711553037756,
      "learning_rate": 1.5142976938729198e-07,
      "loss": 1.602,
      "step": 4542
    },
    {
      "epoch": 0.6851670311439559,
      "grad_norm": 0.28429576765124964,
      "learning_rate": 1.5134068603154113e-07,
      "loss": 1.5316,
      "step": 4543
    },
    {
      "epoch": 0.6853178493326295,
      "grad_norm": 0.31239091894016446,
      "learning_rate": 1.5125163044090767e-07,
      "loss": 1.5024,
      "step": 4544
    },
    {
      "epoch": 0.6854686675213031,
      "grad_norm": 0.24401945059379362,
      "learning_rate": 1.5116260263538713e-07,
      "loss": 1.5675,
      "step": 4545
    },
    {
      "epoch": 0.6856194857099767,
      "grad_norm": 0.27257654534470044,
      "learning_rate": 1.5107360263496878e-07,
      "loss": 1.6108,
      "step": 4546
    },
    {
      "epoch": 0.6857703038986501,
      "grad_norm": 0.2516205155581599,
      "learning_rate": 1.5098463045963584e-07,
      "loss": 1.5333,
      "step": 4547
    },
    {
      "epoch": 0.6859211220873237,
      "grad_norm": 0.23781179235887603,
      "learning_rate": 1.508956861293651e-07,
      "loss": 1.5339,
      "step": 4548
    },
    {
      "epoch": 0.6860719402759973,
      "grad_norm": 0.2520970638415164,
      "learning_rate": 1.508067696641271e-07,
      "loss": 1.5188,
      "step": 4549
    },
    {
      "epoch": 0.6862227584646708,
      "grad_norm": 0.24736223711973349,
      "learning_rate": 1.507178810838862e-07,
      "loss": 1.5573,
      "step": 4550
    },
    {
      "epoch": 0.6863735766533444,
      "grad_norm": 0.2552180571114534,
      "learning_rate": 1.5062902040860053e-07,
      "loss": 1.5815,
      "step": 4551
    },
    {
      "epoch": 0.686524394842018,
      "grad_norm": 0.25558268884153396,
      "learning_rate": 1.5054018765822185e-07,
      "loss": 1.5228,
      "step": 4552
    },
    {
      "epoch": 0.6866752130306915,
      "grad_norm": 0.26345469434362906,
      "learning_rate": 1.5045138285269566e-07,
      "loss": 1.5266,
      "step": 4553
    },
    {
      "epoch": 0.686826031219365,
      "grad_norm": 0.2730774921258875,
      "learning_rate": 1.5036260601196136e-07,
      "loss": 1.5572,
      "step": 4554
    },
    {
      "epoch": 0.6869768494080386,
      "grad_norm": 0.2652309412430363,
      "learning_rate": 1.5027385715595187e-07,
      "loss": 1.5395,
      "step": 4555
    },
    {
      "epoch": 0.6871276675967122,
      "grad_norm": 0.508357645050542,
      "learning_rate": 1.5018513630459372e-07,
      "loss": 1.5837,
      "step": 4556
    },
    {
      "epoch": 0.6872784857853858,
      "grad_norm": 0.2385243692438645,
      "learning_rate": 1.5009644347780765e-07,
      "loss": 1.5148,
      "step": 4557
    },
    {
      "epoch": 0.6874293039740593,
      "grad_norm": 0.25014088917331895,
      "learning_rate": 1.5000777869550753e-07,
      "loss": 1.5505,
      "step": 4558
    },
    {
      "epoch": 0.6875801221627328,
      "grad_norm": 0.3648206648953276,
      "learning_rate": 1.4991914197760124e-07,
      "loss": 1.5474,
      "step": 4559
    },
    {
      "epoch": 0.6877309403514064,
      "grad_norm": 0.24635446185295545,
      "learning_rate": 1.498305333439904e-07,
      "loss": 1.5118,
      "step": 4560
    },
    {
      "epoch": 0.6878817585400799,
      "grad_norm": 0.24884860378935486,
      "learning_rate": 1.4974195281457017e-07,
      "loss": 1.5844,
      "step": 4561
    },
    {
      "epoch": 0.6880325767287535,
      "grad_norm": 0.25932085760973167,
      "learning_rate": 1.4965340040922943e-07,
      "loss": 1.5296,
      "step": 4562
    },
    {
      "epoch": 0.6881833949174271,
      "grad_norm": 0.24962208314785847,
      "learning_rate": 1.4956487614785074e-07,
      "loss": 1.6032,
      "step": 4563
    },
    {
      "epoch": 0.6883342131061005,
      "grad_norm": 0.3078533657792895,
      "learning_rate": 1.4947638005031048e-07,
      "loss": 1.5403,
      "step": 4564
    },
    {
      "epoch": 0.6884850312947741,
      "grad_norm": 0.24532524033327652,
      "learning_rate": 1.4938791213647862e-07,
      "loss": 1.5853,
      "step": 4565
    },
    {
      "epoch": 0.6886358494834477,
      "grad_norm": 0.25105566983553357,
      "learning_rate": 1.4929947242621857e-07,
      "loss": 1.5751,
      "step": 4566
    },
    {
      "epoch": 0.6887866676721213,
      "grad_norm": 0.2600457386477789,
      "learning_rate": 1.4921106093938784e-07,
      "loss": 1.6206,
      "step": 4567
    },
    {
      "epoch": 0.6889374858607948,
      "grad_norm": 0.2549953407241225,
      "learning_rate": 1.4912267769583733e-07,
      "loss": 1.5523,
      "step": 4568
    },
    {
      "epoch": 0.6890883040494684,
      "grad_norm": 0.2686116638084607,
      "learning_rate": 1.4903432271541156e-07,
      "loss": 1.533,
      "step": 4569
    },
    {
      "epoch": 0.6892391222381419,
      "grad_norm": 0.25694000871757877,
      "learning_rate": 1.4894599601794895e-07,
      "loss": 1.558,
      "step": 4570
    },
    {
      "epoch": 0.6893899404268155,
      "grad_norm": 0.25707164553738177,
      "learning_rate": 1.488576976232813e-07,
      "loss": 1.5897,
      "step": 4571
    },
    {
      "epoch": 0.689540758615489,
      "grad_norm": 0.25449614050111674,
      "learning_rate": 1.4876942755123413e-07,
      "loss": 1.6029,
      "step": 4572
    },
    {
      "epoch": 0.6896915768041626,
      "grad_norm": 0.25345246247083175,
      "learning_rate": 1.4868118582162682e-07,
      "loss": 1.6331,
      "step": 4573
    },
    {
      "epoch": 0.6898423949928362,
      "grad_norm": 0.2548159998153135,
      "learning_rate": 1.4859297245427207e-07,
      "loss": 1.5195,
      "step": 4574
    },
    {
      "epoch": 0.6899932131815096,
      "grad_norm": 0.2506490687717234,
      "learning_rate": 1.4850478746897637e-07,
      "loss": 1.5914,
      "step": 4575
    },
    {
      "epoch": 0.6901440313701832,
      "grad_norm": 0.30149312313645693,
      "learning_rate": 1.4841663088553979e-07,
      "loss": 1.557,
      "step": 4576
    },
    {
      "epoch": 0.6902948495588568,
      "grad_norm": 0.28041762145112953,
      "learning_rate": 1.483285027237562e-07,
      "loss": 1.5974,
      "step": 4577
    },
    {
      "epoch": 0.6904456677475304,
      "grad_norm": 0.2530644352677591,
      "learning_rate": 1.482404030034128e-07,
      "loss": 1.4708,
      "step": 4578
    },
    {
      "epoch": 0.6905964859362039,
      "grad_norm": 0.26767673143594023,
      "learning_rate": 1.4815233174429047e-07,
      "loss": 1.5615,
      "step": 4579
    },
    {
      "epoch": 0.6907473041248775,
      "grad_norm": 0.27807131741426094,
      "learning_rate": 1.4806428896616396e-07,
      "loss": 1.6307,
      "step": 4580
    },
    {
      "epoch": 0.690898122313551,
      "grad_norm": 0.2559874783430606,
      "learning_rate": 1.4797627468880136e-07,
      "loss": 1.5253,
      "step": 4581
    },
    {
      "epoch": 0.6910489405022245,
      "grad_norm": 0.36174431325638784,
      "learning_rate": 1.4788828893196438e-07,
      "loss": 1.5906,
      "step": 4582
    },
    {
      "epoch": 0.6911997586908981,
      "grad_norm": 0.26659067323633806,
      "learning_rate": 1.4780033171540844e-07,
      "loss": 1.6182,
      "step": 4583
    },
    {
      "epoch": 0.6913505768795717,
      "grad_norm": 0.26001690653683646,
      "learning_rate": 1.477124030588826e-07,
      "loss": 1.5281,
      "step": 4584
    },
    {
      "epoch": 0.6915013950682453,
      "grad_norm": 0.2669507511015986,
      "learning_rate": 1.4762450298212924e-07,
      "loss": 1.6036,
      "step": 4585
    },
    {
      "epoch": 0.6916522132569188,
      "grad_norm": 0.27047938435631036,
      "learning_rate": 1.4753663150488443e-07,
      "loss": 1.5513,
      "step": 4586
    },
    {
      "epoch": 0.6918030314455923,
      "grad_norm": 0.24458341620251461,
      "learning_rate": 1.4744878864687815e-07,
      "loss": 1.5949,
      "step": 4587
    },
    {
      "epoch": 0.6919538496342659,
      "grad_norm": 0.332652850278529,
      "learning_rate": 1.4736097442783346e-07,
      "loss": 1.5386,
      "step": 4588
    },
    {
      "epoch": 0.6921046678229394,
      "grad_norm": 0.25264908157249905,
      "learning_rate": 1.4727318886746724e-07,
      "loss": 1.5539,
      "step": 4589
    },
    {
      "epoch": 0.692255486011613,
      "grad_norm": 0.24265317520106885,
      "learning_rate": 1.4718543198548996e-07,
      "loss": 1.5467,
      "step": 4590
    },
    {
      "epoch": 0.6924063042002866,
      "grad_norm": 0.23714676531560258,
      "learning_rate": 1.470977038016056e-07,
      "loss": 1.5504,
      "step": 4591
    },
    {
      "epoch": 0.6925571223889601,
      "grad_norm": 0.2573756180396614,
      "learning_rate": 1.4701000433551158e-07,
      "loss": 1.6083,
      "step": 4592
    },
    {
      "epoch": 0.6927079405776336,
      "grad_norm": 0.24096449503649983,
      "learning_rate": 1.4692233360689915e-07,
      "loss": 1.615,
      "step": 4593
    },
    {
      "epoch": 0.6928587587663072,
      "grad_norm": 0.24723077453611716,
      "learning_rate": 1.4683469163545287e-07,
      "loss": 1.5822,
      "step": 4594
    },
    {
      "epoch": 0.6930095769549808,
      "grad_norm": 0.2565972275948653,
      "learning_rate": 1.4674707844085082e-07,
      "loss": 1.5817,
      "step": 4595
    },
    {
      "epoch": 0.6931603951436544,
      "grad_norm": 0.25394017061617835,
      "learning_rate": 1.4665949404276485e-07,
      "loss": 1.6008,
      "step": 4596
    },
    {
      "epoch": 0.6933112133323279,
      "grad_norm": 0.24890272248525033,
      "learning_rate": 1.4657193846086018e-07,
      "loss": 1.5577,
      "step": 4597
    },
    {
      "epoch": 0.6934620315210014,
      "grad_norm": 0.25471972550028754,
      "learning_rate": 1.4648441171479558e-07,
      "loss": 1.487,
      "step": 4598
    },
    {
      "epoch": 0.693612849709675,
      "grad_norm": 0.24686979043513557,
      "learning_rate": 1.4639691382422324e-07,
      "loss": 1.5725,
      "step": 4599
    },
    {
      "epoch": 0.6937636678983485,
      "grad_norm": 0.3777801149755808,
      "learning_rate": 1.4630944480878916e-07,
      "loss": 1.5033,
      "step": 4600
    },
    {
      "epoch": 0.6939144860870221,
      "grad_norm": 0.24966962480364918,
      "learning_rate": 1.4622200468813257e-07,
      "loss": 1.5242,
      "step": 4601
    },
    {
      "epoch": 0.6940653042756957,
      "grad_norm": 0.24990942126298826,
      "learning_rate": 1.4613459348188635e-07,
      "loss": 1.601,
      "step": 4602
    },
    {
      "epoch": 0.6942161224643693,
      "grad_norm": 0.28309591473883483,
      "learning_rate": 1.460472112096769e-07,
      "loss": 1.5873,
      "step": 4603
    },
    {
      "epoch": 0.6943669406530427,
      "grad_norm": 0.240701977509555,
      "learning_rate": 1.4595985789112408e-07,
      "loss": 1.5156,
      "step": 4604
    },
    {
      "epoch": 0.6945177588417163,
      "grad_norm": 0.25553715074585126,
      "learning_rate": 1.458725335458411e-07,
      "loss": 1.5409,
      "step": 4605
    },
    {
      "epoch": 0.6946685770303899,
      "grad_norm": 0.26737231559406555,
      "learning_rate": 1.4578523819343508e-07,
      "loss": 1.4752,
      "step": 4606
    },
    {
      "epoch": 0.6948193952190634,
      "grad_norm": 0.24931512224638422,
      "learning_rate": 1.456979718535062e-07,
      "loss": 1.5891,
      "step": 4607
    },
    {
      "epoch": 0.694970213407737,
      "grad_norm": 0.24713254615524075,
      "learning_rate": 1.4561073454564826e-07,
      "loss": 1.5321,
      "step": 4608
    },
    {
      "epoch": 0.6951210315964105,
      "grad_norm": 0.24274431420453124,
      "learning_rate": 1.4552352628944876e-07,
      "loss": 1.5375,
      "step": 4609
    },
    {
      "epoch": 0.695271849785084,
      "grad_norm": 0.24327648509344132,
      "learning_rate": 1.454363471044883e-07,
      "loss": 1.5397,
      "step": 4610
    },
    {
      "epoch": 0.6954226679737576,
      "grad_norm": 0.42859160895656556,
      "learning_rate": 1.4534919701034128e-07,
      "loss": 1.6108,
      "step": 4611
    },
    {
      "epoch": 0.6955734861624312,
      "grad_norm": 0.2491065048758073,
      "learning_rate": 1.4526207602657532e-07,
      "loss": 1.5006,
      "step": 4612
    },
    {
      "epoch": 0.6957243043511048,
      "grad_norm": 0.2692117787960512,
      "learning_rate": 1.451749841727517e-07,
      "loss": 1.6145,
      "step": 4613
    },
    {
      "epoch": 0.6958751225397783,
      "grad_norm": 0.2536910675646695,
      "learning_rate": 1.4508792146842514e-07,
      "loss": 1.5425,
      "step": 4614
    },
    {
      "epoch": 0.6960259407284518,
      "grad_norm": 0.3287075484496632,
      "learning_rate": 1.4500088793314348e-07,
      "loss": 1.5928,
      "step": 4615
    },
    {
      "epoch": 0.6961767589171254,
      "grad_norm": 0.2407439602141349,
      "learning_rate": 1.4491388358644866e-07,
      "loss": 1.5594,
      "step": 4616
    },
    {
      "epoch": 0.696327577105799,
      "grad_norm": 0.26256589661588803,
      "learning_rate": 1.4482690844787543e-07,
      "loss": 1.5984,
      "step": 4617
    },
    {
      "epoch": 0.6964783952944725,
      "grad_norm": 0.32159675502824847,
      "learning_rate": 1.4473996253695224e-07,
      "loss": 1.5943,
      "step": 4618
    },
    {
      "epoch": 0.6966292134831461,
      "grad_norm": 0.24326918175647477,
      "learning_rate": 1.4465304587320115e-07,
      "loss": 1.5593,
      "step": 4619
    },
    {
      "epoch": 0.6967800316718196,
      "grad_norm": 0.2537403703578292,
      "learning_rate": 1.4456615847613738e-07,
      "loss": 1.5685,
      "step": 4620
    },
    {
      "epoch": 0.6969308498604931,
      "grad_norm": 0.3058001984394793,
      "learning_rate": 1.4447930036526963e-07,
      "loss": 1.5538,
      "step": 4621
    },
    {
      "epoch": 0.6970816680491667,
      "grad_norm": 0.26862945706499347,
      "learning_rate": 1.4439247156010014e-07,
      "loss": 1.6163,
      "step": 4622
    },
    {
      "epoch": 0.6972324862378403,
      "grad_norm": 0.2518795793224312,
      "learning_rate": 1.4430567208012458e-07,
      "loss": 1.6093,
      "step": 4623
    },
    {
      "epoch": 0.6973833044265139,
      "grad_norm": 0.24554849766932724,
      "learning_rate": 1.4421890194483184e-07,
      "loss": 1.5258,
      "step": 4624
    },
    {
      "epoch": 0.6975341226151874,
      "grad_norm": 0.2518564872582991,
      "learning_rate": 1.441321611737043e-07,
      "loss": 1.5754,
      "step": 4625
    },
    {
      "epoch": 0.6976849408038609,
      "grad_norm": 0.25178904071540303,
      "learning_rate": 1.4404544978621797e-07,
      "loss": 1.5468,
      "step": 4626
    },
    {
      "epoch": 0.6978357589925345,
      "grad_norm": 0.26285735665220644,
      "learning_rate": 1.4395876780184194e-07,
      "loss": 1.5592,
      "step": 4627
    },
    {
      "epoch": 0.697986577181208,
      "grad_norm": 0.24616375825149645,
      "learning_rate": 1.438721152400388e-07,
      "loss": 1.558,
      "step": 4628
    },
    {
      "epoch": 0.6981373953698816,
      "grad_norm": 0.27139617239702973,
      "learning_rate": 1.4378549212026475e-07,
      "loss": 1.5659,
      "step": 4629
    },
    {
      "epoch": 0.6982882135585552,
      "grad_norm": 0.26339115341264374,
      "learning_rate": 1.4369889846196914e-07,
      "loss": 1.617,
      "step": 4630
    },
    {
      "epoch": 0.6984390317472288,
      "grad_norm": 0.2499043349546843,
      "learning_rate": 1.4361233428459462e-07,
      "loss": 1.5837,
      "step": 4631
    },
    {
      "epoch": 0.6985898499359022,
      "grad_norm": 0.2678042961990252,
      "learning_rate": 1.4352579960757755e-07,
      "loss": 1.5559,
      "step": 4632
    },
    {
      "epoch": 0.6987406681245758,
      "grad_norm": 0.2550113462690431,
      "learning_rate": 1.4343929445034746e-07,
      "loss": 1.5235,
      "step": 4633
    },
    {
      "epoch": 0.6988914863132494,
      "grad_norm": 0.27815081080646575,
      "learning_rate": 1.4335281883232713e-07,
      "loss": 1.522,
      "step": 4634
    },
    {
      "epoch": 0.699042304501923,
      "grad_norm": 0.25392747519123826,
      "learning_rate": 1.4326637277293308e-07,
      "loss": 1.55,
      "step": 4635
    },
    {
      "epoch": 0.6991931226905965,
      "grad_norm": 0.24728466238713195,
      "learning_rate": 1.431799562915748e-07,
      "loss": 1.6582,
      "step": 4636
    },
    {
      "epoch": 0.69934394087927,
      "grad_norm": 0.23783857063910752,
      "learning_rate": 1.4309356940765538e-07,
      "loss": 1.523,
      "step": 4637
    },
    {
      "epoch": 0.6994947590679436,
      "grad_norm": 0.27091284652224773,
      "learning_rate": 1.430072121405711e-07,
      "loss": 1.5445,
      "step": 4638
    },
    {
      "epoch": 0.6996455772566171,
      "grad_norm": 0.2635842887417288,
      "learning_rate": 1.4292088450971178e-07,
      "loss": 1.5408,
      "step": 4639
    },
    {
      "epoch": 0.6997963954452907,
      "grad_norm": 0.25774520157377045,
      "learning_rate": 1.428345865344605e-07,
      "loss": 1.5434,
      "step": 4640
    },
    {
      "epoch": 0.6999472136339643,
      "grad_norm": 0.29453802949194896,
      "learning_rate": 1.427483182341936e-07,
      "loss": 1.5267,
      "step": 4641
    },
    {
      "epoch": 0.7000980318226379,
      "grad_norm": 0.24766186104982402,
      "learning_rate": 1.4266207962828082e-07,
      "loss": 1.5256,
      "step": 4642
    },
    {
      "epoch": 0.7002488500113113,
      "grad_norm": 0.26043474078314627,
      "learning_rate": 1.4257587073608531e-07,
      "loss": 1.6133,
      "step": 4643
    },
    {
      "epoch": 0.7003996681999849,
      "grad_norm": 0.2755242564839165,
      "learning_rate": 1.424896915769634e-07,
      "loss": 1.4794,
      "step": 4644
    },
    {
      "epoch": 0.7005504863886585,
      "grad_norm": 0.25911969816313285,
      "learning_rate": 1.4240354217026491e-07,
      "loss": 1.5828,
      "step": 4645
    },
    {
      "epoch": 0.700701304577332,
      "grad_norm": 0.25111418115042655,
      "learning_rate": 1.423174225353328e-07,
      "loss": 1.5815,
      "step": 4646
    },
    {
      "epoch": 0.7008521227660056,
      "grad_norm": 0.2484232812334775,
      "learning_rate": 1.4223133269150344e-07,
      "loss": 1.6199,
      "step": 4647
    },
    {
      "epoch": 0.7010029409546792,
      "grad_norm": 0.2466588001579759,
      "learning_rate": 1.421452726581066e-07,
      "loss": 1.6041,
      "step": 4648
    },
    {
      "epoch": 0.7011537591433527,
      "grad_norm": 0.2512178840310022,
      "learning_rate": 1.4205924245446522e-07,
      "loss": 1.5315,
      "step": 4649
    },
    {
      "epoch": 0.7013045773320262,
      "grad_norm": 0.25511669845907525,
      "learning_rate": 1.4197324209989557e-07,
      "loss": 1.5644,
      "step": 4650
    },
    {
      "epoch": 0.7014553955206998,
      "grad_norm": 0.2580678072784764,
      "learning_rate": 1.4188727161370717e-07,
      "loss": 1.5675,
      "step": 4651
    },
    {
      "epoch": 0.7016062137093734,
      "grad_norm": 0.24206855496981436,
      "learning_rate": 1.4180133101520302e-07,
      "loss": 1.5505,
      "step": 4652
    },
    {
      "epoch": 0.701757031898047,
      "grad_norm": 0.255627073851271,
      "learning_rate": 1.4171542032367922e-07,
      "loss": 1.5633,
      "step": 4653
    },
    {
      "epoch": 0.7019078500867204,
      "grad_norm": 0.24664312577496256,
      "learning_rate": 1.4162953955842516e-07,
      "loss": 1.5686,
      "step": 4654
    },
    {
      "epoch": 0.702058668275394,
      "grad_norm": 0.24303956624318634,
      "learning_rate": 1.4154368873872374e-07,
      "loss": 1.5745,
      "step": 4655
    },
    {
      "epoch": 0.7022094864640676,
      "grad_norm": 0.24684430634144772,
      "learning_rate": 1.4145786788385086e-07,
      "loss": 1.5441,
      "step": 4656
    },
    {
      "epoch": 0.7023603046527411,
      "grad_norm": 0.2917308163037401,
      "learning_rate": 1.4137207701307577e-07,
      "loss": 1.6054,
      "step": 4657
    },
    {
      "epoch": 0.7025111228414147,
      "grad_norm": 0.25098386031111397,
      "learning_rate": 1.412863161456611e-07,
      "loss": 1.5738,
      "step": 4658
    },
    {
      "epoch": 0.7026619410300883,
      "grad_norm": 0.2529208126338307,
      "learning_rate": 1.4120058530086269e-07,
      "loss": 1.5741,
      "step": 4659
    },
    {
      "epoch": 0.7028127592187617,
      "grad_norm": 0.24101650356689916,
      "learning_rate": 1.411148844979294e-07,
      "loss": 1.5409,
      "step": 4660
    },
    {
      "epoch": 0.7029635774074353,
      "grad_norm": 0.238650413776694,
      "learning_rate": 1.4102921375610382e-07,
      "loss": 1.5352,
      "step": 4661
    },
    {
      "epoch": 0.7031143955961089,
      "grad_norm": 0.2689066687683917,
      "learning_rate": 1.4094357309462134e-07,
      "loss": 1.5955,
      "step": 4662
    },
    {
      "epoch": 0.7032652137847825,
      "grad_norm": 0.2706659048411641,
      "learning_rate": 1.4085796253271093e-07,
      "loss": 1.562,
      "step": 4663
    },
    {
      "epoch": 0.703416031973456,
      "grad_norm": 0.23829039855906392,
      "learning_rate": 1.4077238208959445e-07,
      "loss": 1.5374,
      "step": 4664
    },
    {
      "epoch": 0.7035668501621295,
      "grad_norm": 0.26114976173614274,
      "learning_rate": 1.4068683178448747e-07,
      "loss": 1.5966,
      "step": 4665
    },
    {
      "epoch": 0.7037176683508031,
      "grad_norm": 0.25768827402905975,
      "learning_rate": 1.4060131163659838e-07,
      "loss": 1.5849,
      "step": 4666
    },
    {
      "epoch": 0.7038684865394766,
      "grad_norm": 0.4189391320885793,
      "learning_rate": 1.4051582166512894e-07,
      "loss": 1.5402,
      "step": 4667
    },
    {
      "epoch": 0.7040193047281502,
      "grad_norm": 0.2483133632121748,
      "learning_rate": 1.4043036188927406e-07,
      "loss": 1.6295,
      "step": 4668
    },
    {
      "epoch": 0.7041701229168238,
      "grad_norm": 0.2385855023535277,
      "learning_rate": 1.4034493232822214e-07,
      "loss": 1.4939,
      "step": 4669
    },
    {
      "epoch": 0.7043209411054974,
      "grad_norm": 0.26169111052317984,
      "learning_rate": 1.4025953300115452e-07,
      "loss": 1.5273,
      "step": 4670
    },
    {
      "epoch": 0.7044717592941708,
      "grad_norm": 0.2585477916676268,
      "learning_rate": 1.4017416392724567e-07,
      "loss": 1.5145,
      "step": 4671
    },
    {
      "epoch": 0.7046225774828444,
      "grad_norm": 0.26120948412477474,
      "learning_rate": 1.400888251256637e-07,
      "loss": 1.5265,
      "step": 4672
    },
    {
      "epoch": 0.704773395671518,
      "grad_norm": 0.24901225629669874,
      "learning_rate": 1.4000351661556953e-07,
      "loss": 1.6174,
      "step": 4673
    },
    {
      "epoch": 0.7049242138601916,
      "grad_norm": 0.24153572524708944,
      "learning_rate": 1.3991823841611734e-07,
      "loss": 1.5245,
      "step": 4674
    },
    {
      "epoch": 0.7050750320488651,
      "grad_norm": 0.5307711694492537,
      "learning_rate": 1.398329905464547e-07,
      "loss": 1.5671,
      "step": 4675
    },
    {
      "epoch": 0.7052258502375387,
      "grad_norm": 0.2579480575203586,
      "learning_rate": 1.397477730257222e-07,
      "loss": 1.5647,
      "step": 4676
    },
    {
      "epoch": 0.7053766684262122,
      "grad_norm": 0.2639878244090322,
      "learning_rate": 1.3966258587305356e-07,
      "loss": 1.5617,
      "step": 4677
    },
    {
      "epoch": 0.7055274866148857,
      "grad_norm": 0.24542207545138842,
      "learning_rate": 1.395774291075759e-07,
      "loss": 1.5895,
      "step": 4678
    },
    {
      "epoch": 0.7056783048035593,
      "grad_norm": 0.2591097894616853,
      "learning_rate": 1.394923027484094e-07,
      "loss": 1.5643,
      "step": 4679
    },
    {
      "epoch": 0.7058291229922329,
      "grad_norm": 0.24816792443713048,
      "learning_rate": 1.3940720681466732e-07,
      "loss": 1.6382,
      "step": 4680
    },
    {
      "epoch": 0.7059799411809065,
      "grad_norm": 0.24660088521437482,
      "learning_rate": 1.3932214132545617e-07,
      "loss": 1.528,
      "step": 4681
    },
    {
      "epoch": 0.7061307593695799,
      "grad_norm": 0.2550707894427213,
      "learning_rate": 1.392371062998757e-07,
      "loss": 1.5442,
      "step": 4682
    },
    {
      "epoch": 0.7062815775582535,
      "grad_norm": 0.25198459987988236,
      "learning_rate": 1.3915210175701875e-07,
      "loss": 1.563,
      "step": 4683
    },
    {
      "epoch": 0.7064323957469271,
      "grad_norm": 0.2737898364630557,
      "learning_rate": 1.390671277159712e-07,
      "loss": 1.5757,
      "step": 4684
    },
    {
      "epoch": 0.7065832139356006,
      "grad_norm": 0.25991857160488113,
      "learning_rate": 1.389821841958123e-07,
      "loss": 1.525,
      "step": 4685
    },
    {
      "epoch": 0.7067340321242742,
      "grad_norm": 0.26009297654677155,
      "learning_rate": 1.3889727121561436e-07,
      "loss": 1.5795,
      "step": 4686
    },
    {
      "epoch": 0.7068848503129478,
      "grad_norm": 0.2547506888702494,
      "learning_rate": 1.3881238879444273e-07,
      "loss": 1.5742,
      "step": 4687
    },
    {
      "epoch": 0.7070356685016213,
      "grad_norm": 0.24743037937256265,
      "learning_rate": 1.3872753695135607e-07,
      "loss": 1.5968,
      "step": 4688
    },
    {
      "epoch": 0.7071864866902948,
      "grad_norm": 0.24740696964137768,
      "learning_rate": 1.386427157054061e-07,
      "loss": 1.4845,
      "step": 4689
    },
    {
      "epoch": 0.7073373048789684,
      "grad_norm": 0.24466560698641626,
      "learning_rate": 1.385579250756375e-07,
      "loss": 1.5184,
      "step": 4690
    },
    {
      "epoch": 0.707488123067642,
      "grad_norm": 0.24718704325105698,
      "learning_rate": 1.3847316508108846e-07,
      "loss": 1.5223,
      "step": 4691
    },
    {
      "epoch": 0.7076389412563155,
      "grad_norm": 0.32800916216340986,
      "learning_rate": 1.3838843574078997e-07,
      "loss": 1.5204,
      "step": 4692
    },
    {
      "epoch": 0.7077897594449891,
      "grad_norm": 0.2868302931470831,
      "learning_rate": 1.383037370737662e-07,
      "loss": 1.5735,
      "step": 4693
    },
    {
      "epoch": 0.7079405776336626,
      "grad_norm": 0.2463838919775735,
      "learning_rate": 1.3821906909903444e-07,
      "loss": 1.5848,
      "step": 4694
    },
    {
      "epoch": 0.7080913958223362,
      "grad_norm": 0.24133956966287837,
      "learning_rate": 1.3813443183560524e-07,
      "loss": 1.6289,
      "step": 4695
    },
    {
      "epoch": 0.7082422140110097,
      "grad_norm": 0.24378465938684218,
      "learning_rate": 1.3804982530248205e-07,
      "loss": 1.4916,
      "step": 4696
    },
    {
      "epoch": 0.7083930321996833,
      "grad_norm": 0.24630088740305392,
      "learning_rate": 1.3796524951866142e-07,
      "loss": 1.5481,
      "step": 4697
    },
    {
      "epoch": 0.7085438503883569,
      "grad_norm": 0.2656040531784327,
      "learning_rate": 1.3788070450313328e-07,
      "loss": 1.5553,
      "step": 4698
    },
    {
      "epoch": 0.7086946685770303,
      "grad_norm": 0.2511559138227458,
      "learning_rate": 1.3779619027488027e-07,
      "loss": 1.5458,
      "step": 4699
    },
    {
      "epoch": 0.7088454867657039,
      "grad_norm": 0.25184229086367793,
      "learning_rate": 1.3771170685287835e-07,
      "loss": 1.5723,
      "step": 4700
    },
    {
      "epoch": 0.7089963049543775,
      "grad_norm": 0.2576597364739768,
      "learning_rate": 1.3762725425609656e-07,
      "loss": 1.441,
      "step": 4701
    },
    {
      "epoch": 0.7091471231430511,
      "grad_norm": 0.2554761744844474,
      "learning_rate": 1.375428325034969e-07,
      "loss": 1.5775,
      "step": 4702
    },
    {
      "epoch": 0.7092979413317246,
      "grad_norm": 0.24178754208210249,
      "learning_rate": 1.3745844161403454e-07,
      "loss": 1.5885,
      "step": 4703
    },
    {
      "epoch": 0.7094487595203982,
      "grad_norm": 0.2518623994941124,
      "learning_rate": 1.373740816066577e-07,
      "loss": 1.5964,
      "step": 4704
    },
    {
      "epoch": 0.7095995777090717,
      "grad_norm": 0.2502385809004188,
      "learning_rate": 1.3728975250030767e-07,
      "loss": 1.5666,
      "step": 4705
    },
    {
      "epoch": 0.7097503958977452,
      "grad_norm": 0.2431014721828363,
      "learning_rate": 1.3720545431391877e-07,
      "loss": 1.486,
      "step": 4706
    },
    {
      "epoch": 0.7099012140864188,
      "grad_norm": 0.3343324071561751,
      "learning_rate": 1.3712118706641833e-07,
      "loss": 1.5682,
      "step": 4707
    },
    {
      "epoch": 0.7100520322750924,
      "grad_norm": 0.24824475114384603,
      "learning_rate": 1.3703695077672695e-07,
      "loss": 1.6006,
      "step": 4708
    },
    {
      "epoch": 0.710202850463766,
      "grad_norm": 0.27672391262975443,
      "learning_rate": 1.36952745463758e-07,
      "loss": 1.5638,
      "step": 4709
    },
    {
      "epoch": 0.7103536686524394,
      "grad_norm": 0.25077329250405694,
      "learning_rate": 1.3686857114641805e-07,
      "loss": 1.5158,
      "step": 4710
    },
    {
      "epoch": 0.710504486841113,
      "grad_norm": 0.3031627755783418,
      "learning_rate": 1.3678442784360678e-07,
      "loss": 1.6167,
      "step": 4711
    },
    {
      "epoch": 0.7106553050297866,
      "grad_norm": 0.24626767774868308,
      "learning_rate": 1.3670031557421674e-07,
      "loss": 1.5919,
      "step": 4712
    },
    {
      "epoch": 0.7108061232184602,
      "grad_norm": 0.24588592958535221,
      "learning_rate": 1.3661623435713348e-07,
      "loss": 1.5194,
      "step": 4713
    },
    {
      "epoch": 0.7109569414071337,
      "grad_norm": 0.2478471317333664,
      "learning_rate": 1.3653218421123588e-07,
      "loss": 1.5131,
      "step": 4714
    },
    {
      "epoch": 0.7111077595958073,
      "grad_norm": 0.275953448782174,
      "learning_rate": 1.3644816515539552e-07,
      "loss": 1.5677,
      "step": 4715
    },
    {
      "epoch": 0.7112585777844808,
      "grad_norm": 0.25129579415503,
      "learning_rate": 1.3636417720847722e-07,
      "loss": 1.5241,
      "step": 4716
    },
    {
      "epoch": 0.7114093959731543,
      "grad_norm": 0.25414855356921834,
      "learning_rate": 1.3628022038933857e-07,
      "loss": 1.5067,
      "step": 4717
    },
    {
      "epoch": 0.7115602141618279,
      "grad_norm": 0.24728432666433134,
      "learning_rate": 1.3619629471683048e-07,
      "loss": 1.6078,
      "step": 4718
    },
    {
      "epoch": 0.7117110323505015,
      "grad_norm": 0.2449076505767851,
      "learning_rate": 1.361124002097966e-07,
      "loss": 1.5409,
      "step": 4719
    },
    {
      "epoch": 0.7118618505391751,
      "grad_norm": 0.24870336121142433,
      "learning_rate": 1.360285368870737e-07,
      "loss": 1.6059,
      "step": 4720
    },
    {
      "epoch": 0.7120126687278486,
      "grad_norm": 0.2504405644553221,
      "learning_rate": 1.3594470476749158e-07,
      "loss": 1.6057,
      "step": 4721
    },
    {
      "epoch": 0.7121634869165221,
      "grad_norm": 0.27633694696709254,
      "learning_rate": 1.3586090386987305e-07,
      "loss": 1.5701,
      "step": 4722
    },
    {
      "epoch": 0.7123143051051957,
      "grad_norm": 0.2649768585335686,
      "learning_rate": 1.3577713421303365e-07,
      "loss": 1.5804,
      "step": 4723
    },
    {
      "epoch": 0.7124651232938692,
      "grad_norm": 0.26194802890270397,
      "learning_rate": 1.3569339581578238e-07,
      "loss": 1.5762,
      "step": 4724
    },
    {
      "epoch": 0.7126159414825428,
      "grad_norm": 0.2464197300271867,
      "learning_rate": 1.3560968869692075e-07,
      "loss": 1.5432,
      "step": 4725
    },
    {
      "epoch": 0.7127667596712164,
      "grad_norm": 1.1086502037535637,
      "learning_rate": 1.355260128752434e-07,
      "loss": 1.6048,
      "step": 4726
    },
    {
      "epoch": 0.7129175778598899,
      "grad_norm": 0.30833236238085443,
      "learning_rate": 1.354423683695382e-07,
      "loss": 1.5721,
      "step": 4727
    },
    {
      "epoch": 0.7130683960485634,
      "grad_norm": 0.3791106109638253,
      "learning_rate": 1.3535875519858568e-07,
      "loss": 1.5757,
      "step": 4728
    },
    {
      "epoch": 0.713219214237237,
      "grad_norm": 0.260077268877803,
      "learning_rate": 1.3527517338115942e-07,
      "loss": 1.5573,
      "step": 4729
    },
    {
      "epoch": 0.7133700324259106,
      "grad_norm": 0.24628343108179016,
      "learning_rate": 1.3519162293602587e-07,
      "loss": 1.5816,
      "step": 4730
    },
    {
      "epoch": 0.7135208506145841,
      "grad_norm": 0.24016223821112434,
      "learning_rate": 1.3510810388194476e-07,
      "loss": 1.4971,
      "step": 4731
    },
    {
      "epoch": 0.7136716688032577,
      "grad_norm": 0.2519599764404375,
      "learning_rate": 1.3502461623766838e-07,
      "loss": 1.5589,
      "step": 4732
    },
    {
      "epoch": 0.7138224869919312,
      "grad_norm": 0.2675908029027234,
      "learning_rate": 1.3494116002194217e-07,
      "loss": 1.5877,
      "step": 4733
    },
    {
      "epoch": 0.7139733051806048,
      "grad_norm": 0.310853199221443,
      "learning_rate": 1.3485773525350451e-07,
      "loss": 1.5517,
      "step": 4734
    },
    {
      "epoch": 0.7141241233692783,
      "grad_norm": 0.25426694369992525,
      "learning_rate": 1.3477434195108672e-07,
      "loss": 1.581,
      "step": 4735
    },
    {
      "epoch": 0.7142749415579519,
      "grad_norm": 0.24852694461858268,
      "learning_rate": 1.346909801334129e-07,
      "loss": 1.4889,
      "step": 4736
    },
    {
      "epoch": 0.7144257597466255,
      "grad_norm": 0.26412638082902884,
      "learning_rate": 1.346076498192004e-07,
      "loss": 1.5109,
      "step": 4737
    },
    {
      "epoch": 0.714576577935299,
      "grad_norm": 0.3346778331399256,
      "learning_rate": 1.3452435102715918e-07,
      "loss": 1.6408,
      "step": 4738
    },
    {
      "epoch": 0.7147273961239725,
      "grad_norm": 0.24639182747383537,
      "learning_rate": 1.3444108377599212e-07,
      "loss": 1.5529,
      "step": 4739
    },
    {
      "epoch": 0.7148782143126461,
      "grad_norm": 0.25655433508352976,
      "learning_rate": 1.3435784808439542e-07,
      "loss": 1.5435,
      "step": 4740
    },
    {
      "epoch": 0.7150290325013197,
      "grad_norm": 0.2717890791231324,
      "learning_rate": 1.3427464397105777e-07,
      "loss": 1.5816,
      "step": 4741
    },
    {
      "epoch": 0.7151798506899932,
      "grad_norm": 0.24936196683161244,
      "learning_rate": 1.3419147145466092e-07,
      "loss": 1.5409,
      "step": 4742
    },
    {
      "epoch": 0.7153306688786668,
      "grad_norm": 0.27944856658442013,
      "learning_rate": 1.3410833055387943e-07,
      "loss": 1.5387,
      "step": 4743
    },
    {
      "epoch": 0.7154814870673403,
      "grad_norm": 0.2568016870564878,
      "learning_rate": 1.3402522128738104e-07,
      "loss": 1.4872,
      "step": 4744
    },
    {
      "epoch": 0.7156323052560138,
      "grad_norm": 0.27254404256969117,
      "learning_rate": 1.339421436738261e-07,
      "loss": 1.6064,
      "step": 4745
    },
    {
      "epoch": 0.7157831234446874,
      "grad_norm": 0.274410169098975,
      "learning_rate": 1.3385909773186788e-07,
      "loss": 1.5203,
      "step": 4746
    },
    {
      "epoch": 0.715933941633361,
      "grad_norm": 0.2880226154380716,
      "learning_rate": 1.3377608348015278e-07,
      "loss": 1.5141,
      "step": 4747
    },
    {
      "epoch": 0.7160847598220346,
      "grad_norm": 0.2594462924354836,
      "learning_rate": 1.3369310093731983e-07,
      "loss": 1.4899,
      "step": 4748
    },
    {
      "epoch": 0.7162355780107081,
      "grad_norm": 0.30696420741370845,
      "learning_rate": 1.33610150122001e-07,
      "loss": 1.5421,
      "step": 4749
    },
    {
      "epoch": 0.7163863961993816,
      "grad_norm": 0.2618505672279436,
      "learning_rate": 1.335272310528212e-07,
      "loss": 1.538,
      "step": 4750
    },
    {
      "epoch": 0.7165372143880552,
      "grad_norm": 0.462287871540447,
      "learning_rate": 1.3344434374839822e-07,
      "loss": 1.5347,
      "step": 4751
    },
    {
      "epoch": 0.7166880325767288,
      "grad_norm": 0.24555954596563262,
      "learning_rate": 1.3336148822734253e-07,
      "loss": 1.4691,
      "step": 4752
    },
    {
      "epoch": 0.7168388507654023,
      "grad_norm": 0.35095253187412284,
      "learning_rate": 1.3327866450825782e-07,
      "loss": 1.5425,
      "step": 4753
    },
    {
      "epoch": 0.7169896689540759,
      "grad_norm": 0.2502740043949193,
      "learning_rate": 1.3319587260974025e-07,
      "loss": 1.5323,
      "step": 4754
    },
    {
      "epoch": 0.7171404871427494,
      "grad_norm": 0.2507176225991556,
      "learning_rate": 1.331131125503791e-07,
      "loss": 1.5308,
      "step": 4755
    },
    {
      "epoch": 0.7172913053314229,
      "grad_norm": 0.2819272418342701,
      "learning_rate": 1.330303843487563e-07,
      "loss": 1.5434,
      "step": 4756
    },
    {
      "epoch": 0.7174421235200965,
      "grad_norm": 0.2509324282898924,
      "learning_rate": 1.3294768802344692e-07,
      "loss": 1.5538,
      "step": 4757
    },
    {
      "epoch": 0.7175929417087701,
      "grad_norm": 0.24737899237106037,
      "learning_rate": 1.328650235930186e-07,
      "loss": 1.4796,
      "step": 4758
    },
    {
      "epoch": 0.7177437598974437,
      "grad_norm": 0.26206163566396135,
      "learning_rate": 1.327823910760318e-07,
      "loss": 1.516,
      "step": 4759
    },
    {
      "epoch": 0.7178945780861172,
      "grad_norm": 0.6149272723503363,
      "learning_rate": 1.3269979049104014e-07,
      "loss": 1.585,
      "step": 4760
    },
    {
      "epoch": 0.7180453962747907,
      "grad_norm": 0.25192326864076015,
      "learning_rate": 1.3261722185658974e-07,
      "loss": 1.5538,
      "step": 4761
    },
    {
      "epoch": 0.7181962144634643,
      "grad_norm": 0.24960088406697814,
      "learning_rate": 1.3253468519121963e-07,
      "loss": 1.5544,
      "step": 4762
    },
    {
      "epoch": 0.7183470326521378,
      "grad_norm": 0.2752348350354927,
      "learning_rate": 1.324521805134618e-07,
      "loss": 1.5641,
      "step": 4763
    },
    {
      "epoch": 0.7184978508408114,
      "grad_norm": 0.2498284925081655,
      "learning_rate": 1.3236970784184088e-07,
      "loss": 1.6076,
      "step": 4764
    },
    {
      "epoch": 0.718648669029485,
      "grad_norm": 0.24993634033255005,
      "learning_rate": 1.3228726719487437e-07,
      "loss": 1.6529,
      "step": 4765
    },
    {
      "epoch": 0.7187994872181586,
      "grad_norm": 0.24857935421463165,
      "learning_rate": 1.3220485859107267e-07,
      "loss": 1.5523,
      "step": 4766
    },
    {
      "epoch": 0.718950305406832,
      "grad_norm": 0.27442449103397504,
      "learning_rate": 1.321224820489389e-07,
      "loss": 1.6139,
      "step": 4767
    },
    {
      "epoch": 0.7191011235955056,
      "grad_norm": 0.2631059019473054,
      "learning_rate": 1.320401375869689e-07,
      "loss": 1.5582,
      "step": 4768
    },
    {
      "epoch": 0.7192519417841792,
      "grad_norm": 0.2751035641520621,
      "learning_rate": 1.3195782522365145e-07,
      "loss": 1.5074,
      "step": 4769
    },
    {
      "epoch": 0.7194027599728527,
      "grad_norm": 0.254206834439695,
      "learning_rate": 1.3187554497746813e-07,
      "loss": 1.5125,
      "step": 4770
    },
    {
      "epoch": 0.7195535781615263,
      "grad_norm": 0.26443731820936195,
      "learning_rate": 1.3179329686689323e-07,
      "loss": 1.4735,
      "step": 4771
    },
    {
      "epoch": 0.7197043963501998,
      "grad_norm": 0.27480922400219626,
      "learning_rate": 1.3171108091039374e-07,
      "loss": 1.5646,
      "step": 4772
    },
    {
      "epoch": 0.7198552145388734,
      "grad_norm": 0.2668415492180347,
      "learning_rate": 1.316288971264297e-07,
      "loss": 1.5483,
      "step": 4773
    },
    {
      "epoch": 0.7200060327275469,
      "grad_norm": 0.27113682079889084,
      "learning_rate": 1.3154674553345368e-07,
      "loss": 1.5722,
      "step": 4774
    },
    {
      "epoch": 0.7201568509162205,
      "grad_norm": 0.24571186963288189,
      "learning_rate": 1.31464626149911e-07,
      "loss": 1.5348,
      "step": 4775
    },
    {
      "epoch": 0.7203076691048941,
      "grad_norm": 0.33501354542952616,
      "learning_rate": 1.313825389942401e-07,
      "loss": 1.5666,
      "step": 4776
    },
    {
      "epoch": 0.7204584872935677,
      "grad_norm": 0.2708282584987015,
      "learning_rate": 1.3130048408487177e-07,
      "loss": 1.5743,
      "step": 4777
    },
    {
      "epoch": 0.7206093054822411,
      "grad_norm": 0.2560038596933287,
      "learning_rate": 1.3121846144022963e-07,
      "loss": 1.5117,
      "step": 4778
    },
    {
      "epoch": 0.7207601236709147,
      "grad_norm": 0.2460218550726402,
      "learning_rate": 1.3113647107873042e-07,
      "loss": 1.5944,
      "step": 4779
    },
    {
      "epoch": 0.7209109418595883,
      "grad_norm": 0.2540661878105851,
      "learning_rate": 1.310545130187832e-07,
      "loss": 1.5551,
      "step": 4780
    },
    {
      "epoch": 0.7210617600482618,
      "grad_norm": 0.26535813628475846,
      "learning_rate": 1.3097258727878996e-07,
      "loss": 1.5921,
      "step": 4781
    },
    {
      "epoch": 0.7212125782369354,
      "grad_norm": 0.25510044653757036,
      "learning_rate": 1.3089069387714534e-07,
      "loss": 1.5921,
      "step": 4782
    },
    {
      "epoch": 0.721363396425609,
      "grad_norm": 0.258672877029231,
      "learning_rate": 1.30808832832237e-07,
      "loss": 1.5307,
      "step": 4783
    },
    {
      "epoch": 0.7215142146142824,
      "grad_norm": 0.2515153158348181,
      "learning_rate": 1.3072700416244492e-07,
      "loss": 1.564,
      "step": 4784
    },
    {
      "epoch": 0.721665032802956,
      "grad_norm": 0.25595079191995307,
      "learning_rate": 1.3064520788614202e-07,
      "loss": 1.5595,
      "step": 4785
    },
    {
      "epoch": 0.7218158509916296,
      "grad_norm": 0.2503056974253421,
      "learning_rate": 1.3056344402169418e-07,
      "loss": 1.5445,
      "step": 4786
    },
    {
      "epoch": 0.7219666691803032,
      "grad_norm": 0.23401901746378476,
      "learning_rate": 1.3048171258745955e-07,
      "loss": 1.5035,
      "step": 4787
    },
    {
      "epoch": 0.7221174873689767,
      "grad_norm": 0.2654912607361534,
      "learning_rate": 1.3040001360178925e-07,
      "loss": 1.6015,
      "step": 4788
    },
    {
      "epoch": 0.7222683055576502,
      "grad_norm": 0.3003362992871323,
      "learning_rate": 1.3031834708302715e-07,
      "loss": 1.5482,
      "step": 4789
    },
    {
      "epoch": 0.7224191237463238,
      "grad_norm": 0.2459141520442908,
      "learning_rate": 1.3023671304950976e-07,
      "loss": 1.5025,
      "step": 4790
    },
    {
      "epoch": 0.7225699419349974,
      "grad_norm": 0.2769796726755768,
      "learning_rate": 1.3015511151956614e-07,
      "loss": 1.5705,
      "step": 4791
    },
    {
      "epoch": 0.7227207601236709,
      "grad_norm": 0.2628715158476944,
      "learning_rate": 1.3007354251151844e-07,
      "loss": 1.5736,
      "step": 4792
    },
    {
      "epoch": 0.7228715783123445,
      "grad_norm": 0.232517209113232,
      "learning_rate": 1.299920060436812e-07,
      "loss": 1.5009,
      "step": 4793
    },
    {
      "epoch": 0.7230223965010181,
      "grad_norm": 0.250920557316729,
      "learning_rate": 1.2991050213436166e-07,
      "loss": 1.5482,
      "step": 4794
    },
    {
      "epoch": 0.7231732146896915,
      "grad_norm": 0.2692022742934645,
      "learning_rate": 1.298290308018598e-07,
      "loss": 1.4868,
      "step": 4795
    },
    {
      "epoch": 0.7233240328783651,
      "grad_norm": 0.2830417321359642,
      "learning_rate": 1.2974759206446845e-07,
      "loss": 1.5895,
      "step": 4796
    },
    {
      "epoch": 0.7234748510670387,
      "grad_norm": 0.25235529470522844,
      "learning_rate": 1.296661859404729e-07,
      "loss": 1.5513,
      "step": 4797
    },
    {
      "epoch": 0.7236256692557123,
      "grad_norm": 0.26965882097816113,
      "learning_rate": 1.2958481244815111e-07,
      "loss": 1.5082,
      "step": 4798
    },
    {
      "epoch": 0.7237764874443858,
      "grad_norm": 0.37111415105454276,
      "learning_rate": 1.29503471605774e-07,
      "loss": 1.4878,
      "step": 4799
    },
    {
      "epoch": 0.7239273056330593,
      "grad_norm": 0.3637365626005936,
      "learning_rate": 1.2942216343160478e-07,
      "loss": 1.5448,
      "step": 4800
    },
    {
      "epoch": 0.7240781238217329,
      "grad_norm": 0.25419613760280135,
      "learning_rate": 1.2934088794389954e-07,
      "loss": 1.5455,
      "step": 4801
    },
    {
      "epoch": 0.7242289420104064,
      "grad_norm": 0.24881891337070325,
      "learning_rate": 1.2925964516090703e-07,
      "loss": 1.5603,
      "step": 4802
    },
    {
      "epoch": 0.72437976019908,
      "grad_norm": 0.2655133874520916,
      "learning_rate": 1.2917843510086866e-07,
      "loss": 1.61,
      "step": 4803
    },
    {
      "epoch": 0.7245305783877536,
      "grad_norm": 0.24981081472759925,
      "learning_rate": 1.2909725778201842e-07,
      "loss": 1.5664,
      "step": 4804
    },
    {
      "epoch": 0.7246813965764272,
      "grad_norm": 0.2618097018011796,
      "learning_rate": 1.2901611322258285e-07,
      "loss": 1.5684,
      "step": 4805
    },
    {
      "epoch": 0.7248322147651006,
      "grad_norm": 0.2461547827592507,
      "learning_rate": 1.2893500144078148e-07,
      "loss": 1.5833,
      "step": 4806
    },
    {
      "epoch": 0.7249830329537742,
      "grad_norm": 0.3153575492477137,
      "learning_rate": 1.2885392245482618e-07,
      "loss": 1.6003,
      "step": 4807
    },
    {
      "epoch": 0.7251338511424478,
      "grad_norm": 0.2555928410426974,
      "learning_rate": 1.287728762829214e-07,
      "loss": 1.5403,
      "step": 4808
    },
    {
      "epoch": 0.7252846693311213,
      "grad_norm": 0.30337841154326795,
      "learning_rate": 1.2869186294326463e-07,
      "loss": 1.5782,
      "step": 4809
    },
    {
      "epoch": 0.7254354875197949,
      "grad_norm": 0.3306463957817853,
      "learning_rate": 1.286108824540456e-07,
      "loss": 1.5603,
      "step": 4810
    },
    {
      "epoch": 0.7255863057084685,
      "grad_norm": 0.2454194997307586,
      "learning_rate": 1.2852993483344664e-07,
      "loss": 1.5024,
      "step": 4811
    },
    {
      "epoch": 0.725737123897142,
      "grad_norm": 0.26104941804853266,
      "learning_rate": 1.284490200996431e-07,
      "loss": 1.5667,
      "step": 4812
    },
    {
      "epoch": 0.7258879420858155,
      "grad_norm": 0.25637271753239255,
      "learning_rate": 1.2836813827080263e-07,
      "loss": 1.5568,
      "step": 4813
    },
    {
      "epoch": 0.7260387602744891,
      "grad_norm": 0.2648634274855605,
      "learning_rate": 1.2828728936508537e-07,
      "loss": 1.5988,
      "step": 4814
    },
    {
      "epoch": 0.7261895784631627,
      "grad_norm": 0.2612922500483929,
      "learning_rate": 1.282064734006445e-07,
      "loss": 1.5545,
      "step": 4815
    },
    {
      "epoch": 0.7263403966518363,
      "grad_norm": 0.28560409625042155,
      "learning_rate": 1.2812569039562548e-07,
      "loss": 1.6042,
      "step": 4816
    },
    {
      "epoch": 0.7264912148405097,
      "grad_norm": 0.2518002809513442,
      "learning_rate": 1.2804494036816636e-07,
      "loss": 1.4589,
      "step": 4817
    },
    {
      "epoch": 0.7266420330291833,
      "grad_norm": 0.24302328381493488,
      "learning_rate": 1.279642233363979e-07,
      "loss": 1.6153,
      "step": 4818
    },
    {
      "epoch": 0.7267928512178569,
      "grad_norm": 0.27424878212349574,
      "learning_rate": 1.2788353931844347e-07,
      "loss": 1.6389,
      "step": 4819
    },
    {
      "epoch": 0.7269436694065304,
      "grad_norm": 0.2432496072800537,
      "learning_rate": 1.2780288833241905e-07,
      "loss": 1.5211,
      "step": 4820
    },
    {
      "epoch": 0.727094487595204,
      "grad_norm": 0.24526076941733485,
      "learning_rate": 1.277222703964329e-07,
      "loss": 1.5853,
      "step": 4821
    },
    {
      "epoch": 0.7272453057838776,
      "grad_norm": 0.2758571399825683,
      "learning_rate": 1.276416855285864e-07,
      "loss": 1.517,
      "step": 4822
    },
    {
      "epoch": 0.727396123972551,
      "grad_norm": 0.2523044002321618,
      "learning_rate": 1.27561133746973e-07,
      "loss": 1.5637,
      "step": 4823
    },
    {
      "epoch": 0.7275469421612246,
      "grad_norm": 0.2448839969754353,
      "learning_rate": 1.274806150696789e-07,
      "loss": 1.5458,
      "step": 4824
    },
    {
      "epoch": 0.7276977603498982,
      "grad_norm": 0.26831978639492615,
      "learning_rate": 1.2740012951478304e-07,
      "loss": 1.5429,
      "step": 4825
    },
    {
      "epoch": 0.7278485785385718,
      "grad_norm": 0.3017575431328631,
      "learning_rate": 1.2731967710035667e-07,
      "loss": 1.5537,
      "step": 4826
    },
    {
      "epoch": 0.7279993967272453,
      "grad_norm": 0.25548910611664444,
      "learning_rate": 1.2723925784446366e-07,
      "loss": 1.5139,
      "step": 4827
    },
    {
      "epoch": 0.7281502149159189,
      "grad_norm": 0.2505657459733443,
      "learning_rate": 1.271588717651606e-07,
      "loss": 1.4835,
      "step": 4828
    },
    {
      "epoch": 0.7283010331045924,
      "grad_norm": 0.2570799187421738,
      "learning_rate": 1.2707851888049646e-07,
      "loss": 1.5586,
      "step": 4829
    },
    {
      "epoch": 0.728451851293266,
      "grad_norm": 0.24539688464693027,
      "learning_rate": 1.2699819920851274e-07,
      "loss": 1.5456,
      "step": 4830
    },
    {
      "epoch": 0.7286026694819395,
      "grad_norm": 0.25000991554635804,
      "learning_rate": 1.2691791276724353e-07,
      "loss": 1.5591,
      "step": 4831
    },
    {
      "epoch": 0.7287534876706131,
      "grad_norm": 0.24951798023113553,
      "learning_rate": 1.2683765957471563e-07,
      "loss": 1.5319,
      "step": 4832
    },
    {
      "epoch": 0.7289043058592867,
      "grad_norm": 0.3061178431026231,
      "learning_rate": 1.2675743964894808e-07,
      "loss": 1.5712,
      "step": 4833
    },
    {
      "epoch": 0.7290551240479601,
      "grad_norm": 0.26458100141929913,
      "learning_rate": 1.2667725300795256e-07,
      "loss": 1.6257,
      "step": 4834
    },
    {
      "epoch": 0.7292059422366337,
      "grad_norm": 0.3171487655534102,
      "learning_rate": 1.265970996697335e-07,
      "loss": 1.5573,
      "step": 4835
    },
    {
      "epoch": 0.7293567604253073,
      "grad_norm": 0.25811069191209124,
      "learning_rate": 1.2651697965228747e-07,
      "loss": 1.5435,
      "step": 4836
    },
    {
      "epoch": 0.7295075786139809,
      "grad_norm": 0.2565389245468662,
      "learning_rate": 1.2643689297360377e-07,
      "loss": 1.5934,
      "step": 4837
    },
    {
      "epoch": 0.7296583968026544,
      "grad_norm": 0.2633714387416033,
      "learning_rate": 1.263568396516643e-07,
      "loss": 1.5242,
      "step": 4838
    },
    {
      "epoch": 0.729809214991328,
      "grad_norm": 0.2553323568346628,
      "learning_rate": 1.262768197044433e-07,
      "loss": 1.576,
      "step": 4839
    },
    {
      "epoch": 0.7299600331800015,
      "grad_norm": 0.3301487630751055,
      "learning_rate": 1.2619683314990747e-07,
      "loss": 1.5358,
      "step": 4840
    },
    {
      "epoch": 0.730110851368675,
      "grad_norm": 0.25633908493026725,
      "learning_rate": 1.2611688000601637e-07,
      "loss": 1.5936,
      "step": 4841
    },
    {
      "epoch": 0.7302616695573486,
      "grad_norm": 0.24960387878766888,
      "learning_rate": 1.2603696029072158e-07,
      "loss": 1.533,
      "step": 4842
    },
    {
      "epoch": 0.7304124877460222,
      "grad_norm": 0.3349952372116707,
      "learning_rate": 1.2595707402196753e-07,
      "loss": 1.5558,
      "step": 4843
    },
    {
      "epoch": 0.7305633059346958,
      "grad_norm": 0.26827627120663905,
      "learning_rate": 1.258772212176909e-07,
      "loss": 1.5489,
      "step": 4844
    },
    {
      "epoch": 0.7307141241233692,
      "grad_norm": 0.26202005410067175,
      "learning_rate": 1.2579740189582114e-07,
      "loss": 1.6241,
      "step": 4845
    },
    {
      "epoch": 0.7308649423120428,
      "grad_norm": 0.23730845402549805,
      "learning_rate": 1.257176160742799e-07,
      "loss": 1.5512,
      "step": 4846
    },
    {
      "epoch": 0.7310157605007164,
      "grad_norm": 0.24628036544363902,
      "learning_rate": 1.2563786377098134e-07,
      "loss": 1.5274,
      "step": 4847
    },
    {
      "epoch": 0.73116657868939,
      "grad_norm": 0.3359114463154828,
      "learning_rate": 1.2555814500383234e-07,
      "loss": 1.5514,
      "step": 4848
    },
    {
      "epoch": 0.7313173968780635,
      "grad_norm": 0.2545087267482347,
      "learning_rate": 1.2547845979073202e-07,
      "loss": 1.5457,
      "step": 4849
    },
    {
      "epoch": 0.7314682150667371,
      "grad_norm": 0.2385362845552149,
      "learning_rate": 1.2539880814957195e-07,
      "loss": 1.6271,
      "step": 4850
    },
    {
      "epoch": 0.7316190332554106,
      "grad_norm": 0.2630604683576181,
      "learning_rate": 1.2531919009823638e-07,
      "loss": 1.5229,
      "step": 4851
    },
    {
      "epoch": 0.7317698514440841,
      "grad_norm": 0.2413971760254369,
      "learning_rate": 1.2523960565460183e-07,
      "loss": 1.6014,
      "step": 4852
    },
    {
      "epoch": 0.7319206696327577,
      "grad_norm": 0.35491120802072545,
      "learning_rate": 1.2516005483653728e-07,
      "loss": 1.5709,
      "step": 4853
    },
    {
      "epoch": 0.7320714878214313,
      "grad_norm": 0.23795666653362516,
      "learning_rate": 1.2508053766190428e-07,
      "loss": 1.6314,
      "step": 4854
    },
    {
      "epoch": 0.7322223060101049,
      "grad_norm": 0.3201093612781402,
      "learning_rate": 1.2500105414855672e-07,
      "loss": 1.6001,
      "step": 4855
    },
    {
      "epoch": 0.7323731241987784,
      "grad_norm": 0.2570636957802805,
      "learning_rate": 1.2492160431434095e-07,
      "loss": 1.5927,
      "step": 4856
    },
    {
      "epoch": 0.7325239423874519,
      "grad_norm": 0.27672499262231987,
      "learning_rate": 1.248421881770957e-07,
      "loss": 1.5212,
      "step": 4857
    },
    {
      "epoch": 0.7326747605761255,
      "grad_norm": 0.2881789131100355,
      "learning_rate": 1.2476280575465236e-07,
      "loss": 1.5669,
      "step": 4858
    },
    {
      "epoch": 0.732825578764799,
      "grad_norm": 0.2512900593411515,
      "learning_rate": 1.2468345706483454e-07,
      "loss": 1.5854,
      "step": 4859
    },
    {
      "epoch": 0.7329763969534726,
      "grad_norm": 0.24430168831316637,
      "learning_rate": 1.246041421254582e-07,
      "loss": 1.6113,
      "step": 4860
    },
    {
      "epoch": 0.7331272151421462,
      "grad_norm": 0.24967757059348708,
      "learning_rate": 1.2452486095433207e-07,
      "loss": 1.566,
      "step": 4861
    },
    {
      "epoch": 0.7332780333308196,
      "grad_norm": 0.2460779172665949,
      "learning_rate": 1.2444561356925696e-07,
      "loss": 1.5459,
      "step": 4862
    },
    {
      "epoch": 0.7334288515194932,
      "grad_norm": 0.24547609377071855,
      "learning_rate": 1.2436639998802617e-07,
      "loss": 1.5637,
      "step": 4863
    },
    {
      "epoch": 0.7335796697081668,
      "grad_norm": 0.25436731808135615,
      "learning_rate": 1.2428722022842558e-07,
      "loss": 1.5807,
      "step": 4864
    },
    {
      "epoch": 0.7337304878968404,
      "grad_norm": 0.31255660965857957,
      "learning_rate": 1.242080743082333e-07,
      "loss": 1.6048,
      "step": 4865
    },
    {
      "epoch": 0.7338813060855139,
      "grad_norm": 0.24347193523993743,
      "learning_rate": 1.241289622452198e-07,
      "loss": 1.609,
      "step": 4866
    },
    {
      "epoch": 0.7340321242741875,
      "grad_norm": 0.24839489010233448,
      "learning_rate": 1.240498840571482e-07,
      "loss": 1.5773,
      "step": 4867
    },
    {
      "epoch": 0.734182942462861,
      "grad_norm": 0.25858908645724965,
      "learning_rate": 1.239708397617738e-07,
      "loss": 1.5358,
      "step": 4868
    },
    {
      "epoch": 0.7343337606515346,
      "grad_norm": 0.25517992246543114,
      "learning_rate": 1.2389182937684424e-07,
      "loss": 1.4918,
      "step": 4869
    },
    {
      "epoch": 0.7344845788402081,
      "grad_norm": 0.2739609838242563,
      "learning_rate": 1.2381285292009975e-07,
      "loss": 1.5208,
      "step": 4870
    },
    {
      "epoch": 0.7346353970288817,
      "grad_norm": 0.2588505484162014,
      "learning_rate": 1.2373391040927288e-07,
      "loss": 1.5892,
      "step": 4871
    },
    {
      "epoch": 0.7347862152175553,
      "grad_norm": 0.2467211738657118,
      "learning_rate": 1.236550018620885e-07,
      "loss": 1.5012,
      "step": 4872
    },
    {
      "epoch": 0.7349370334062288,
      "grad_norm": 0.2505211482849064,
      "learning_rate": 1.2357612729626386e-07,
      "loss": 1.5041,
      "step": 4873
    },
    {
      "epoch": 0.7350878515949023,
      "grad_norm": 0.23844532247069256,
      "learning_rate": 1.2349728672950852e-07,
      "loss": 1.5505,
      "step": 4874
    },
    {
      "epoch": 0.7352386697835759,
      "grad_norm": 0.25889367117582207,
      "learning_rate": 1.2341848017952463e-07,
      "loss": 1.528,
      "step": 4875
    },
    {
      "epoch": 0.7353894879722495,
      "grad_norm": 0.2632746655943858,
      "learning_rate": 1.2333970766400645e-07,
      "loss": 1.6027,
      "step": 4876
    },
    {
      "epoch": 0.735540306160923,
      "grad_norm": 0.26375667831932015,
      "learning_rate": 1.232609692006407e-07,
      "loss": 1.5342,
      "step": 4877
    },
    {
      "epoch": 0.7356911243495966,
      "grad_norm": 0.2636727335172054,
      "learning_rate": 1.2318226480710657e-07,
      "loss": 1.5426,
      "step": 4878
    },
    {
      "epoch": 0.7358419425382701,
      "grad_norm": 0.2391237416136801,
      "learning_rate": 1.231035945010754e-07,
      "loss": 1.5567,
      "step": 4879
    },
    {
      "epoch": 0.7359927607269436,
      "grad_norm": 0.25500269314353563,
      "learning_rate": 1.2302495830021088e-07,
      "loss": 1.573,
      "step": 4880
    },
    {
      "epoch": 0.7361435789156172,
      "grad_norm": 0.24175138586985964,
      "learning_rate": 1.2294635622216932e-07,
      "loss": 1.4994,
      "step": 4881
    },
    {
      "epoch": 0.7362943971042908,
      "grad_norm": 0.2729612009679162,
      "learning_rate": 1.2286778828459906e-07,
      "loss": 1.594,
      "step": 4882
    },
    {
      "epoch": 0.7364452152929644,
      "grad_norm": 0.33025350959697874,
      "learning_rate": 1.2278925450514088e-07,
      "loss": 1.5365,
      "step": 4883
    },
    {
      "epoch": 0.7365960334816379,
      "grad_norm": 0.24528978937390591,
      "learning_rate": 1.2271075490142795e-07,
      "loss": 1.527,
      "step": 4884
    },
    {
      "epoch": 0.7367468516703114,
      "grad_norm": 0.24587315632508772,
      "learning_rate": 1.2263228949108575e-07,
      "loss": 1.5086,
      "step": 4885
    },
    {
      "epoch": 0.736897669858985,
      "grad_norm": 0.24091396943889565,
      "learning_rate": 1.2255385829173198e-07,
      "loss": 1.5683,
      "step": 4886
    },
    {
      "epoch": 0.7370484880476585,
      "grad_norm": 0.26317210667279417,
      "learning_rate": 1.2247546132097664e-07,
      "loss": 1.5471,
      "step": 4887
    },
    {
      "epoch": 0.7371993062363321,
      "grad_norm": 0.45806378602570813,
      "learning_rate": 1.2239709859642233e-07,
      "loss": 1.5186,
      "step": 4888
    },
    {
      "epoch": 0.7373501244250057,
      "grad_norm": 0.2603858399133861,
      "learning_rate": 1.223187701356637e-07,
      "loss": 1.5954,
      "step": 4889
    },
    {
      "epoch": 0.7375009426136792,
      "grad_norm": 0.2440198533434365,
      "learning_rate": 1.2224047595628764e-07,
      "loss": 1.5374,
      "step": 4890
    },
    {
      "epoch": 0.7376517608023527,
      "grad_norm": 0.8741362887191947,
      "learning_rate": 1.2216221607587366e-07,
      "loss": 1.559,
      "step": 4891
    },
    {
      "epoch": 0.7378025789910263,
      "grad_norm": 0.24756208490889292,
      "learning_rate": 1.2208399051199331e-07,
      "loss": 1.5677,
      "step": 4892
    },
    {
      "epoch": 0.7379533971796999,
      "grad_norm": 0.3713112296550961,
      "learning_rate": 1.2200579928221041e-07,
      "loss": 1.5299,
      "step": 4893
    },
    {
      "epoch": 0.7381042153683735,
      "grad_norm": 0.24478060913393343,
      "learning_rate": 1.2192764240408133e-07,
      "loss": 1.5512,
      "step": 4894
    },
    {
      "epoch": 0.738255033557047,
      "grad_norm": 0.24810953221364054,
      "learning_rate": 1.218495198951545e-07,
      "loss": 1.6259,
      "step": 4895
    },
    {
      "epoch": 0.7384058517457205,
      "grad_norm": 0.3194982608501394,
      "learning_rate": 1.2177143177297062e-07,
      "loss": 1.4957,
      "step": 4896
    },
    {
      "epoch": 0.7385566699343941,
      "grad_norm": 0.23913340094219115,
      "learning_rate": 1.2169337805506286e-07,
      "loss": 1.5925,
      "step": 4897
    },
    {
      "epoch": 0.7387074881230676,
      "grad_norm": 0.2401729841984577,
      "learning_rate": 1.2161535875895652e-07,
      "loss": 1.5484,
      "step": 4898
    },
    {
      "epoch": 0.7388583063117412,
      "grad_norm": 0.2560492767790417,
      "learning_rate": 1.215373739021692e-07,
      "loss": 1.571,
      "step": 4899
    },
    {
      "epoch": 0.7390091245004148,
      "grad_norm": 0.2533453774118272,
      "learning_rate": 1.214594235022107e-07,
      "loss": 1.5562,
      "step": 4900
    },
    {
      "epoch": 0.7391599426890884,
      "grad_norm": 0.24954588639369182,
      "learning_rate": 1.2138150757658327e-07,
      "loss": 1.5795,
      "step": 4901
    },
    {
      "epoch": 0.7393107608777618,
      "grad_norm": 0.3961490119640425,
      "learning_rate": 1.2130362614278125e-07,
      "loss": 1.5271,
      "step": 4902
    },
    {
      "epoch": 0.7394615790664354,
      "grad_norm": 0.24917330164657575,
      "learning_rate": 1.2122577921829124e-07,
      "loss": 1.5983,
      "step": 4903
    },
    {
      "epoch": 0.739612397255109,
      "grad_norm": 0.2498110920325101,
      "learning_rate": 1.211479668205923e-07,
      "loss": 1.6059,
      "step": 4904
    },
    {
      "epoch": 0.7397632154437825,
      "grad_norm": 0.404323150035056,
      "learning_rate": 1.210701889671554e-07,
      "loss": 1.5374,
      "step": 4905
    },
    {
      "epoch": 0.7399140336324561,
      "grad_norm": 0.25752710970601544,
      "learning_rate": 1.20992445675444e-07,
      "loss": 1.5844,
      "step": 4906
    },
    {
      "epoch": 0.7400648518211296,
      "grad_norm": 0.26327930214752393,
      "learning_rate": 1.209147369629138e-07,
      "loss": 1.5387,
      "step": 4907
    },
    {
      "epoch": 0.7402156700098032,
      "grad_norm": 0.26222434865951005,
      "learning_rate": 1.208370628470126e-07,
      "loss": 1.5743,
      "step": 4908
    },
    {
      "epoch": 0.7403664881984767,
      "grad_norm": 0.2464825262906466,
      "learning_rate": 1.2075942334518043e-07,
      "loss": 1.5022,
      "step": 4909
    },
    {
      "epoch": 0.7405173063871503,
      "grad_norm": 0.24335101661561845,
      "learning_rate": 1.206818184748498e-07,
      "loss": 1.5018,
      "step": 4910
    },
    {
      "epoch": 0.7406681245758239,
      "grad_norm": 0.25616286972518576,
      "learning_rate": 1.2060424825344513e-07,
      "loss": 1.5709,
      "step": 4911
    },
    {
      "epoch": 0.7408189427644974,
      "grad_norm": 0.2556399833872158,
      "learning_rate": 1.2052671269838326e-07,
      "loss": 1.4909,
      "step": 4912
    },
    {
      "epoch": 0.7409697609531709,
      "grad_norm": 0.27448818647694906,
      "learning_rate": 1.2044921182707306e-07,
      "loss": 1.5212,
      "step": 4913
    },
    {
      "epoch": 0.7411205791418445,
      "grad_norm": 0.27039827192529786,
      "learning_rate": 1.203717456569159e-07,
      "loss": 1.4644,
      "step": 4914
    },
    {
      "epoch": 0.7412713973305181,
      "grad_norm": 0.2517974792155157,
      "learning_rate": 1.202943142053051e-07,
      "loss": 1.5436,
      "step": 4915
    },
    {
      "epoch": 0.7414222155191916,
      "grad_norm": 0.2514834079126791,
      "learning_rate": 1.2021691748962622e-07,
      "loss": 1.5402,
      "step": 4916
    },
    {
      "epoch": 0.7415730337078652,
      "grad_norm": 0.2475307711527151,
      "learning_rate": 1.2013955552725722e-07,
      "loss": 1.5476,
      "step": 4917
    },
    {
      "epoch": 0.7417238518965388,
      "grad_norm": 0.25423670019438954,
      "learning_rate": 1.2006222833556805e-07,
      "loss": 1.6096,
      "step": 4918
    },
    {
      "epoch": 0.7418746700852122,
      "grad_norm": 0.27267555069412147,
      "learning_rate": 1.1998493593192084e-07,
      "loss": 1.5231,
      "step": 4919
    },
    {
      "epoch": 0.7420254882738858,
      "grad_norm": 0.26121014975893625,
      "learning_rate": 1.1990767833367011e-07,
      "loss": 1.5762,
      "step": 4920
    },
    {
      "epoch": 0.7421763064625594,
      "grad_norm": 0.2595423583273133,
      "learning_rate": 1.1983045555816236e-07,
      "loss": 1.5498,
      "step": 4921
    },
    {
      "epoch": 0.742327124651233,
      "grad_norm": 0.2541893990545665,
      "learning_rate": 1.1975326762273634e-07,
      "loss": 1.5345,
      "step": 4922
    },
    {
      "epoch": 0.7424779428399065,
      "grad_norm": 0.24344151161699468,
      "learning_rate": 1.1967611454472306e-07,
      "loss": 1.52,
      "step": 4923
    },
    {
      "epoch": 0.74262876102858,
      "grad_norm": 0.24929965046703223,
      "learning_rate": 1.1959899634144562e-07,
      "loss": 1.575,
      "step": 4924
    },
    {
      "epoch": 0.7427795792172536,
      "grad_norm": 0.26155719723606896,
      "learning_rate": 1.1952191303021926e-07,
      "loss": 1.5896,
      "step": 4925
    },
    {
      "epoch": 0.7429303974059271,
      "grad_norm": 0.26977541332904076,
      "learning_rate": 1.1944486462835139e-07,
      "loss": 1.5467,
      "step": 4926
    },
    {
      "epoch": 0.7430812155946007,
      "grad_norm": 0.27565973767509366,
      "learning_rate": 1.1936785115314173e-07,
      "loss": 1.5633,
      "step": 4927
    },
    {
      "epoch": 0.7432320337832743,
      "grad_norm": 0.25178798058837637,
      "learning_rate": 1.1929087262188203e-07,
      "loss": 1.5646,
      "step": 4928
    },
    {
      "epoch": 0.7433828519719479,
      "grad_norm": 0.2527947496709433,
      "learning_rate": 1.192139290518561e-07,
      "loss": 1.5178,
      "step": 4929
    },
    {
      "epoch": 0.7435336701606213,
      "grad_norm": 0.2510570946797427,
      "learning_rate": 1.1913702046034014e-07,
      "loss": 1.5101,
      "step": 4930
    },
    {
      "epoch": 0.7436844883492949,
      "grad_norm": 0.25647391759096383,
      "learning_rate": 1.1906014686460233e-07,
      "loss": 1.6167,
      "step": 4931
    },
    {
      "epoch": 0.7438353065379685,
      "grad_norm": 0.5134223535051595,
      "learning_rate": 1.1898330828190293e-07,
      "loss": 1.567,
      "step": 4932
    },
    {
      "epoch": 0.743986124726642,
      "grad_norm": 0.3244622011687557,
      "learning_rate": 1.1890650472949463e-07,
      "loss": 1.5749,
      "step": 4933
    },
    {
      "epoch": 0.7441369429153156,
      "grad_norm": 0.2515454757116323,
      "learning_rate": 1.1882973622462195e-07,
      "loss": 1.5309,
      "step": 4934
    },
    {
      "epoch": 0.7442877611039891,
      "grad_norm": 0.2450180040181453,
      "learning_rate": 1.1875300278452166e-07,
      "loss": 1.5503,
      "step": 4935
    },
    {
      "epoch": 0.7444385792926627,
      "grad_norm": 0.30533582464679626,
      "learning_rate": 1.1867630442642259e-07,
      "loss": 1.5621,
      "step": 4936
    },
    {
      "epoch": 0.7445893974813362,
      "grad_norm": 0.26862067115477706,
      "learning_rate": 1.185996411675459e-07,
      "loss": 1.5826,
      "step": 4937
    },
    {
      "epoch": 0.7447402156700098,
      "grad_norm": 0.24082233081492946,
      "learning_rate": 1.1852301302510466e-07,
      "loss": 1.533,
      "step": 4938
    },
    {
      "epoch": 0.7448910338586834,
      "grad_norm": 0.2594698908196981,
      "learning_rate": 1.1844642001630402e-07,
      "loss": 1.5032,
      "step": 4939
    },
    {
      "epoch": 0.745041852047357,
      "grad_norm": 0.2559417207583824,
      "learning_rate": 1.1836986215834152e-07,
      "loss": 1.5699,
      "step": 4940
    },
    {
      "epoch": 0.7451926702360304,
      "grad_norm": 0.2812017700416851,
      "learning_rate": 1.1829333946840654e-07,
      "loss": 1.494,
      "step": 4941
    },
    {
      "epoch": 0.745343488424704,
      "grad_norm": 0.2468609173351973,
      "learning_rate": 1.1821685196368056e-07,
      "loss": 1.5215,
      "step": 4942
    },
    {
      "epoch": 0.7454943066133776,
      "grad_norm": 0.24703477600416315,
      "learning_rate": 1.1814039966133743e-07,
      "loss": 1.5627,
      "step": 4943
    },
    {
      "epoch": 0.7456451248020511,
      "grad_norm": 0.26930351840059186,
      "learning_rate": 1.1806398257854283e-07,
      "loss": 1.6403,
      "step": 4944
    },
    {
      "epoch": 0.7457959429907247,
      "grad_norm": 0.2484494879334176,
      "learning_rate": 1.1798760073245456e-07,
      "loss": 1.5543,
      "step": 4945
    },
    {
      "epoch": 0.7459467611793983,
      "grad_norm": 0.2598991579530039,
      "learning_rate": 1.179112541402227e-07,
      "loss": 1.5641,
      "step": 4946
    },
    {
      "epoch": 0.7460975793680718,
      "grad_norm": 0.2499149483355594,
      "learning_rate": 1.178349428189892e-07,
      "loss": 1.6063,
      "step": 4947
    },
    {
      "epoch": 0.7462483975567453,
      "grad_norm": 0.26364334144722557,
      "learning_rate": 1.177586667858882e-07,
      "loss": 1.5886,
      "step": 4948
    },
    {
      "epoch": 0.7463992157454189,
      "grad_norm": 0.24273355169903088,
      "learning_rate": 1.176824260580458e-07,
      "loss": 1.5432,
      "step": 4949
    },
    {
      "epoch": 0.7465500339340925,
      "grad_norm": 0.248398952345595,
      "learning_rate": 1.1760622065258045e-07,
      "loss": 1.5106,
      "step": 4950
    },
    {
      "epoch": 0.746700852122766,
      "grad_norm": 0.2591335359742776,
      "learning_rate": 1.1753005058660235e-07,
      "loss": 1.6749,
      "step": 4951
    },
    {
      "epoch": 0.7468516703114395,
      "grad_norm": 0.2685687249343771,
      "learning_rate": 1.1745391587721384e-07,
      "loss": 1.5833,
      "step": 4952
    },
    {
      "epoch": 0.7470024885001131,
      "grad_norm": 0.260740723925062,
      "learning_rate": 1.1737781654150954e-07,
      "loss": 1.5123,
      "step": 4953
    },
    {
      "epoch": 0.7471533066887867,
      "grad_norm": 0.254406185396711,
      "learning_rate": 1.1730175259657592e-07,
      "loss": 1.6124,
      "step": 4954
    },
    {
      "epoch": 0.7473041248774602,
      "grad_norm": 0.249995359524058,
      "learning_rate": 1.1722572405949142e-07,
      "loss": 1.6095,
      "step": 4955
    },
    {
      "epoch": 0.7474549430661338,
      "grad_norm": 0.40428727869443287,
      "learning_rate": 1.1714973094732685e-07,
      "loss": 1.5825,
      "step": 4956
    },
    {
      "epoch": 0.7476057612548074,
      "grad_norm": 0.2707533573954845,
      "learning_rate": 1.170737732771448e-07,
      "loss": 1.5751,
      "step": 4957
    },
    {
      "epoch": 0.7477565794434808,
      "grad_norm": 0.24719276947553667,
      "learning_rate": 1.1699785106599988e-07,
      "loss": 1.4995,
      "step": 4958
    },
    {
      "epoch": 0.7479073976321544,
      "grad_norm": 0.37578501917362633,
      "learning_rate": 1.1692196433093901e-07,
      "loss": 1.535,
      "step": 4959
    },
    {
      "epoch": 0.748058215820828,
      "grad_norm": 0.2815482979032689,
      "learning_rate": 1.1684611308900089e-07,
      "loss": 1.5078,
      "step": 4960
    },
    {
      "epoch": 0.7482090340095016,
      "grad_norm": 0.23479697436304425,
      "learning_rate": 1.1677029735721633e-07,
      "loss": 1.5504,
      "step": 4961
    },
    {
      "epoch": 0.7483598521981751,
      "grad_norm": 0.2573652262763795,
      "learning_rate": 1.1669451715260808e-07,
      "loss": 1.5825,
      "step": 4962
    },
    {
      "epoch": 0.7485106703868487,
      "grad_norm": 0.2531541201062456,
      "learning_rate": 1.1661877249219119e-07,
      "loss": 1.5803,
      "step": 4963
    },
    {
      "epoch": 0.7486614885755222,
      "grad_norm": 0.23667926303795242,
      "learning_rate": 1.1654306339297243e-07,
      "loss": 1.4844,
      "step": 4964
    },
    {
      "epoch": 0.7488123067641957,
      "grad_norm": 0.2478635136891246,
      "learning_rate": 1.1646738987195066e-07,
      "loss": 1.6326,
      "step": 4965
    },
    {
      "epoch": 0.7489631249528693,
      "grad_norm": 0.2392680276111686,
      "learning_rate": 1.1639175194611691e-07,
      "loss": 1.517,
      "step": 4966
    },
    {
      "epoch": 0.7491139431415429,
      "grad_norm": 0.28320944578932267,
      "learning_rate": 1.1631614963245402e-07,
      "loss": 1.6226,
      "step": 4967
    },
    {
      "epoch": 0.7492647613302165,
      "grad_norm": 0.24608625349818133,
      "learning_rate": 1.1624058294793687e-07,
      "loss": 1.5865,
      "step": 4968
    },
    {
      "epoch": 0.7494155795188899,
      "grad_norm": 0.24516006928790682,
      "learning_rate": 1.1616505190953247e-07,
      "loss": 1.5116,
      "step": 4969
    },
    {
      "epoch": 0.7495663977075635,
      "grad_norm": 0.24647472487682642,
      "learning_rate": 1.1608955653419973e-07,
      "loss": 1.5507,
      "step": 4970
    },
    {
      "epoch": 0.7497172158962371,
      "grad_norm": 0.24875229566089416,
      "learning_rate": 1.1601409683888945e-07,
      "loss": 1.4911,
      "step": 4971
    },
    {
      "epoch": 0.7498680340849107,
      "grad_norm": 0.23487532169749784,
      "learning_rate": 1.159386728405447e-07,
      "loss": 1.5221,
      "step": 4972
    },
    {
      "epoch": 0.7500188522735842,
      "grad_norm": 0.25496571236082377,
      "learning_rate": 1.1586328455610026e-07,
      "loss": 1.5835,
      "step": 4973
    },
    {
      "epoch": 0.7501696704622578,
      "grad_norm": 0.2596088183079915,
      "learning_rate": 1.1578793200248302e-07,
      "loss": 1.5166,
      "step": 4974
    },
    {
      "epoch": 0.7503204886509313,
      "grad_norm": 0.27419554174727,
      "learning_rate": 1.1571261519661174e-07,
      "loss": 1.5739,
      "step": 4975
    },
    {
      "epoch": 0.7504713068396048,
      "grad_norm": 0.2543888042401833,
      "learning_rate": 1.1563733415539738e-07,
      "loss": 1.559,
      "step": 4976
    },
    {
      "epoch": 0.7506221250282784,
      "grad_norm": 0.25937322181867734,
      "learning_rate": 1.1556208889574265e-07,
      "loss": 1.5478,
      "step": 4977
    },
    {
      "epoch": 0.750772943216952,
      "grad_norm": 0.28582306238223393,
      "learning_rate": 1.1548687943454221e-07,
      "loss": 1.5822,
      "step": 4978
    },
    {
      "epoch": 0.7509237614056256,
      "grad_norm": 0.24300248886885564,
      "learning_rate": 1.15411705788683e-07,
      "loss": 1.521,
      "step": 4979
    },
    {
      "epoch": 0.751074579594299,
      "grad_norm": 0.2466945390160691,
      "learning_rate": 1.1533656797504351e-07,
      "loss": 1.6157,
      "step": 4980
    },
    {
      "epoch": 0.7512253977829726,
      "grad_norm": 0.5555991220101545,
      "learning_rate": 1.1526146601049439e-07,
      "loss": 1.5467,
      "step": 4981
    },
    {
      "epoch": 0.7513762159716462,
      "grad_norm": 0.25727078930771363,
      "learning_rate": 1.1518639991189831e-07,
      "loss": 1.5622,
      "step": 4982
    },
    {
      "epoch": 0.7515270341603197,
      "grad_norm": 0.2451247504649221,
      "learning_rate": 1.1511136969610975e-07,
      "loss": 1.5494,
      "step": 4983
    },
    {
      "epoch": 0.7516778523489933,
      "grad_norm": 0.2447929104257427,
      "learning_rate": 1.1503637537997508e-07,
      "loss": 1.5519,
      "step": 4984
    },
    {
      "epoch": 0.7518286705376669,
      "grad_norm": 0.24481084601332812,
      "learning_rate": 1.1496141698033284e-07,
      "loss": 1.5932,
      "step": 4985
    },
    {
      "epoch": 0.7519794887263404,
      "grad_norm": 0.3341531795440303,
      "learning_rate": 1.1488649451401336e-07,
      "loss": 1.5481,
      "step": 4986
    },
    {
      "epoch": 0.7521303069150139,
      "grad_norm": 0.24060085308458376,
      "learning_rate": 1.1481160799783889e-07,
      "loss": 1.6076,
      "step": 4987
    },
    {
      "epoch": 0.7522811251036875,
      "grad_norm": 0.2607819323107068,
      "learning_rate": 1.1473675744862353e-07,
      "loss": 1.6102,
      "step": 4988
    },
    {
      "epoch": 0.7524319432923611,
      "grad_norm": 0.4569374987845713,
      "learning_rate": 1.1466194288317359e-07,
      "loss": 1.5122,
      "step": 4989
    },
    {
      "epoch": 0.7525827614810346,
      "grad_norm": 0.2642753303846815,
      "learning_rate": 1.1458716431828701e-07,
      "loss": 1.6151,
      "step": 4990
    },
    {
      "epoch": 0.7527335796697082,
      "grad_norm": 0.24128226500742275,
      "learning_rate": 1.1451242177075373e-07,
      "loss": 1.5247,
      "step": 4991
    },
    {
      "epoch": 0.7528843978583817,
      "grad_norm": 0.2464668191518985,
      "learning_rate": 1.1443771525735575e-07,
      "loss": 1.574,
      "step": 4992
    },
    {
      "epoch": 0.7530352160470553,
      "grad_norm": 0.24845291887234755,
      "learning_rate": 1.1436304479486679e-07,
      "loss": 1.5154,
      "step": 4993
    },
    {
      "epoch": 0.7531860342357288,
      "grad_norm": 0.2611909477100272,
      "learning_rate": 1.1428841040005243e-07,
      "loss": 1.6241,
      "step": 4994
    },
    {
      "epoch": 0.7533368524244024,
      "grad_norm": 0.25661868482566025,
      "learning_rate": 1.142138120896705e-07,
      "loss": 1.5354,
      "step": 4995
    },
    {
      "epoch": 0.753487670613076,
      "grad_norm": 0.2798025246675307,
      "learning_rate": 1.1413924988047034e-07,
      "loss": 1.5309,
      "step": 4996
    },
    {
      "epoch": 0.7536384888017494,
      "grad_norm": 0.27912914616880197,
      "learning_rate": 1.1406472378919329e-07,
      "loss": 1.5718,
      "step": 4997
    },
    {
      "epoch": 0.753789306990423,
      "grad_norm": 0.24277232163790605,
      "learning_rate": 1.1399023383257277e-07,
      "loss": 1.5643,
      "step": 4998
    },
    {
      "epoch": 0.7539401251790966,
      "grad_norm": 0.3617782813224853,
      "learning_rate": 1.1391578002733388e-07,
      "loss": 1.5847,
      "step": 4999
    },
    {
      "epoch": 0.7540909433677702,
      "grad_norm": 0.26007769529671915,
      "learning_rate": 1.1384136239019368e-07,
      "loss": 1.5742,
      "step": 5000
    },
    {
      "epoch": 0.7542417615564437,
      "grad_norm": 0.26579375940669586,
      "learning_rate": 1.13766980937861e-07,
      "loss": 1.5151,
      "step": 5001
    },
    {
      "epoch": 0.7543925797451173,
      "grad_norm": 0.2720902322803907,
      "learning_rate": 1.1369263568703679e-07,
      "loss": 1.5097,
      "step": 5002
    },
    {
      "epoch": 0.7545433979337908,
      "grad_norm": 0.2511906572314177,
      "learning_rate": 1.1361832665441365e-07,
      "loss": 1.5099,
      "step": 5003
    },
    {
      "epoch": 0.7546942161224643,
      "grad_norm": 0.24154146744530233,
      "learning_rate": 1.1354405385667607e-07,
      "loss": 1.5492,
      "step": 5004
    },
    {
      "epoch": 0.7548450343111379,
      "grad_norm": 0.2572353015729826,
      "learning_rate": 1.1346981731050058e-07,
      "loss": 1.6045,
      "step": 5005
    },
    {
      "epoch": 0.7549958524998115,
      "grad_norm": 0.25829188720635793,
      "learning_rate": 1.133956170325554e-07,
      "loss": 1.5816,
      "step": 5006
    },
    {
      "epoch": 0.7551466706884851,
      "grad_norm": 0.2774856071266561,
      "learning_rate": 1.1332145303950055e-07,
      "loss": 1.5602,
      "step": 5007
    },
    {
      "epoch": 0.7552974888771586,
      "grad_norm": 0.2721287551465047,
      "learning_rate": 1.1324732534798817e-07,
      "loss": 1.5472,
      "step": 5008
    },
    {
      "epoch": 0.7554483070658321,
      "grad_norm": 0.2436616598675525,
      "learning_rate": 1.1317323397466201e-07,
      "loss": 1.6173,
      "step": 5009
    },
    {
      "epoch": 0.7555991252545057,
      "grad_norm": 0.25715980408099304,
      "learning_rate": 1.1309917893615767e-07,
      "loss": 1.5151,
      "step": 5010
    },
    {
      "epoch": 0.7557499434431793,
      "grad_norm": 0.2464845622169951,
      "learning_rate": 1.1302516024910279e-07,
      "loss": 1.6045,
      "step": 5011
    },
    {
      "epoch": 0.7559007616318528,
      "grad_norm": 0.2672272217687471,
      "learning_rate": 1.1295117793011667e-07,
      "loss": 1.5588,
      "step": 5012
    },
    {
      "epoch": 0.7560515798205264,
      "grad_norm": 0.2424124619486496,
      "learning_rate": 1.1287723199581049e-07,
      "loss": 1.6073,
      "step": 5013
    },
    {
      "epoch": 0.7562023980091999,
      "grad_norm": 0.2584854770320498,
      "learning_rate": 1.128033224627872e-07,
      "loss": 1.5845,
      "step": 5014
    },
    {
      "epoch": 0.7563532161978734,
      "grad_norm": 0.2836166456867082,
      "learning_rate": 1.1272944934764175e-07,
      "loss": 1.6881,
      "step": 5015
    },
    {
      "epoch": 0.756504034386547,
      "grad_norm": 0.24450066706031187,
      "learning_rate": 1.1265561266696079e-07,
      "loss": 1.5112,
      "step": 5016
    },
    {
      "epoch": 0.7566548525752206,
      "grad_norm": 0.2352581205448825,
      "learning_rate": 1.125818124373227e-07,
      "loss": 1.5355,
      "step": 5017
    },
    {
      "epoch": 0.7568056707638942,
      "grad_norm": 0.24222512360340281,
      "learning_rate": 1.1250804867529792e-07,
      "loss": 1.5726,
      "step": 5018
    },
    {
      "epoch": 0.7569564889525677,
      "grad_norm": 0.26113164150923096,
      "learning_rate": 1.1243432139744849e-07,
      "loss": 1.5462,
      "step": 5019
    },
    {
      "epoch": 0.7571073071412412,
      "grad_norm": 0.24774383004217806,
      "learning_rate": 1.1236063062032832e-07,
      "loss": 1.6045,
      "step": 5020
    },
    {
      "epoch": 0.7572581253299148,
      "grad_norm": 0.2895263852510736,
      "learning_rate": 1.1228697636048318e-07,
      "loss": 1.5447,
      "step": 5021
    },
    {
      "epoch": 0.7574089435185883,
      "grad_norm": 0.24985005566487262,
      "learning_rate": 1.1221335863445062e-07,
      "loss": 1.54,
      "step": 5022
    },
    {
      "epoch": 0.7575597617072619,
      "grad_norm": 0.30047437792071696,
      "learning_rate": 1.1213977745875986e-07,
      "loss": 1.5912,
      "step": 5023
    },
    {
      "epoch": 0.7577105798959355,
      "grad_norm": 0.2559631034377228,
      "learning_rate": 1.1206623284993205e-07,
      "loss": 1.576,
      "step": 5024
    },
    {
      "epoch": 0.757861398084609,
      "grad_norm": 0.24488841104514433,
      "learning_rate": 1.1199272482448016e-07,
      "loss": 1.5611,
      "step": 5025
    },
    {
      "epoch": 0.7580122162732825,
      "grad_norm": 0.2506081363400735,
      "learning_rate": 1.1191925339890885e-07,
      "loss": 1.5457,
      "step": 5026
    },
    {
      "epoch": 0.7581630344619561,
      "grad_norm": 0.2548257218134441,
      "learning_rate": 1.1184581858971454e-07,
      "loss": 1.5363,
      "step": 5027
    },
    {
      "epoch": 0.7583138526506297,
      "grad_norm": 0.2599352467154265,
      "learning_rate": 1.1177242041338558e-07,
      "loss": 1.5705,
      "step": 5028
    },
    {
      "epoch": 0.7584646708393032,
      "grad_norm": 0.2992245833249952,
      "learning_rate": 1.1169905888640197e-07,
      "loss": 1.5595,
      "step": 5029
    },
    {
      "epoch": 0.7586154890279768,
      "grad_norm": 0.26473812364540256,
      "learning_rate": 1.1162573402523539e-07,
      "loss": 1.5335,
      "step": 5030
    },
    {
      "epoch": 0.7587663072166503,
      "grad_norm": 0.24633036947358974,
      "learning_rate": 1.115524458463496e-07,
      "loss": 1.516,
      "step": 5031
    },
    {
      "epoch": 0.7589171254053239,
      "grad_norm": 0.2593413769776572,
      "learning_rate": 1.1147919436619982e-07,
      "loss": 1.6015,
      "step": 5032
    },
    {
      "epoch": 0.7590679435939974,
      "grad_norm": 0.2752753592917604,
      "learning_rate": 1.114059796012331e-07,
      "loss": 1.5842,
      "step": 5033
    },
    {
      "epoch": 0.759218761782671,
      "grad_norm": 0.2764460999783728,
      "learning_rate": 1.1133280156788841e-07,
      "loss": 1.5341,
      "step": 5034
    },
    {
      "epoch": 0.7593695799713446,
      "grad_norm": 0.24793911244162387,
      "learning_rate": 1.1125966028259628e-07,
      "loss": 1.6194,
      "step": 5035
    },
    {
      "epoch": 0.7595203981600182,
      "grad_norm": 0.26031083680762124,
      "learning_rate": 1.1118655576177906e-07,
      "loss": 1.5888,
      "step": 5036
    },
    {
      "epoch": 0.7596712163486916,
      "grad_norm": 0.2483042361062299,
      "learning_rate": 1.1111348802185078e-07,
      "loss": 1.5365,
      "step": 5037
    },
    {
      "epoch": 0.7598220345373652,
      "grad_norm": 0.247028271129063,
      "learning_rate": 1.1104045707921743e-07,
      "loss": 1.5584,
      "step": 5038
    },
    {
      "epoch": 0.7599728527260388,
      "grad_norm": 0.27686028713949173,
      "learning_rate": 1.1096746295027647e-07,
      "loss": 1.5153,
      "step": 5039
    },
    {
      "epoch": 0.7601236709147123,
      "grad_norm": 0.2546933802016954,
      "learning_rate": 1.1089450565141718e-07,
      "loss": 1.6124,
      "step": 5040
    },
    {
      "epoch": 0.7602744891033859,
      "grad_norm": 0.29014179257294564,
      "learning_rate": 1.1082158519902072e-07,
      "loss": 1.5824,
      "step": 5041
    },
    {
      "epoch": 0.7604253072920594,
      "grad_norm": 0.2604184056232571,
      "learning_rate": 1.1074870160945976e-07,
      "loss": 1.6372,
      "step": 5042
    },
    {
      "epoch": 0.760576125480733,
      "grad_norm": 0.25121624505515716,
      "learning_rate": 1.1067585489909878e-07,
      "loss": 1.5824,
      "step": 5043
    },
    {
      "epoch": 0.7607269436694065,
      "grad_norm": 0.282952493155054,
      "learning_rate": 1.1060304508429406e-07,
      "loss": 1.6134,
      "step": 5044
    },
    {
      "epoch": 0.7608777618580801,
      "grad_norm": 0.46881777025757665,
      "learning_rate": 1.1053027218139351e-07,
      "loss": 1.4651,
      "step": 5045
    },
    {
      "epoch": 0.7610285800467537,
      "grad_norm": 0.24818367502647792,
      "learning_rate": 1.1045753620673667e-07,
      "loss": 1.5725,
      "step": 5046
    },
    {
      "epoch": 0.7611793982354272,
      "grad_norm": 0.24944519143597316,
      "learning_rate": 1.1038483717665503e-07,
      "loss": 1.5695,
      "step": 5047
    },
    {
      "epoch": 0.7613302164241007,
      "grad_norm": 0.2537350616836102,
      "learning_rate": 1.1031217510747156e-07,
      "loss": 1.5754,
      "step": 5048
    },
    {
      "epoch": 0.7614810346127743,
      "grad_norm": 0.2466758581154355,
      "learning_rate": 1.1023955001550103e-07,
      "loss": 1.5672,
      "step": 5049
    },
    {
      "epoch": 0.7616318528014479,
      "grad_norm": 0.274456611738853,
      "learning_rate": 1.1016696191704979e-07,
      "loss": 1.5828,
      "step": 5050
    },
    {
      "epoch": 0.7617826709901214,
      "grad_norm": 0.24776615534578775,
      "learning_rate": 1.1009441082841614e-07,
      "loss": 1.6203,
      "step": 5051
    },
    {
      "epoch": 0.761933489178795,
      "grad_norm": 0.27968409507398106,
      "learning_rate": 1.1002189676588985e-07,
      "loss": 1.5833,
      "step": 5052
    },
    {
      "epoch": 0.7620843073674686,
      "grad_norm": 0.2570880513767808,
      "learning_rate": 1.0994941974575237e-07,
      "loss": 1.5976,
      "step": 5053
    },
    {
      "epoch": 0.762235125556142,
      "grad_norm": 0.5737033159122047,
      "learning_rate": 1.0987697978427702e-07,
      "loss": 1.5633,
      "step": 5054
    },
    {
      "epoch": 0.7623859437448156,
      "grad_norm": 0.247452268839846,
      "learning_rate": 1.0980457689772862e-07,
      "loss": 1.5988,
      "step": 5055
    },
    {
      "epoch": 0.7625367619334892,
      "grad_norm": 0.2438038408104602,
      "learning_rate": 1.0973221110236363e-07,
      "loss": 1.5329,
      "step": 5056
    },
    {
      "epoch": 0.7626875801221628,
      "grad_norm": 0.2730328055823197,
      "learning_rate": 1.096598824144305e-07,
      "loss": 1.5276,
      "step": 5057
    },
    {
      "epoch": 0.7628383983108363,
      "grad_norm": 0.2570461385203547,
      "learning_rate": 1.0958759085016896e-07,
      "loss": 1.5214,
      "step": 5058
    },
    {
      "epoch": 0.7629892164995098,
      "grad_norm": 0.24501258526815628,
      "learning_rate": 1.095153364258106e-07,
      "loss": 1.5642,
      "step": 5059
    },
    {
      "epoch": 0.7631400346881834,
      "grad_norm": 0.25496296302770444,
      "learning_rate": 1.0944311915757869e-07,
      "loss": 1.5868,
      "step": 5060
    },
    {
      "epoch": 0.7632908528768569,
      "grad_norm": 0.25572422863875494,
      "learning_rate": 1.0937093906168807e-07,
      "loss": 1.5427,
      "step": 5061
    },
    {
      "epoch": 0.7634416710655305,
      "grad_norm": 0.29023173011170667,
      "learning_rate": 1.0929879615434529e-07,
      "loss": 1.5807,
      "step": 5062
    },
    {
      "epoch": 0.7635924892542041,
      "grad_norm": 0.26293886464282606,
      "learning_rate": 1.0922669045174846e-07,
      "loss": 1.5797,
      "step": 5063
    },
    {
      "epoch": 0.7637433074428777,
      "grad_norm": 0.3782063987669869,
      "learning_rate": 1.0915462197008754e-07,
      "loss": 1.5318,
      "step": 5064
    },
    {
      "epoch": 0.7638941256315511,
      "grad_norm": 0.28029250684341045,
      "learning_rate": 1.0908259072554397e-07,
      "loss": 1.5944,
      "step": 5065
    },
    {
      "epoch": 0.7640449438202247,
      "grad_norm": 0.25338732580070844,
      "learning_rate": 1.0901059673429072e-07,
      "loss": 1.5392,
      "step": 5066
    },
    {
      "epoch": 0.7641957620088983,
      "grad_norm": 0.2500040273740219,
      "learning_rate": 1.0893864001249275e-07,
      "loss": 1.5127,
      "step": 5067
    },
    {
      "epoch": 0.7643465801975718,
      "grad_norm": 0.2629491004831539,
      "learning_rate": 1.0886672057630634e-07,
      "loss": 1.5428,
      "step": 5068
    },
    {
      "epoch": 0.7644973983862454,
      "grad_norm": 0.2555813459761103,
      "learning_rate": 1.0879483844187939e-07,
      "loss": 1.5993,
      "step": 5069
    },
    {
      "epoch": 0.7646482165749189,
      "grad_norm": 0.2511057797810466,
      "learning_rate": 1.0872299362535173e-07,
      "loss": 1.5218,
      "step": 5070
    },
    {
      "epoch": 0.7647990347635925,
      "grad_norm": 0.24639035790289063,
      "learning_rate": 1.0865118614285451e-07,
      "loss": 1.5672,
      "step": 5071
    },
    {
      "epoch": 0.764949852952266,
      "grad_norm": 0.25237923393152023,
      "learning_rate": 1.0857941601051059e-07,
      "loss": 1.5934,
      "step": 5072
    },
    {
      "epoch": 0.7651006711409396,
      "grad_norm": 0.24358021773191016,
      "learning_rate": 1.085076832444345e-07,
      "loss": 1.5214,
      "step": 5073
    },
    {
      "epoch": 0.7652514893296132,
      "grad_norm": 0.2757877174872806,
      "learning_rate": 1.0843598786073233e-07,
      "loss": 1.5243,
      "step": 5074
    },
    {
      "epoch": 0.7654023075182868,
      "grad_norm": 0.24732749051442404,
      "learning_rate": 1.0836432987550176e-07,
      "loss": 1.5443,
      "step": 5075
    },
    {
      "epoch": 0.7655531257069602,
      "grad_norm": 0.2519313151788913,
      "learning_rate": 1.0829270930483204e-07,
      "loss": 1.5767,
      "step": 5076
    },
    {
      "epoch": 0.7657039438956338,
      "grad_norm": 0.35168934252187173,
      "learning_rate": 1.0822112616480412e-07,
      "loss": 1.5573,
      "step": 5077
    },
    {
      "epoch": 0.7658547620843074,
      "grad_norm": 0.2848108080477447,
      "learning_rate": 1.081495804714906e-07,
      "loss": 1.5374,
      "step": 5078
    },
    {
      "epoch": 0.7660055802729809,
      "grad_norm": 0.455667879865247,
      "learning_rate": 1.0807807224095547e-07,
      "loss": 1.6263,
      "step": 5079
    },
    {
      "epoch": 0.7661563984616545,
      "grad_norm": 0.25405772398517595,
      "learning_rate": 1.0800660148925436e-07,
      "loss": 1.5841,
      "step": 5080
    },
    {
      "epoch": 0.7663072166503281,
      "grad_norm": 0.253593517656124,
      "learning_rate": 1.0793516823243468e-07,
      "loss": 1.5692,
      "step": 5081
    },
    {
      "epoch": 0.7664580348390015,
      "grad_norm": 0.3045819421057285,
      "learning_rate": 1.078637724865352e-07,
      "loss": 1.5518,
      "step": 5082
    },
    {
      "epoch": 0.7666088530276751,
      "grad_norm": 0.4177817143813259,
      "learning_rate": 1.0779241426758627e-07,
      "loss": 1.5759,
      "step": 5083
    },
    {
      "epoch": 0.7667596712163487,
      "grad_norm": 0.24682410044797307,
      "learning_rate": 1.0772109359161003e-07,
      "loss": 1.5823,
      "step": 5084
    },
    {
      "epoch": 0.7669104894050223,
      "grad_norm": 0.2693160890115367,
      "learning_rate": 1.0764981047461996e-07,
      "loss": 1.591,
      "step": 5085
    },
    {
      "epoch": 0.7670613075936958,
      "grad_norm": 0.2505752972828048,
      "learning_rate": 1.0757856493262116e-07,
      "loss": 1.594,
      "step": 5086
    },
    {
      "epoch": 0.7672121257823693,
      "grad_norm": 0.3321816099078353,
      "learning_rate": 1.0750735698161045e-07,
      "loss": 1.5276,
      "step": 5087
    },
    {
      "epoch": 0.7673629439710429,
      "grad_norm": 0.4899980373642501,
      "learning_rate": 1.0743618663757602e-07,
      "loss": 1.5406,
      "step": 5088
    },
    {
      "epoch": 0.7675137621597165,
      "grad_norm": 0.30735719382349896,
      "learning_rate": 1.0736505391649762e-07,
      "loss": 1.5763,
      "step": 5089
    },
    {
      "epoch": 0.76766458034839,
      "grad_norm": 0.24252898756360913,
      "learning_rate": 1.0729395883434675e-07,
      "loss": 1.5664,
      "step": 5090
    },
    {
      "epoch": 0.7678153985370636,
      "grad_norm": 0.26556494556336524,
      "learning_rate": 1.0722290140708628e-07,
      "loss": 1.6136,
      "step": 5091
    },
    {
      "epoch": 0.7679662167257372,
      "grad_norm": 0.25334258561603834,
      "learning_rate": 1.0715188165067066e-07,
      "loss": 1.6102,
      "step": 5092
    },
    {
      "epoch": 0.7681170349144106,
      "grad_norm": 0.25168076140626744,
      "learning_rate": 1.0708089958104576e-07,
      "loss": 1.5315,
      "step": 5093
    },
    {
      "epoch": 0.7682678531030842,
      "grad_norm": 0.24437736466101306,
      "learning_rate": 1.0700995521414938e-07,
      "loss": 1.5378,
      "step": 5094
    },
    {
      "epoch": 0.7684186712917578,
      "grad_norm": 0.2706173777643933,
      "learning_rate": 1.0693904856591043e-07,
      "loss": 1.5875,
      "step": 5095
    },
    {
      "epoch": 0.7685694894804314,
      "grad_norm": 0.26091900532545664,
      "learning_rate": 1.068681796522495e-07,
      "loss": 1.6139,
      "step": 5096
    },
    {
      "epoch": 0.7687203076691049,
      "grad_norm": 0.2668382964154832,
      "learning_rate": 1.0679734848907885e-07,
      "loss": 1.588,
      "step": 5097
    },
    {
      "epoch": 0.7688711258577785,
      "grad_norm": 0.24883145644711072,
      "learning_rate": 1.0672655509230205e-07,
      "loss": 1.4768,
      "step": 5098
    },
    {
      "epoch": 0.769021944046452,
      "grad_norm": 0.27166435863492955,
      "learning_rate": 1.0665579947781425e-07,
      "loss": 1.5197,
      "step": 5099
    },
    {
      "epoch": 0.7691727622351255,
      "grad_norm": 0.2772294693142326,
      "learning_rate": 1.0658508166150224e-07,
      "loss": 1.523,
      "step": 5100
    },
    {
      "epoch": 0.7693235804237991,
      "grad_norm": 0.26345014775586284,
      "learning_rate": 1.0651440165924419e-07,
      "loss": 1.6231,
      "step": 5101
    },
    {
      "epoch": 0.7694743986124727,
      "grad_norm": 0.2625428479171914,
      "learning_rate": 1.0644375948690975e-07,
      "loss": 1.5537,
      "step": 5102
    },
    {
      "epoch": 0.7696252168011463,
      "grad_norm": 0.24545151495665093,
      "learning_rate": 1.0637315516036027e-07,
      "loss": 1.6497,
      "step": 5103
    },
    {
      "epoch": 0.7697760349898197,
      "grad_norm": 0.29316282299242524,
      "learning_rate": 1.063025886954484e-07,
      "loss": 1.591,
      "step": 5104
    },
    {
      "epoch": 0.7699268531784933,
      "grad_norm": 0.2483624076119689,
      "learning_rate": 1.0623206010801837e-07,
      "loss": 1.552,
      "step": 5105
    },
    {
      "epoch": 0.7700776713671669,
      "grad_norm": 0.25246809965954947,
      "learning_rate": 1.0616156941390586e-07,
      "loss": 1.572,
      "step": 5106
    },
    {
      "epoch": 0.7702284895558404,
      "grad_norm": 0.24407033874182604,
      "learning_rate": 1.0609111662893822e-07,
      "loss": 1.5878,
      "step": 5107
    },
    {
      "epoch": 0.770379307744514,
      "grad_norm": 0.24774824490308997,
      "learning_rate": 1.0602070176893404e-07,
      "loss": 1.5668,
      "step": 5108
    },
    {
      "epoch": 0.7705301259331876,
      "grad_norm": 0.24146342902464263,
      "learning_rate": 1.059503248497035e-07,
      "loss": 1.5541,
      "step": 5109
    },
    {
      "epoch": 0.7706809441218611,
      "grad_norm": 0.23805835167014303,
      "learning_rate": 1.0587998588704834e-07,
      "loss": 1.495,
      "step": 5110
    },
    {
      "epoch": 0.7708317623105346,
      "grad_norm": 0.42185474556300606,
      "learning_rate": 1.0580968489676167e-07,
      "loss": 1.5745,
      "step": 5111
    },
    {
      "epoch": 0.7709825804992082,
      "grad_norm": 0.2408951418800673,
      "learning_rate": 1.0573942189462803e-07,
      "loss": 1.5365,
      "step": 5112
    },
    {
      "epoch": 0.7711333986878818,
      "grad_norm": 0.42443576706367386,
      "learning_rate": 1.0566919689642368e-07,
      "loss": 1.545,
      "step": 5113
    },
    {
      "epoch": 0.7712842168765554,
      "grad_norm": 0.26429294303783457,
      "learning_rate": 1.0559900991791606e-07,
      "loss": 1.4837,
      "step": 5114
    },
    {
      "epoch": 0.7714350350652288,
      "grad_norm": 0.2831171957305455,
      "learning_rate": 1.0552886097486416e-07,
      "loss": 1.5854,
      "step": 5115
    },
    {
      "epoch": 0.7715858532539024,
      "grad_norm": 0.25376435672152076,
      "learning_rate": 1.0545875008301861e-07,
      "loss": 1.5179,
      "step": 5116
    },
    {
      "epoch": 0.771736671442576,
      "grad_norm": 0.24585108407865713,
      "learning_rate": 1.0538867725812123e-07,
      "loss": 1.5157,
      "step": 5117
    },
    {
      "epoch": 0.7718874896312495,
      "grad_norm": 0.2667675570241291,
      "learning_rate": 1.0531864251590542e-07,
      "loss": 1.5122,
      "step": 5118
    },
    {
      "epoch": 0.7720383078199231,
      "grad_norm": 0.25634356723734913,
      "learning_rate": 1.0524864587209598e-07,
      "loss": 1.6761,
      "step": 5119
    },
    {
      "epoch": 0.7721891260085967,
      "grad_norm": 0.40152559169486846,
      "learning_rate": 1.051786873424093e-07,
      "loss": 1.6062,
      "step": 5120
    },
    {
      "epoch": 0.7723399441972701,
      "grad_norm": 0.5199439778760333,
      "learning_rate": 1.0510876694255308e-07,
      "loss": 1.5619,
      "step": 5121
    },
    {
      "epoch": 0.7724907623859437,
      "grad_norm": 0.2580699708718366,
      "learning_rate": 1.0503888468822639e-07,
      "loss": 1.5517,
      "step": 5122
    },
    {
      "epoch": 0.7726415805746173,
      "grad_norm": 0.2545315956830306,
      "learning_rate": 1.0496904059511994e-07,
      "loss": 1.5236,
      "step": 5123
    },
    {
      "epoch": 0.7727923987632909,
      "grad_norm": 0.24856695616944577,
      "learning_rate": 1.0489923467891576e-07,
      "loss": 1.5601,
      "step": 5124
    },
    {
      "epoch": 0.7729432169519644,
      "grad_norm": 0.25293373694071336,
      "learning_rate": 1.0482946695528719e-07,
      "loss": 1.5245,
      "step": 5125
    },
    {
      "epoch": 0.773094035140638,
      "grad_norm": 0.24443089481446287,
      "learning_rate": 1.0475973743989927e-07,
      "loss": 1.5907,
      "step": 5126
    },
    {
      "epoch": 0.7732448533293115,
      "grad_norm": 0.272129843089619,
      "learning_rate": 1.0469004614840823e-07,
      "loss": 1.5011,
      "step": 5127
    },
    {
      "epoch": 0.773395671517985,
      "grad_norm": 0.2619860150616335,
      "learning_rate": 1.0462039309646177e-07,
      "loss": 1.6184,
      "step": 5128
    },
    {
      "epoch": 0.7735464897066586,
      "grad_norm": 0.24823666256911225,
      "learning_rate": 1.0455077829969911e-07,
      "loss": 1.5534,
      "step": 5129
    },
    {
      "epoch": 0.7736973078953322,
      "grad_norm": 0.27543584201099786,
      "learning_rate": 1.0448120177375075e-07,
      "loss": 1.5553,
      "step": 5130
    },
    {
      "epoch": 0.7738481260840058,
      "grad_norm": 0.2454650406977061,
      "learning_rate": 1.0441166353423864e-07,
      "loss": 1.5141,
      "step": 5131
    },
    {
      "epoch": 0.7739989442726792,
      "grad_norm": 0.24311302458080847,
      "learning_rate": 1.0434216359677609e-07,
      "loss": 1.5771,
      "step": 5132
    },
    {
      "epoch": 0.7741497624613528,
      "grad_norm": 0.2510070298633273,
      "learning_rate": 1.0427270197696798e-07,
      "loss": 1.5942,
      "step": 5133
    },
    {
      "epoch": 0.7743005806500264,
      "grad_norm": 0.24700988295710286,
      "learning_rate": 1.0420327869041043e-07,
      "loss": 1.5162,
      "step": 5134
    },
    {
      "epoch": 0.7744513988387,
      "grad_norm": 0.9863476661457141,
      "learning_rate": 1.0413389375269089e-07,
      "loss": 1.5606,
      "step": 5135
    },
    {
      "epoch": 0.7746022170273735,
      "grad_norm": 0.2501003593389527,
      "learning_rate": 1.0406454717938842e-07,
      "loss": 1.5663,
      "step": 5136
    },
    {
      "epoch": 0.7747530352160471,
      "grad_norm": 0.2686564153867668,
      "learning_rate": 1.0399523898607329e-07,
      "loss": 1.5506,
      "step": 5137
    },
    {
      "epoch": 0.7749038534047206,
      "grad_norm": 0.2520650053807712,
      "learning_rate": 1.0392596918830716e-07,
      "loss": 1.5425,
      "step": 5138
    },
    {
      "epoch": 0.7750546715933941,
      "grad_norm": 0.25532459334127455,
      "learning_rate": 1.0385673780164326e-07,
      "loss": 1.5156,
      "step": 5139
    },
    {
      "epoch": 0.7752054897820677,
      "grad_norm": 0.25637021843688246,
      "learning_rate": 1.0378754484162597e-07,
      "loss": 1.5837,
      "step": 5140
    },
    {
      "epoch": 0.7753563079707413,
      "grad_norm": 0.2615259072232839,
      "learning_rate": 1.0371839032379104e-07,
      "loss": 1.5349,
      "step": 5141
    },
    {
      "epoch": 0.7755071261594149,
      "grad_norm": 0.24496714282901963,
      "learning_rate": 1.036492742636658e-07,
      "loss": 1.552,
      "step": 5142
    },
    {
      "epoch": 0.7756579443480883,
      "grad_norm": 0.2613104991310393,
      "learning_rate": 1.0358019667676879e-07,
      "loss": 1.5723,
      "step": 5143
    },
    {
      "epoch": 0.7758087625367619,
      "grad_norm": 0.2639473884716577,
      "learning_rate": 1.0351115757860992e-07,
      "loss": 1.4878,
      "step": 5144
    },
    {
      "epoch": 0.7759595807254355,
      "grad_norm": 0.2808735969070876,
      "learning_rate": 1.0344215698469044e-07,
      "loss": 1.6,
      "step": 5145
    },
    {
      "epoch": 0.776110398914109,
      "grad_norm": 0.2491555953757118,
      "learning_rate": 1.0337319491050307e-07,
      "loss": 1.5844,
      "step": 5146
    },
    {
      "epoch": 0.7762612171027826,
      "grad_norm": 0.25991931785708977,
      "learning_rate": 1.0330427137153175e-07,
      "loss": 1.5831,
      "step": 5147
    },
    {
      "epoch": 0.7764120352914562,
      "grad_norm": 0.2742991791455993,
      "learning_rate": 1.0323538638325183e-07,
      "loss": 1.5945,
      "step": 5148
    },
    {
      "epoch": 0.7765628534801297,
      "grad_norm": 0.25883039506044575,
      "learning_rate": 1.0316653996113003e-07,
      "loss": 1.5559,
      "step": 5149
    },
    {
      "epoch": 0.7767136716688032,
      "grad_norm": 0.2655681031376377,
      "learning_rate": 1.0309773212062436e-07,
      "loss": 1.5658,
      "step": 5150
    },
    {
      "epoch": 0.7768644898574768,
      "grad_norm": 0.300007684282825,
      "learning_rate": 1.030289628771841e-07,
      "loss": 1.5492,
      "step": 5151
    },
    {
      "epoch": 0.7770153080461504,
      "grad_norm": 0.25078131996972297,
      "learning_rate": 1.0296023224625012e-07,
      "loss": 1.561,
      "step": 5152
    },
    {
      "epoch": 0.777166126234824,
      "grad_norm": 0.2518685712142501,
      "learning_rate": 1.0289154024325433e-07,
      "loss": 1.5263,
      "step": 5153
    },
    {
      "epoch": 0.7773169444234975,
      "grad_norm": 0.2369554858026924,
      "learning_rate": 1.0282288688362012e-07,
      "loss": 1.5365,
      "step": 5154
    },
    {
      "epoch": 0.777467762612171,
      "grad_norm": 0.2523849726536303,
      "learning_rate": 1.0275427218276209e-07,
      "loss": 1.6622,
      "step": 5155
    },
    {
      "epoch": 0.7776185808008446,
      "grad_norm": 0.3445991104335428,
      "learning_rate": 1.0268569615608635e-07,
      "loss": 1.6526,
      "step": 5156
    },
    {
      "epoch": 0.7777693989895181,
      "grad_norm": 0.32051184261339966,
      "learning_rate": 1.0261715881899019e-07,
      "loss": 1.5817,
      "step": 5157
    },
    {
      "epoch": 0.7779202171781917,
      "grad_norm": 0.4438827106717608,
      "learning_rate": 1.0254866018686214e-07,
      "loss": 1.5408,
      "step": 5158
    },
    {
      "epoch": 0.7780710353668653,
      "grad_norm": 0.2504474562113185,
      "learning_rate": 1.0248020027508231e-07,
      "loss": 1.5919,
      "step": 5159
    },
    {
      "epoch": 0.7782218535555387,
      "grad_norm": 0.2487038606727896,
      "learning_rate": 1.024117790990218e-07,
      "loss": 1.6189,
      "step": 5160
    },
    {
      "epoch": 0.7783726717442123,
      "grad_norm": 0.24337959048412816,
      "learning_rate": 1.0234339667404318e-07,
      "loss": 1.6063,
      "step": 5161
    },
    {
      "epoch": 0.7785234899328859,
      "grad_norm": 0.2789888112152276,
      "learning_rate": 1.0227505301550037e-07,
      "loss": 1.4946,
      "step": 5162
    },
    {
      "epoch": 0.7786743081215595,
      "grad_norm": 0.3383774441602223,
      "learning_rate": 1.0220674813873847e-07,
      "loss": 1.5146,
      "step": 5163
    },
    {
      "epoch": 0.778825126310233,
      "grad_norm": 0.41575331098938056,
      "learning_rate": 1.0213848205909386e-07,
      "loss": 1.5281,
      "step": 5164
    },
    {
      "epoch": 0.7789759444989066,
      "grad_norm": 0.2470574744341523,
      "learning_rate": 1.0207025479189435e-07,
      "loss": 1.5721,
      "step": 5165
    },
    {
      "epoch": 0.7791267626875801,
      "grad_norm": 0.24746981011226615,
      "learning_rate": 1.0200206635245892e-07,
      "loss": 1.5174,
      "step": 5166
    },
    {
      "epoch": 0.7792775808762537,
      "grad_norm": 0.254621670888393,
      "learning_rate": 1.0193391675609786e-07,
      "loss": 1.6554,
      "step": 5167
    },
    {
      "epoch": 0.7794283990649272,
      "grad_norm": 0.2517442422822756,
      "learning_rate": 1.0186580601811265e-07,
      "loss": 1.5588,
      "step": 5168
    },
    {
      "epoch": 0.7795792172536008,
      "grad_norm": 0.2524935229273349,
      "learning_rate": 1.017977341537963e-07,
      "loss": 1.5498,
      "step": 5169
    },
    {
      "epoch": 0.7797300354422744,
      "grad_norm": 0.2637390863959784,
      "learning_rate": 1.0172970117843282e-07,
      "loss": 1.5733,
      "step": 5170
    },
    {
      "epoch": 0.779880853630948,
      "grad_norm": 0.28437078199905974,
      "learning_rate": 1.0166170710729755e-07,
      "loss": 1.5658,
      "step": 5171
    },
    {
      "epoch": 0.7800316718196214,
      "grad_norm": 0.2673622905711399,
      "learning_rate": 1.0159375195565729e-07,
      "loss": 1.5313,
      "step": 5172
    },
    {
      "epoch": 0.780182490008295,
      "grad_norm": 0.24615532045152091,
      "learning_rate": 1.0152583573876983e-07,
      "loss": 1.5598,
      "step": 5173
    },
    {
      "epoch": 0.7803333081969686,
      "grad_norm": 0.26648977590140616,
      "learning_rate": 1.0145795847188433e-07,
      "loss": 1.4859,
      "step": 5174
    },
    {
      "epoch": 0.7804841263856421,
      "grad_norm": 0.25841066994939277,
      "learning_rate": 1.0139012017024135e-07,
      "loss": 1.5512,
      "step": 5175
    },
    {
      "epoch": 0.7806349445743157,
      "grad_norm": 0.24448615213183508,
      "learning_rate": 1.0132232084907243e-07,
      "loss": 1.5373,
      "step": 5176
    },
    {
      "epoch": 0.7807857627629892,
      "grad_norm": 0.24528000389884158,
      "learning_rate": 1.0125456052360045e-07,
      "loss": 1.5653,
      "step": 5177
    },
    {
      "epoch": 0.7809365809516627,
      "grad_norm": 0.2474044667383555,
      "learning_rate": 1.0118683920903975e-07,
      "loss": 1.5745,
      "step": 5178
    },
    {
      "epoch": 0.7810873991403363,
      "grad_norm": 0.2761230065419814,
      "learning_rate": 1.0111915692059566e-07,
      "loss": 1.6278,
      "step": 5179
    },
    {
      "epoch": 0.7812382173290099,
      "grad_norm": 0.43992774047553906,
      "learning_rate": 1.010515136734648e-07,
      "loss": 1.5916,
      "step": 5180
    },
    {
      "epoch": 0.7813890355176835,
      "grad_norm": 0.24466744172378638,
      "learning_rate": 1.0098390948283499e-07,
      "loss": 1.5438,
      "step": 5181
    },
    {
      "epoch": 0.781539853706357,
      "grad_norm": 0.24309196970415786,
      "learning_rate": 1.0091634436388549e-07,
      "loss": 1.5997,
      "step": 5182
    },
    {
      "epoch": 0.7816906718950305,
      "grad_norm": 0.24513875337909635,
      "learning_rate": 1.0084881833178652e-07,
      "loss": 1.5485,
      "step": 5183
    },
    {
      "epoch": 0.7818414900837041,
      "grad_norm": 0.6544087327244111,
      "learning_rate": 1.0078133140169965e-07,
      "loss": 1.5465,
      "step": 5184
    },
    {
      "epoch": 0.7819923082723776,
      "grad_norm": 0.25270703405805567,
      "learning_rate": 1.0071388358877775e-07,
      "loss": 1.5103,
      "step": 5185
    },
    {
      "epoch": 0.7821431264610512,
      "grad_norm": 0.4938550184353543,
      "learning_rate": 1.0064647490816472e-07,
      "loss": 1.5426,
      "step": 5186
    },
    {
      "epoch": 0.7822939446497248,
      "grad_norm": 0.26403755528150696,
      "learning_rate": 1.0057910537499578e-07,
      "loss": 1.611,
      "step": 5187
    },
    {
      "epoch": 0.7824447628383983,
      "grad_norm": 0.24171431971485452,
      "learning_rate": 1.0051177500439742e-07,
      "loss": 1.588,
      "step": 5188
    },
    {
      "epoch": 0.7825955810270718,
      "grad_norm": 0.23879309098552787,
      "learning_rate": 1.0044448381148726e-07,
      "loss": 1.6279,
      "step": 5189
    },
    {
      "epoch": 0.7827463992157454,
      "grad_norm": 0.24518655059112635,
      "learning_rate": 1.0037723181137403e-07,
      "loss": 1.5294,
      "step": 5190
    },
    {
      "epoch": 0.782897217404419,
      "grad_norm": 0.48801982637795555,
      "learning_rate": 1.0031001901915789e-07,
      "loss": 1.5268,
      "step": 5191
    },
    {
      "epoch": 0.7830480355930926,
      "grad_norm": 0.246061195139339,
      "learning_rate": 1.0024284544993003e-07,
      "loss": 1.6452,
      "step": 5192
    },
    {
      "epoch": 0.7831988537817661,
      "grad_norm": 0.25484960572036536,
      "learning_rate": 1.0017571111877283e-07,
      "loss": 1.4964,
      "step": 5193
    },
    {
      "epoch": 0.7833496719704396,
      "grad_norm": 0.2641759867426765,
      "learning_rate": 1.0010861604075987e-07,
      "loss": 1.5249,
      "step": 5194
    },
    {
      "epoch": 0.7835004901591132,
      "grad_norm": 0.2749696841708133,
      "learning_rate": 1.0004156023095611e-07,
      "loss": 1.5734,
      "step": 5195
    },
    {
      "epoch": 0.7836513083477867,
      "grad_norm": 0.2400846391621469,
      "learning_rate": 9.99745437044174e-08,
      "loss": 1.5363,
      "step": 5196
    },
    {
      "epoch": 0.7838021265364603,
      "grad_norm": 0.2503792022857372,
      "learning_rate": 9.99075664761909e-08,
      "loss": 1.5607,
      "step": 5197
    },
    {
      "epoch": 0.7839529447251339,
      "grad_norm": 0.25126265300389844,
      "learning_rate": 9.984062856131502e-08,
      "loss": 1.5441,
      "step": 5198
    },
    {
      "epoch": 0.7841037629138075,
      "grad_norm": 0.2677246673958432,
      "learning_rate": 9.977372997481922e-08,
      "loss": 1.5171,
      "step": 5199
    },
    {
      "epoch": 0.7842545811024809,
      "grad_norm": 0.27186577521057703,
      "learning_rate": 9.970687073172415e-08,
      "loss": 1.5189,
      "step": 5200
    },
    {
      "epoch": 0.7844053992911545,
      "grad_norm": 0.24286649058717943,
      "learning_rate": 9.964005084704177e-08,
      "loss": 1.5378,
      "step": 5201
    },
    {
      "epoch": 0.7845562174798281,
      "grad_norm": 0.25296634340355056,
      "learning_rate": 9.957327033577498e-08,
      "loss": 1.5879,
      "step": 5202
    },
    {
      "epoch": 0.7847070356685016,
      "grad_norm": 0.24729615374562958,
      "learning_rate": 9.950652921291796e-08,
      "loss": 1.5808,
      "step": 5203
    },
    {
      "epoch": 0.7848578538571752,
      "grad_norm": 0.2759538028414474,
      "learning_rate": 9.943982749345615e-08,
      "loss": 1.5551,
      "step": 5204
    },
    {
      "epoch": 0.7850086720458487,
      "grad_norm": 0.24625468461101757,
      "learning_rate": 9.937316519236592e-08,
      "loss": 1.5504,
      "step": 5205
    },
    {
      "epoch": 0.7851594902345223,
      "grad_norm": 0.2551362944408527,
      "learning_rate": 9.93065423246149e-08,
      "loss": 1.6395,
      "step": 5206
    },
    {
      "epoch": 0.7853103084231958,
      "grad_norm": 0.24286043771088153,
      "learning_rate": 9.923995890516184e-08,
      "loss": 1.545,
      "step": 5207
    },
    {
      "epoch": 0.7854611266118694,
      "grad_norm": 0.24377166566239084,
      "learning_rate": 9.917341494895678e-08,
      "loss": 1.6084,
      "step": 5208
    },
    {
      "epoch": 0.785611944800543,
      "grad_norm": 0.24603192824167494,
      "learning_rate": 9.910691047094072e-08,
      "loss": 1.564,
      "step": 5209
    },
    {
      "epoch": 0.7857627629892165,
      "grad_norm": 0.2426345984172691,
      "learning_rate": 9.904044548604573e-08,
      "loss": 1.4926,
      "step": 5210
    },
    {
      "epoch": 0.78591358117789,
      "grad_norm": 0.25474503066062437,
      "learning_rate": 9.897402000919533e-08,
      "loss": 1.5594,
      "step": 5211
    },
    {
      "epoch": 0.7860643993665636,
      "grad_norm": 0.24394729119339043,
      "learning_rate": 9.890763405530386e-08,
      "loss": 1.5512,
      "step": 5212
    },
    {
      "epoch": 0.7862152175552372,
      "grad_norm": 0.2517485169788096,
      "learning_rate": 9.884128763927688e-08,
      "loss": 1.5152,
      "step": 5213
    },
    {
      "epoch": 0.7863660357439107,
      "grad_norm": 0.24075771254911466,
      "learning_rate": 9.877498077601117e-08,
      "loss": 1.5828,
      "step": 5214
    },
    {
      "epoch": 0.7865168539325843,
      "grad_norm": 0.2576568077914992,
      "learning_rate": 9.870871348039452e-08,
      "loss": 1.5435,
      "step": 5215
    },
    {
      "epoch": 0.7866676721212579,
      "grad_norm": 0.2488922033748577,
      "learning_rate": 9.864248576730583e-08,
      "loss": 1.5554,
      "step": 5216
    },
    {
      "epoch": 0.7868184903099313,
      "grad_norm": 0.24442321649992887,
      "learning_rate": 9.857629765161521e-08,
      "loss": 1.5341,
      "step": 5217
    },
    {
      "epoch": 0.7869693084986049,
      "grad_norm": 0.268373426064988,
      "learning_rate": 9.851014914818382e-08,
      "loss": 1.5983,
      "step": 5218
    },
    {
      "epoch": 0.7871201266872785,
      "grad_norm": 0.27454531743987626,
      "learning_rate": 9.844404027186387e-08,
      "loss": 1.616,
      "step": 5219
    },
    {
      "epoch": 0.7872709448759521,
      "grad_norm": 0.24735204104979536,
      "learning_rate": 9.837797103749871e-08,
      "loss": 1.6034,
      "step": 5220
    },
    {
      "epoch": 0.7874217630646256,
      "grad_norm": 0.24328973411274415,
      "learning_rate": 9.831194145992291e-08,
      "loss": 1.5682,
      "step": 5221
    },
    {
      "epoch": 0.7875725812532991,
      "grad_norm": 0.2849190394836823,
      "learning_rate": 9.824595155396198e-08,
      "loss": 1.6179,
      "step": 5222
    },
    {
      "epoch": 0.7877233994419727,
      "grad_norm": 0.2868891873378707,
      "learning_rate": 9.818000133443252e-08,
      "loss": 1.514,
      "step": 5223
    },
    {
      "epoch": 0.7878742176306462,
      "grad_norm": 0.29915894733704507,
      "learning_rate": 9.811409081614238e-08,
      "loss": 1.5292,
      "step": 5224
    },
    {
      "epoch": 0.7880250358193198,
      "grad_norm": 0.25681452067867805,
      "learning_rate": 9.804822001389034e-08,
      "loss": 1.5641,
      "step": 5225
    },
    {
      "epoch": 0.7881758540079934,
      "grad_norm": 0.5143721419834408,
      "learning_rate": 9.798238894246627e-08,
      "loss": 1.5118,
      "step": 5226
    },
    {
      "epoch": 0.788326672196667,
      "grad_norm": 0.24450340290190584,
      "learning_rate": 9.791659761665125e-08,
      "loss": 1.5542,
      "step": 5227
    },
    {
      "epoch": 0.7884774903853404,
      "grad_norm": 0.25923111975771784,
      "learning_rate": 9.785084605121731e-08,
      "loss": 1.553,
      "step": 5228
    },
    {
      "epoch": 0.788628308574014,
      "grad_norm": 0.576592704674624,
      "learning_rate": 9.778513426092749e-08,
      "loss": 1.5407,
      "step": 5229
    },
    {
      "epoch": 0.7887791267626876,
      "grad_norm": 0.25600492943228165,
      "learning_rate": 9.77194622605362e-08,
      "loss": 1.5689,
      "step": 5230
    },
    {
      "epoch": 0.7889299449513612,
      "grad_norm": 0.24986631977565338,
      "learning_rate": 9.765383006478861e-08,
      "loss": 1.5635,
      "step": 5231
    },
    {
      "epoch": 0.7890807631400347,
      "grad_norm": 0.2553981532136114,
      "learning_rate": 9.758823768842106e-08,
      "loss": 1.5453,
      "step": 5232
    },
    {
      "epoch": 0.7892315813287082,
      "grad_norm": 0.2693895420664751,
      "learning_rate": 9.752268514616086e-08,
      "loss": 1.5,
      "step": 5233
    },
    {
      "epoch": 0.7893823995173818,
      "grad_norm": 0.24512872950012005,
      "learning_rate": 9.745717245272664e-08,
      "loss": 1.5462,
      "step": 5234
    },
    {
      "epoch": 0.7895332177060553,
      "grad_norm": 0.29255385539188583,
      "learning_rate": 9.739169962282781e-08,
      "loss": 1.6196,
      "step": 5235
    },
    {
      "epoch": 0.7896840358947289,
      "grad_norm": 0.24636829891140796,
      "learning_rate": 9.732626667116489e-08,
      "loss": 1.5105,
      "step": 5236
    },
    {
      "epoch": 0.7898348540834025,
      "grad_norm": 0.24649437419168208,
      "learning_rate": 9.72608736124296e-08,
      "loss": 1.523,
      "step": 5237
    },
    {
      "epoch": 0.7899856722720761,
      "grad_norm": 0.24873170942575037,
      "learning_rate": 9.719552046130452e-08,
      "loss": 1.6408,
      "step": 5238
    },
    {
      "epoch": 0.7901364904607495,
      "grad_norm": 0.2548158490497884,
      "learning_rate": 9.713020723246324e-08,
      "loss": 1.567,
      "step": 5239
    },
    {
      "epoch": 0.7902873086494231,
      "grad_norm": 0.42845738937801925,
      "learning_rate": 9.706493394057067e-08,
      "loss": 1.6048,
      "step": 5240
    },
    {
      "epoch": 0.7904381268380967,
      "grad_norm": 0.2844146144024818,
      "learning_rate": 9.699970060028247e-08,
      "loss": 1.6515,
      "step": 5241
    },
    {
      "epoch": 0.7905889450267702,
      "grad_norm": 0.24872106490086418,
      "learning_rate": 9.693450722624542e-08,
      "loss": 1.6053,
      "step": 5242
    },
    {
      "epoch": 0.7907397632154438,
      "grad_norm": 0.32871423293171437,
      "learning_rate": 9.68693538330973e-08,
      "loss": 1.5577,
      "step": 5243
    },
    {
      "epoch": 0.7908905814041174,
      "grad_norm": 0.2728321330895966,
      "learning_rate": 9.6804240435467e-08,
      "loss": 1.5897,
      "step": 5244
    },
    {
      "epoch": 0.7910413995927909,
      "grad_norm": 0.24331676661144996,
      "learning_rate": 9.673916704797438e-08,
      "loss": 1.5672,
      "step": 5245
    },
    {
      "epoch": 0.7911922177814644,
      "grad_norm": 0.26678977606867443,
      "learning_rate": 9.667413368523023e-08,
      "loss": 1.596,
      "step": 5246
    },
    {
      "epoch": 0.791343035970138,
      "grad_norm": 0.31409689319400996,
      "learning_rate": 9.660914036183653e-08,
      "loss": 1.5556,
      "step": 5247
    },
    {
      "epoch": 0.7914938541588116,
      "grad_norm": 0.25646123946223787,
      "learning_rate": 9.654418709238619e-08,
      "loss": 1.4908,
      "step": 5248
    },
    {
      "epoch": 0.7916446723474851,
      "grad_norm": 0.2567826501041337,
      "learning_rate": 9.647927389146295e-08,
      "loss": 1.6256,
      "step": 5249
    },
    {
      "epoch": 0.7917954905361586,
      "grad_norm": 0.34114059546512304,
      "learning_rate": 9.641440077364188e-08,
      "loss": 1.51,
      "step": 5250
    },
    {
      "epoch": 0.7919463087248322,
      "grad_norm": 0.2836721666035514,
      "learning_rate": 9.634956775348883e-08,
      "loss": 1.5525,
      "step": 5251
    },
    {
      "epoch": 0.7920971269135058,
      "grad_norm": 0.23733357097386582,
      "learning_rate": 9.628477484556065e-08,
      "loss": 1.484,
      "step": 5252
    },
    {
      "epoch": 0.7922479451021793,
      "grad_norm": 0.25173384900472523,
      "learning_rate": 9.622002206440533e-08,
      "loss": 1.5501,
      "step": 5253
    },
    {
      "epoch": 0.7923987632908529,
      "grad_norm": 0.3801763344859971,
      "learning_rate": 9.615530942456168e-08,
      "loss": 1.517,
      "step": 5254
    },
    {
      "epoch": 0.7925495814795265,
      "grad_norm": 0.24662474824068642,
      "learning_rate": 9.609063694055964e-08,
      "loss": 1.5852,
      "step": 5255
    },
    {
      "epoch": 0.7927003996681999,
      "grad_norm": 0.26165667966294054,
      "learning_rate": 9.602600462691991e-08,
      "loss": 1.5105,
      "step": 5256
    },
    {
      "epoch": 0.7928512178568735,
      "grad_norm": 0.2962893428088458,
      "learning_rate": 9.596141249815455e-08,
      "loss": 1.528,
      "step": 5257
    },
    {
      "epoch": 0.7930020360455471,
      "grad_norm": 0.2650559965583865,
      "learning_rate": 9.589686056876624e-08,
      "loss": 1.5849,
      "step": 5258
    },
    {
      "epoch": 0.7931528542342207,
      "grad_norm": 0.2716594700912555,
      "learning_rate": 9.583234885324873e-08,
      "loss": 1.5872,
      "step": 5259
    },
    {
      "epoch": 0.7933036724228942,
      "grad_norm": 0.2619807567043884,
      "learning_rate": 9.576787736608693e-08,
      "loss": 1.5347,
      "step": 5260
    },
    {
      "epoch": 0.7934544906115678,
      "grad_norm": 0.2643445440132794,
      "learning_rate": 9.570344612175643e-08,
      "loss": 1.5998,
      "step": 5261
    },
    {
      "epoch": 0.7936053088002413,
      "grad_norm": 0.25171253202602445,
      "learning_rate": 9.563905513472395e-08,
      "loss": 1.5692,
      "step": 5262
    },
    {
      "epoch": 0.7937561269889148,
      "grad_norm": 0.27166046322517085,
      "learning_rate": 9.55747044194472e-08,
      "loss": 1.5505,
      "step": 5263
    },
    {
      "epoch": 0.7939069451775884,
      "grad_norm": 0.2535333741114826,
      "learning_rate": 9.551039399037475e-08,
      "loss": 1.5318,
      "step": 5264
    },
    {
      "epoch": 0.794057763366262,
      "grad_norm": 0.2507836488559641,
      "learning_rate": 9.544612386194612e-08,
      "loss": 1.5812,
      "step": 5265
    },
    {
      "epoch": 0.7942085815549356,
      "grad_norm": 0.2539476403745853,
      "learning_rate": 9.538189404859192e-08,
      "loss": 1.4597,
      "step": 5266
    },
    {
      "epoch": 0.794359399743609,
      "grad_norm": 0.2592222527971778,
      "learning_rate": 9.531770456473355e-08,
      "loss": 1.5503,
      "step": 5267
    },
    {
      "epoch": 0.7945102179322826,
      "grad_norm": 0.2598003965601039,
      "learning_rate": 9.525355542478348e-08,
      "loss": 1.5817,
      "step": 5268
    },
    {
      "epoch": 0.7946610361209562,
      "grad_norm": 0.24709229152520368,
      "learning_rate": 9.51894466431449e-08,
      "loss": 1.5441,
      "step": 5269
    },
    {
      "epoch": 0.7948118543096298,
      "grad_norm": 0.26921306649993715,
      "learning_rate": 9.51253782342123e-08,
      "loss": 1.5136,
      "step": 5270
    },
    {
      "epoch": 0.7949626724983033,
      "grad_norm": 0.26097673845803504,
      "learning_rate": 9.50613502123708e-08,
      "loss": 1.5609,
      "step": 5271
    },
    {
      "epoch": 0.7951134906869769,
      "grad_norm": 0.26364841943998557,
      "learning_rate": 9.49973625919965e-08,
      "loss": 1.5982,
      "step": 5272
    },
    {
      "epoch": 0.7952643088756504,
      "grad_norm": 0.28794281341300015,
      "learning_rate": 9.493341538745664e-08,
      "loss": 1.573,
      "step": 5273
    },
    {
      "epoch": 0.7954151270643239,
      "grad_norm": 0.33487655215897294,
      "learning_rate": 9.48695086131091e-08,
      "loss": 1.5709,
      "step": 5274
    },
    {
      "epoch": 0.7955659452529975,
      "grad_norm": 0.24130890877284242,
      "learning_rate": 9.480564228330283e-08,
      "loss": 1.5723,
      "step": 5275
    },
    {
      "epoch": 0.7957167634416711,
      "grad_norm": 0.2625366907215809,
      "learning_rate": 9.474181641237778e-08,
      "loss": 1.5325,
      "step": 5276
    },
    {
      "epoch": 0.7958675816303447,
      "grad_norm": 0.26232962210393057,
      "learning_rate": 9.467803101466462e-08,
      "loss": 1.5742,
      "step": 5277
    },
    {
      "epoch": 0.7960183998190181,
      "grad_norm": 0.23504685894343988,
      "learning_rate": 9.461428610448502e-08,
      "loss": 1.5213,
      "step": 5278
    },
    {
      "epoch": 0.7961692180076917,
      "grad_norm": 0.26532676636442065,
      "learning_rate": 9.455058169615163e-08,
      "loss": 1.5344,
      "step": 5279
    },
    {
      "epoch": 0.7963200361963653,
      "grad_norm": 0.24510599024024202,
      "learning_rate": 9.448691780396792e-08,
      "loss": 1.5386,
      "step": 5280
    },
    {
      "epoch": 0.7964708543850388,
      "grad_norm": 1.031700399816234,
      "learning_rate": 9.44232944422283e-08,
      "loss": 1.6055,
      "step": 5281
    },
    {
      "epoch": 0.7966216725737124,
      "grad_norm": 0.26191947447633585,
      "learning_rate": 9.4359711625218e-08,
      "loss": 1.5396,
      "step": 5282
    },
    {
      "epoch": 0.796772490762386,
      "grad_norm": 0.25671463940734085,
      "learning_rate": 9.42961693672133e-08,
      "loss": 1.5745,
      "step": 5283
    },
    {
      "epoch": 0.7969233089510595,
      "grad_norm": 0.24064938910739106,
      "learning_rate": 9.42326676824812e-08,
      "loss": 1.5457,
      "step": 5284
    },
    {
      "epoch": 0.797074127139733,
      "grad_norm": 0.3121141778522017,
      "learning_rate": 9.416920658527983e-08,
      "loss": 1.5594,
      "step": 5285
    },
    {
      "epoch": 0.7972249453284066,
      "grad_norm": 0.24925526232670184,
      "learning_rate": 9.410578608985786e-08,
      "loss": 1.5715,
      "step": 5286
    },
    {
      "epoch": 0.7973757635170802,
      "grad_norm": 0.2635196049687862,
      "learning_rate": 9.40424062104552e-08,
      "loss": 1.5411,
      "step": 5287
    },
    {
      "epoch": 0.7975265817057537,
      "grad_norm": 0.27385968980861314,
      "learning_rate": 9.397906696130243e-08,
      "loss": 1.5748,
      "step": 5288
    },
    {
      "epoch": 0.7976773998944273,
      "grad_norm": 0.24713953186058749,
      "learning_rate": 9.391576835662096e-08,
      "loss": 1.5194,
      "step": 5289
    },
    {
      "epoch": 0.7978282180831008,
      "grad_norm": 0.30874219514176704,
      "learning_rate": 9.38525104106233e-08,
      "loss": 1.5026,
      "step": 5290
    },
    {
      "epoch": 0.7979790362717744,
      "grad_norm": 0.25061358837709624,
      "learning_rate": 9.378929313751266e-08,
      "loss": 1.5311,
      "step": 5291
    },
    {
      "epoch": 0.7981298544604479,
      "grad_norm": 0.24458375046614347,
      "learning_rate": 9.37261165514831e-08,
      "loss": 1.5807,
      "step": 5292
    },
    {
      "epoch": 0.7982806726491215,
      "grad_norm": 0.23696780363506526,
      "learning_rate": 9.366298066671973e-08,
      "loss": 1.6163,
      "step": 5293
    },
    {
      "epoch": 0.7984314908377951,
      "grad_norm": 0.26478478952353585,
      "learning_rate": 9.359988549739828e-08,
      "loss": 1.5109,
      "step": 5294
    },
    {
      "epoch": 0.7985823090264685,
      "grad_norm": 0.2501202826056413,
      "learning_rate": 9.353683105768544e-08,
      "loss": 1.5546,
      "step": 5295
    },
    {
      "epoch": 0.7987331272151421,
      "grad_norm": 0.2445865178732147,
      "learning_rate": 9.34738173617389e-08,
      "loss": 1.5804,
      "step": 5296
    },
    {
      "epoch": 0.7988839454038157,
      "grad_norm": 0.2883543741249625,
      "learning_rate": 9.341084442370697e-08,
      "loss": 1.5302,
      "step": 5297
    },
    {
      "epoch": 0.7990347635924893,
      "grad_norm": 0.25099696195827753,
      "learning_rate": 9.334791225772893e-08,
      "loss": 1.556,
      "step": 5298
    },
    {
      "epoch": 0.7991855817811628,
      "grad_norm": 0.23455636692257625,
      "learning_rate": 9.328502087793482e-08,
      "loss": 1.5932,
      "step": 5299
    },
    {
      "epoch": 0.7993363999698364,
      "grad_norm": 0.2728960714104501,
      "learning_rate": 9.322217029844573e-08,
      "loss": 1.577,
      "step": 5300
    },
    {
      "epoch": 0.7994872181585099,
      "grad_norm": 0.27709587109866557,
      "learning_rate": 9.315936053337333e-08,
      "loss": 1.5744,
      "step": 5301
    },
    {
      "epoch": 0.7996380363471834,
      "grad_norm": 0.25922070011737874,
      "learning_rate": 9.309659159682024e-08,
      "loss": 1.5334,
      "step": 5302
    },
    {
      "epoch": 0.799788854535857,
      "grad_norm": 0.2508166462275185,
      "learning_rate": 9.303386350288e-08,
      "loss": 1.5497,
      "step": 5303
    },
    {
      "epoch": 0.7999396727245306,
      "grad_norm": 0.25757627776221476,
      "learning_rate": 9.297117626563686e-08,
      "loss": 1.56,
      "step": 5304
    },
    {
      "epoch": 0.8000904909132042,
      "grad_norm": 0.25218239326517317,
      "learning_rate": 9.290852989916581e-08,
      "loss": 1.5865,
      "step": 5305
    },
    {
      "epoch": 0.8002413091018777,
      "grad_norm": 0.25401844710881205,
      "learning_rate": 9.284592441753298e-08,
      "loss": 1.5299,
      "step": 5306
    },
    {
      "epoch": 0.8003921272905512,
      "grad_norm": 0.2811756303548548,
      "learning_rate": 9.278335983479504e-08,
      "loss": 1.5726,
      "step": 5307
    },
    {
      "epoch": 0.8005429454792248,
      "grad_norm": 0.24685673296726654,
      "learning_rate": 9.272083616499948e-08,
      "loss": 1.5447,
      "step": 5308
    },
    {
      "epoch": 0.8006937636678984,
      "grad_norm": 0.2541167362442535,
      "learning_rate": 9.265835342218481e-08,
      "loss": 1.6192,
      "step": 5309
    },
    {
      "epoch": 0.8008445818565719,
      "grad_norm": 0.2541763550342351,
      "learning_rate": 9.259591162038019e-08,
      "loss": 1.5799,
      "step": 5310
    },
    {
      "epoch": 0.8009954000452455,
      "grad_norm": 0.24062327291007268,
      "learning_rate": 9.253351077360558e-08,
      "loss": 1.5336,
      "step": 5311
    },
    {
      "epoch": 0.801146218233919,
      "grad_norm": 0.24377484811412073,
      "learning_rate": 9.247115089587176e-08,
      "loss": 1.6561,
      "step": 5312
    },
    {
      "epoch": 0.8012970364225925,
      "grad_norm": 0.2619299920309576,
      "learning_rate": 9.240883200118046e-08,
      "loss": 1.5988,
      "step": 5313
    },
    {
      "epoch": 0.8014478546112661,
      "grad_norm": 0.25311637076284305,
      "learning_rate": 9.234655410352403e-08,
      "loss": 1.5791,
      "step": 5314
    },
    {
      "epoch": 0.8015986727999397,
      "grad_norm": 0.24374721927243836,
      "learning_rate": 9.228431721688557e-08,
      "loss": 1.5806,
      "step": 5315
    },
    {
      "epoch": 0.8017494909886133,
      "grad_norm": 0.2423027621176886,
      "learning_rate": 9.222212135523924e-08,
      "loss": 1.5073,
      "step": 5316
    },
    {
      "epoch": 0.8019003091772868,
      "grad_norm": 0.24484754116073318,
      "learning_rate": 9.215996653254977e-08,
      "loss": 1.5388,
      "step": 5317
    },
    {
      "epoch": 0.8020511273659603,
      "grad_norm": 0.2615928740002736,
      "learning_rate": 9.20978527627726e-08,
      "loss": 1.5687,
      "step": 5318
    },
    {
      "epoch": 0.8022019455546339,
      "grad_norm": 0.2516112458899224,
      "learning_rate": 9.203578005985427e-08,
      "loss": 1.5286,
      "step": 5319
    },
    {
      "epoch": 0.8023527637433074,
      "grad_norm": 0.25795245742735634,
      "learning_rate": 9.197374843773182e-08,
      "loss": 1.5351,
      "step": 5320
    },
    {
      "epoch": 0.802503581931981,
      "grad_norm": 0.2481199157870698,
      "learning_rate": 9.191175791033308e-08,
      "loss": 1.5578,
      "step": 5321
    },
    {
      "epoch": 0.8026544001206546,
      "grad_norm": 0.26970641849821136,
      "learning_rate": 9.184980849157689e-08,
      "loss": 1.6133,
      "step": 5322
    },
    {
      "epoch": 0.802805218309328,
      "grad_norm": 0.40884427230147374,
      "learning_rate": 9.17879001953726e-08,
      "loss": 1.5045,
      "step": 5323
    },
    {
      "epoch": 0.8029560364980016,
      "grad_norm": 0.26584043367342286,
      "learning_rate": 9.172603303562046e-08,
      "loss": 1.596,
      "step": 5324
    },
    {
      "epoch": 0.8031068546866752,
      "grad_norm": 0.24749377355422092,
      "learning_rate": 9.166420702621136e-08,
      "loss": 1.6246,
      "step": 5325
    },
    {
      "epoch": 0.8032576728753488,
      "grad_norm": 0.26131795595872453,
      "learning_rate": 9.160242218102718e-08,
      "loss": 1.5267,
      "step": 5326
    },
    {
      "epoch": 0.8034084910640223,
      "grad_norm": 0.24781848229229342,
      "learning_rate": 9.154067851394035e-08,
      "loss": 1.5093,
      "step": 5327
    },
    {
      "epoch": 0.8035593092526959,
      "grad_norm": 0.25944684942346824,
      "learning_rate": 9.147897603881405e-08,
      "loss": 1.566,
      "step": 5328
    },
    {
      "epoch": 0.8037101274413694,
      "grad_norm": 0.2621925392436018,
      "learning_rate": 9.14173147695024e-08,
      "loss": 1.6605,
      "step": 5329
    },
    {
      "epoch": 0.803860945630043,
      "grad_norm": 0.301692620618676,
      "learning_rate": 9.135569471985013e-08,
      "loss": 1.593,
      "step": 5330
    },
    {
      "epoch": 0.8040117638187165,
      "grad_norm": 0.25014830888623185,
      "learning_rate": 9.129411590369263e-08,
      "loss": 1.5825,
      "step": 5331
    },
    {
      "epoch": 0.8041625820073901,
      "grad_norm": 0.25920278714820694,
      "learning_rate": 9.123257833485629e-08,
      "loss": 1.5638,
      "step": 5332
    },
    {
      "epoch": 0.8043134001960637,
      "grad_norm": 0.25065550082065197,
      "learning_rate": 9.117108202715803e-08,
      "loss": 1.5359,
      "step": 5333
    },
    {
      "epoch": 0.8044642183847373,
      "grad_norm": 0.24765080141677795,
      "learning_rate": 9.110962699440544e-08,
      "loss": 1.4746,
      "step": 5334
    },
    {
      "epoch": 0.8046150365734107,
      "grad_norm": 0.38934375374357866,
      "learning_rate": 9.104821325039715e-08,
      "loss": 1.5905,
      "step": 5335
    },
    {
      "epoch": 0.8047658547620843,
      "grad_norm": 0.24887350237507042,
      "learning_rate": 9.098684080892226e-08,
      "loss": 1.5848,
      "step": 5336
    },
    {
      "epoch": 0.8049166729507579,
      "grad_norm": 0.250743742241653,
      "learning_rate": 9.092550968376067e-08,
      "loss": 1.5651,
      "step": 5337
    },
    {
      "epoch": 0.8050674911394314,
      "grad_norm": 0.24632833456968029,
      "learning_rate": 9.08642198886829e-08,
      "loss": 1.5344,
      "step": 5338
    },
    {
      "epoch": 0.805218309328105,
      "grad_norm": 0.25992383642761396,
      "learning_rate": 9.080297143745049e-08,
      "loss": 1.5933,
      "step": 5339
    },
    {
      "epoch": 0.8053691275167785,
      "grad_norm": 0.259916730791826,
      "learning_rate": 9.07417643438154e-08,
      "loss": 1.5681,
      "step": 5340
    },
    {
      "epoch": 0.805519945705452,
      "grad_norm": 0.23859363293643113,
      "learning_rate": 9.068059862152031e-08,
      "loss": 1.5128,
      "step": 5341
    },
    {
      "epoch": 0.8056707638941256,
      "grad_norm": 0.24195583611150917,
      "learning_rate": 9.061947428429889e-08,
      "loss": 1.5135,
      "step": 5342
    },
    {
      "epoch": 0.8058215820827992,
      "grad_norm": 0.2431272382246692,
      "learning_rate": 9.055839134587527e-08,
      "loss": 1.6044,
      "step": 5343
    },
    {
      "epoch": 0.8059724002714728,
      "grad_norm": 0.25546031392204177,
      "learning_rate": 9.049734981996424e-08,
      "loss": 1.5352,
      "step": 5344
    },
    {
      "epoch": 0.8061232184601463,
      "grad_norm": 0.2712017381244022,
      "learning_rate": 9.043634972027155e-08,
      "loss": 1.5402,
      "step": 5345
    },
    {
      "epoch": 0.8062740366488198,
      "grad_norm": 0.274957572626681,
      "learning_rate": 9.037539106049345e-08,
      "loss": 1.5007,
      "step": 5346
    },
    {
      "epoch": 0.8064248548374934,
      "grad_norm": 0.25596132221287465,
      "learning_rate": 9.031447385431684e-08,
      "loss": 1.5955,
      "step": 5347
    },
    {
      "epoch": 0.806575673026167,
      "grad_norm": 0.24419676172022808,
      "learning_rate": 9.025359811541955e-08,
      "loss": 1.5236,
      "step": 5348
    },
    {
      "epoch": 0.8067264912148405,
      "grad_norm": 0.24905555635542445,
      "learning_rate": 9.01927638574699e-08,
      "loss": 1.5338,
      "step": 5349
    },
    {
      "epoch": 0.8068773094035141,
      "grad_norm": 0.2656314969267587,
      "learning_rate": 9.013197109412696e-08,
      "loss": 1.5723,
      "step": 5350
    },
    {
      "epoch": 0.8070281275921877,
      "grad_norm": 0.2623964491270413,
      "learning_rate": 9.007121983904037e-08,
      "loss": 1.5422,
      "step": 5351
    },
    {
      "epoch": 0.8071789457808611,
      "grad_norm": 0.25206837273975874,
      "learning_rate": 9.00105101058507e-08,
      "loss": 1.5458,
      "step": 5352
    },
    {
      "epoch": 0.8073297639695347,
      "grad_norm": 0.2492787955366975,
      "learning_rate": 8.994984190818903e-08,
      "loss": 1.5401,
      "step": 5353
    },
    {
      "epoch": 0.8074805821582083,
      "grad_norm": 0.28513932967276034,
      "learning_rate": 8.988921525967705e-08,
      "loss": 1.5206,
      "step": 5354
    },
    {
      "epoch": 0.8076314003468819,
      "grad_norm": 0.2584154267957481,
      "learning_rate": 8.982863017392732e-08,
      "loss": 1.5066,
      "step": 5355
    },
    {
      "epoch": 0.8077822185355554,
      "grad_norm": 0.254232521760263,
      "learning_rate": 8.976808666454291e-08,
      "loss": 1.6354,
      "step": 5356
    },
    {
      "epoch": 0.8079330367242289,
      "grad_norm": 0.2475693806936419,
      "learning_rate": 8.970758474511752e-08,
      "loss": 1.571,
      "step": 5357
    },
    {
      "epoch": 0.8080838549129025,
      "grad_norm": 0.25953367558525964,
      "learning_rate": 8.96471244292357e-08,
      "loss": 1.5508,
      "step": 5358
    },
    {
      "epoch": 0.808234673101576,
      "grad_norm": 0.5908046759790779,
      "learning_rate": 8.958670573047255e-08,
      "loss": 1.5292,
      "step": 5359
    },
    {
      "epoch": 0.8083854912902496,
      "grad_norm": 0.2551127997023796,
      "learning_rate": 8.952632866239374e-08,
      "loss": 1.5858,
      "step": 5360
    },
    {
      "epoch": 0.8085363094789232,
      "grad_norm": 0.252857184965345,
      "learning_rate": 8.946599323855578e-08,
      "loss": 1.6471,
      "step": 5361
    },
    {
      "epoch": 0.8086871276675968,
      "grad_norm": 0.2781140975758444,
      "learning_rate": 8.94056994725057e-08,
      "loss": 1.5221,
      "step": 5362
    },
    {
      "epoch": 0.8088379458562702,
      "grad_norm": 0.2465808512536548,
      "learning_rate": 8.934544737778118e-08,
      "loss": 1.6323,
      "step": 5363
    },
    {
      "epoch": 0.8089887640449438,
      "grad_norm": 0.2573413992597441,
      "learning_rate": 8.928523696791055e-08,
      "loss": 1.5326,
      "step": 5364
    },
    {
      "epoch": 0.8091395822336174,
      "grad_norm": 0.2462046847547708,
      "learning_rate": 8.922506825641285e-08,
      "loss": 1.6586,
      "step": 5365
    },
    {
      "epoch": 0.809290400422291,
      "grad_norm": 0.25037190609185495,
      "learning_rate": 8.916494125679772e-08,
      "loss": 1.5154,
      "step": 5366
    },
    {
      "epoch": 0.8094412186109645,
      "grad_norm": 0.25180805274015566,
      "learning_rate": 8.910485598256532e-08,
      "loss": 1.5549,
      "step": 5367
    },
    {
      "epoch": 0.809592036799638,
      "grad_norm": 0.25981895801806054,
      "learning_rate": 8.904481244720668e-08,
      "loss": 1.5517,
      "step": 5368
    },
    {
      "epoch": 0.8097428549883116,
      "grad_norm": 0.2631646512048799,
      "learning_rate": 8.898481066420324e-08,
      "loss": 1.6065,
      "step": 5369
    },
    {
      "epoch": 0.8098936731769851,
      "grad_norm": 0.24932562170508332,
      "learning_rate": 8.892485064702712e-08,
      "loss": 1.5323,
      "step": 5370
    },
    {
      "epoch": 0.8100444913656587,
      "grad_norm": 0.24436695030970323,
      "learning_rate": 8.88649324091412e-08,
      "loss": 1.5663,
      "step": 5371
    },
    {
      "epoch": 0.8101953095543323,
      "grad_norm": 0.3269133308214128,
      "learning_rate": 8.880505596399882e-08,
      "loss": 1.508,
      "step": 5372
    },
    {
      "epoch": 0.8103461277430059,
      "grad_norm": 0.24294520604001518,
      "learning_rate": 8.874522132504394e-08,
      "loss": 1.5495,
      "step": 5373
    },
    {
      "epoch": 0.8104969459316793,
      "grad_norm": 0.24278171864311007,
      "learning_rate": 8.868542850571121e-08,
      "loss": 1.5866,
      "step": 5374
    },
    {
      "epoch": 0.8106477641203529,
      "grad_norm": 0.4172245566715417,
      "learning_rate": 8.862567751942585e-08,
      "loss": 1.6476,
      "step": 5375
    },
    {
      "epoch": 0.8107985823090265,
      "grad_norm": 0.24977377337571738,
      "learning_rate": 8.856596837960376e-08,
      "loss": 1.633,
      "step": 5376
    },
    {
      "epoch": 0.8109494004977,
      "grad_norm": 0.29329456679490906,
      "learning_rate": 8.850630109965125e-08,
      "loss": 1.4838,
      "step": 5377
    },
    {
      "epoch": 0.8111002186863736,
      "grad_norm": 0.25501453774994753,
      "learning_rate": 8.844667569296549e-08,
      "loss": 1.6073,
      "step": 5378
    },
    {
      "epoch": 0.8112510368750472,
      "grad_norm": 0.24424479931122867,
      "learning_rate": 8.838709217293407e-08,
      "loss": 1.4627,
      "step": 5379
    },
    {
      "epoch": 0.8114018550637206,
      "grad_norm": 0.260366118063312,
      "learning_rate": 8.832755055293517e-08,
      "loss": 1.5528,
      "step": 5380
    },
    {
      "epoch": 0.8115526732523942,
      "grad_norm": 0.4415329037429282,
      "learning_rate": 8.826805084633773e-08,
      "loss": 1.5864,
      "step": 5381
    },
    {
      "epoch": 0.8117034914410678,
      "grad_norm": 0.2507284828558614,
      "learning_rate": 8.820859306650114e-08,
      "loss": 1.5795,
      "step": 5382
    },
    {
      "epoch": 0.8118543096297414,
      "grad_norm": 0.26354637795152236,
      "learning_rate": 8.814917722677528e-08,
      "loss": 1.5162,
      "step": 5383
    },
    {
      "epoch": 0.8120051278184149,
      "grad_norm": 0.25882985694211996,
      "learning_rate": 8.80898033405009e-08,
      "loss": 1.501,
      "step": 5384
    },
    {
      "epoch": 0.8121559460070884,
      "grad_norm": 0.24968693494284327,
      "learning_rate": 8.80304714210091e-08,
      "loss": 1.5276,
      "step": 5385
    },
    {
      "epoch": 0.812306764195762,
      "grad_norm": 0.255317334408465,
      "learning_rate": 8.79711814816216e-08,
      "loss": 1.5747,
      "step": 5386
    },
    {
      "epoch": 0.8124575823844356,
      "grad_norm": 0.24498829617791465,
      "learning_rate": 8.791193353565071e-08,
      "loss": 1.5387,
      "step": 5387
    },
    {
      "epoch": 0.8126084005731091,
      "grad_norm": 0.2503163183131022,
      "learning_rate": 8.785272759639941e-08,
      "loss": 1.481,
      "step": 5388
    },
    {
      "epoch": 0.8127592187617827,
      "grad_norm": 0.25392590128550246,
      "learning_rate": 8.779356367716106e-08,
      "loss": 1.576,
      "step": 5389
    },
    {
      "epoch": 0.8129100369504563,
      "grad_norm": 0.2693098078403543,
      "learning_rate": 8.773444179121969e-08,
      "loss": 1.5928,
      "step": 5390
    },
    {
      "epoch": 0.8130608551391297,
      "grad_norm": 0.2658132062565851,
      "learning_rate": 8.767536195184996e-08,
      "loss": 1.6161,
      "step": 5391
    },
    {
      "epoch": 0.8132116733278033,
      "grad_norm": 0.25606448372362856,
      "learning_rate": 8.761632417231696e-08,
      "loss": 1.5879,
      "step": 5392
    },
    {
      "epoch": 0.8133624915164769,
      "grad_norm": 0.2674645289633079,
      "learning_rate": 8.755732846587636e-08,
      "loss": 1.5681,
      "step": 5393
    },
    {
      "epoch": 0.8135133097051505,
      "grad_norm": 0.2383322629257858,
      "learning_rate": 8.74983748457745e-08,
      "loss": 1.5701,
      "step": 5394
    },
    {
      "epoch": 0.813664127893824,
      "grad_norm": 0.24718613518887608,
      "learning_rate": 8.743946332524812e-08,
      "loss": 1.6227,
      "step": 5395
    },
    {
      "epoch": 0.8138149460824976,
      "grad_norm": 0.2548462219820224,
      "learning_rate": 8.738059391752456e-08,
      "loss": 1.5131,
      "step": 5396
    },
    {
      "epoch": 0.8139657642711711,
      "grad_norm": 0.2727091785242275,
      "learning_rate": 8.732176663582182e-08,
      "loss": 1.6031,
      "step": 5397
    },
    {
      "epoch": 0.8141165824598446,
      "grad_norm": 0.24118012755987908,
      "learning_rate": 8.726298149334824e-08,
      "loss": 1.5522,
      "step": 5398
    },
    {
      "epoch": 0.8142674006485182,
      "grad_norm": 0.24933804716672786,
      "learning_rate": 8.72042385033028e-08,
      "loss": 1.5314,
      "step": 5399
    },
    {
      "epoch": 0.8144182188371918,
      "grad_norm": 0.25911084966975867,
      "learning_rate": 8.7145537678875e-08,
      "loss": 1.5313,
      "step": 5400
    },
    {
      "epoch": 0.8145690370258654,
      "grad_norm": 0.3206244188727088,
      "learning_rate": 8.708687903324499e-08,
      "loss": 1.5206,
      "step": 5401
    },
    {
      "epoch": 0.8147198552145388,
      "grad_norm": 0.272195332433252,
      "learning_rate": 8.702826257958325e-08,
      "loss": 1.567,
      "step": 5402
    },
    {
      "epoch": 0.8148706734032124,
      "grad_norm": 0.2790391207210687,
      "learning_rate": 8.696968833105083e-08,
      "loss": 1.46,
      "step": 5403
    },
    {
      "epoch": 0.815021491591886,
      "grad_norm": 0.2394075464882588,
      "learning_rate": 8.691115630079953e-08,
      "loss": 1.5442,
      "step": 5404
    },
    {
      "epoch": 0.8151723097805595,
      "grad_norm": 0.2676690166921416,
      "learning_rate": 8.685266650197136e-08,
      "loss": 1.5634,
      "step": 5405
    },
    {
      "epoch": 0.8153231279692331,
      "grad_norm": 0.25896659649800485,
      "learning_rate": 8.679421894769897e-08,
      "loss": 1.597,
      "step": 5406
    },
    {
      "epoch": 0.8154739461579067,
      "grad_norm": 0.38741568974714197,
      "learning_rate": 8.673581365110564e-08,
      "loss": 1.5574,
      "step": 5407
    },
    {
      "epoch": 0.8156247643465802,
      "grad_norm": 0.2508110110315982,
      "learning_rate": 8.667745062530502e-08,
      "loss": 1.5147,
      "step": 5408
    },
    {
      "epoch": 0.8157755825352537,
      "grad_norm": 0.261320500360099,
      "learning_rate": 8.66191298834012e-08,
      "loss": 1.5707,
      "step": 5409
    },
    {
      "epoch": 0.8159264007239273,
      "grad_norm": 0.32222062350335506,
      "learning_rate": 8.656085143848907e-08,
      "loss": 1.5306,
      "step": 5410
    },
    {
      "epoch": 0.8160772189126009,
      "grad_norm": 0.2517801262747225,
      "learning_rate": 8.650261530365372e-08,
      "loss": 1.5784,
      "step": 5411
    },
    {
      "epoch": 0.8162280371012745,
      "grad_norm": 0.24409824743779973,
      "learning_rate": 8.644442149197089e-08,
      "loss": 1.5261,
      "step": 5412
    },
    {
      "epoch": 0.8163788552899479,
      "grad_norm": 0.24812057004238666,
      "learning_rate": 8.638627001650676e-08,
      "loss": 1.5896,
      "step": 5413
    },
    {
      "epoch": 0.8165296734786215,
      "grad_norm": 0.2474975475339188,
      "learning_rate": 8.632816089031808e-08,
      "loss": 1.5342,
      "step": 5414
    },
    {
      "epoch": 0.8166804916672951,
      "grad_norm": 0.27549008020970944,
      "learning_rate": 8.627009412645197e-08,
      "loss": 1.5721,
      "step": 5415
    },
    {
      "epoch": 0.8168313098559686,
      "grad_norm": 0.256658953687918,
      "learning_rate": 8.621206973794617e-08,
      "loss": 1.5522,
      "step": 5416
    },
    {
      "epoch": 0.8169821280446422,
      "grad_norm": 0.2613980832353188,
      "learning_rate": 8.615408773782884e-08,
      "loss": 1.5271,
      "step": 5417
    },
    {
      "epoch": 0.8171329462333158,
      "grad_norm": 0.24113470747515664,
      "learning_rate": 8.609614813911861e-08,
      "loss": 1.5702,
      "step": 5418
    },
    {
      "epoch": 0.8172837644219892,
      "grad_norm": 0.25148116486031524,
      "learning_rate": 8.603825095482459e-08,
      "loss": 1.6049,
      "step": 5419
    },
    {
      "epoch": 0.8174345826106628,
      "grad_norm": 0.33292332289003057,
      "learning_rate": 8.59803961979464e-08,
      "loss": 1.6006,
      "step": 5420
    },
    {
      "epoch": 0.8175854007993364,
      "grad_norm": 0.2546474181958,
      "learning_rate": 8.592258388147416e-08,
      "loss": 1.565,
      "step": 5421
    },
    {
      "epoch": 0.81773621898801,
      "grad_norm": 0.2444825543043335,
      "learning_rate": 8.586481401838833e-08,
      "loss": 1.5486,
      "step": 5422
    },
    {
      "epoch": 0.8178870371766835,
      "grad_norm": 0.2912897847245712,
      "learning_rate": 8.580708662166004e-08,
      "loss": 1.5549,
      "step": 5423
    },
    {
      "epoch": 0.8180378553653571,
      "grad_norm": 0.2460550010359619,
      "learning_rate": 8.57494017042507e-08,
      "loss": 1.5158,
      "step": 5424
    },
    {
      "epoch": 0.8181886735540306,
      "grad_norm": 0.24223485160905156,
      "learning_rate": 8.569175927911227e-08,
      "loss": 1.5291,
      "step": 5425
    },
    {
      "epoch": 0.8183394917427042,
      "grad_norm": 0.2703071094741545,
      "learning_rate": 8.563415935918708e-08,
      "loss": 1.5467,
      "step": 5426
    },
    {
      "epoch": 0.8184903099313777,
      "grad_norm": 0.2583706245350999,
      "learning_rate": 8.55766019574081e-08,
      "loss": 1.5992,
      "step": 5427
    },
    {
      "epoch": 0.8186411281200513,
      "grad_norm": 0.33994776617608374,
      "learning_rate": 8.551908708669864e-08,
      "loss": 1.5422,
      "step": 5428
    },
    {
      "epoch": 0.8187919463087249,
      "grad_norm": 0.26498457659972996,
      "learning_rate": 8.546161475997232e-08,
      "loss": 1.5438,
      "step": 5429
    },
    {
      "epoch": 0.8189427644973983,
      "grad_norm": 0.24547134501992465,
      "learning_rate": 8.540418499013353e-08,
      "loss": 1.4488,
      "step": 5430
    },
    {
      "epoch": 0.8190935826860719,
      "grad_norm": 0.24827873309112944,
      "learning_rate": 8.534679779007686e-08,
      "loss": 1.5305,
      "step": 5431
    },
    {
      "epoch": 0.8192444008747455,
      "grad_norm": 0.24132056654808884,
      "learning_rate": 8.528945317268735e-08,
      "loss": 1.5605,
      "step": 5432
    },
    {
      "epoch": 0.8193952190634191,
      "grad_norm": 0.2485427899850471,
      "learning_rate": 8.523215115084062e-08,
      "loss": 1.6074,
      "step": 5433
    },
    {
      "epoch": 0.8195460372520926,
      "grad_norm": 0.2631974283968051,
      "learning_rate": 8.51748917374026e-08,
      "loss": 1.6224,
      "step": 5434
    },
    {
      "epoch": 0.8196968554407662,
      "grad_norm": 0.24912974465370194,
      "learning_rate": 8.511767494522968e-08,
      "loss": 1.5186,
      "step": 5435
    },
    {
      "epoch": 0.8198476736294397,
      "grad_norm": 0.2737340974688945,
      "learning_rate": 8.506050078716876e-08,
      "loss": 1.6187,
      "step": 5436
    },
    {
      "epoch": 0.8199984918181132,
      "grad_norm": 0.25737401441478086,
      "learning_rate": 8.500336927605706e-08,
      "loss": 1.5347,
      "step": 5437
    },
    {
      "epoch": 0.8201493100067868,
      "grad_norm": 0.2560828165673062,
      "learning_rate": 8.494628042472223e-08,
      "loss": 1.5867,
      "step": 5438
    },
    {
      "epoch": 0.8203001281954604,
      "grad_norm": 0.25727026274899956,
      "learning_rate": 8.488923424598243e-08,
      "loss": 1.5816,
      "step": 5439
    },
    {
      "epoch": 0.820450946384134,
      "grad_norm": 0.2565306185789074,
      "learning_rate": 8.483223075264622e-08,
      "loss": 1.5893,
      "step": 5440
    },
    {
      "epoch": 0.8206017645728075,
      "grad_norm": 0.2537942028175509,
      "learning_rate": 8.477526995751247e-08,
      "loss": 1.5945,
      "step": 5441
    },
    {
      "epoch": 0.820752582761481,
      "grad_norm": 0.2733803871603261,
      "learning_rate": 8.471835187337053e-08,
      "loss": 1.5049,
      "step": 5442
    },
    {
      "epoch": 0.8209034009501546,
      "grad_norm": 0.2534012718871278,
      "learning_rate": 8.466147651300029e-08,
      "loss": 1.5302,
      "step": 5443
    },
    {
      "epoch": 0.8210542191388281,
      "grad_norm": 0.2573195107412062,
      "learning_rate": 8.460464388917184e-08,
      "loss": 1.5285,
      "step": 5444
    },
    {
      "epoch": 0.8212050373275017,
      "grad_norm": 0.2494917233132808,
      "learning_rate": 8.454785401464567e-08,
      "loss": 1.5708,
      "step": 5445
    },
    {
      "epoch": 0.8213558555161753,
      "grad_norm": 0.2738697235871858,
      "learning_rate": 8.449110690217295e-08,
      "loss": 1.5621,
      "step": 5446
    },
    {
      "epoch": 0.8215066737048488,
      "grad_norm": 0.2437305077525022,
      "learning_rate": 8.443440256449493e-08,
      "loss": 1.556,
      "step": 5447
    },
    {
      "epoch": 0.8216574918935223,
      "grad_norm": 0.29219678686720424,
      "learning_rate": 8.437774101434338e-08,
      "loss": 1.5577,
      "step": 5448
    },
    {
      "epoch": 0.8218083100821959,
      "grad_norm": 0.25668764384134174,
      "learning_rate": 8.432112226444057e-08,
      "loss": 1.5769,
      "step": 5449
    },
    {
      "epoch": 0.8219591282708695,
      "grad_norm": 0.2556210092097604,
      "learning_rate": 8.426454632749896e-08,
      "loss": 1.5504,
      "step": 5450
    },
    {
      "epoch": 0.822109946459543,
      "grad_norm": 0.24933253322043045,
      "learning_rate": 8.420801321622157e-08,
      "loss": 1.506,
      "step": 5451
    },
    {
      "epoch": 0.8222607646482166,
      "grad_norm": 0.2635338886111469,
      "learning_rate": 8.415152294330161e-08,
      "loss": 1.5611,
      "step": 5452
    },
    {
      "epoch": 0.8224115828368901,
      "grad_norm": 0.25933779608687785,
      "learning_rate": 8.409507552142293e-08,
      "loss": 1.5929,
      "step": 5453
    },
    {
      "epoch": 0.8225624010255637,
      "grad_norm": 0.2639221458810809,
      "learning_rate": 8.403867096325956e-08,
      "loss": 1.5985,
      "step": 5454
    },
    {
      "epoch": 0.8227132192142372,
      "grad_norm": 0.2561121934952759,
      "learning_rate": 8.398230928147593e-08,
      "loss": 1.5388,
      "step": 5455
    },
    {
      "epoch": 0.8228640374029108,
      "grad_norm": 0.24240913723860447,
      "learning_rate": 8.392599048872697e-08,
      "loss": 1.5786,
      "step": 5456
    },
    {
      "epoch": 0.8230148555915844,
      "grad_norm": 0.2876358324857365,
      "learning_rate": 8.386971459765781e-08,
      "loss": 1.5611,
      "step": 5457
    },
    {
      "epoch": 0.8231656737802578,
      "grad_norm": 0.24344122254880732,
      "learning_rate": 8.3813481620904e-08,
      "loss": 1.5714,
      "step": 5458
    },
    {
      "epoch": 0.8233164919689314,
      "grad_norm": 0.2714474288976156,
      "learning_rate": 8.375729157109159e-08,
      "loss": 1.5763,
      "step": 5459
    },
    {
      "epoch": 0.823467310157605,
      "grad_norm": 0.25141151167514675,
      "learning_rate": 8.370114446083685e-08,
      "loss": 1.5913,
      "step": 5460
    },
    {
      "epoch": 0.8236181283462786,
      "grad_norm": 0.24774573760534288,
      "learning_rate": 8.364504030274636e-08,
      "loss": 1.598,
      "step": 5461
    },
    {
      "epoch": 0.8237689465349521,
      "grad_norm": 0.24693739103143914,
      "learning_rate": 8.358897910941723e-08,
      "loss": 1.5434,
      "step": 5462
    },
    {
      "epoch": 0.8239197647236257,
      "grad_norm": 0.24586716731845726,
      "learning_rate": 8.35329608934368e-08,
      "loss": 1.5677,
      "step": 5463
    },
    {
      "epoch": 0.8240705829122992,
      "grad_norm": 0.24333422423598058,
      "learning_rate": 8.347698566738278e-08,
      "loss": 1.5386,
      "step": 5464
    },
    {
      "epoch": 0.8242214011009728,
      "grad_norm": 0.2705858912166686,
      "learning_rate": 8.342105344382319e-08,
      "loss": 1.5513,
      "step": 5465
    },
    {
      "epoch": 0.8243722192896463,
      "grad_norm": 0.2513852238797157,
      "learning_rate": 8.336516423531658e-08,
      "loss": 1.5458,
      "step": 5466
    },
    {
      "epoch": 0.8245230374783199,
      "grad_norm": 0.2677073243030713,
      "learning_rate": 8.330931805441161e-08,
      "loss": 1.484,
      "step": 5467
    },
    {
      "epoch": 0.8246738556669935,
      "grad_norm": 0.2417387343568687,
      "learning_rate": 8.32535149136473e-08,
      "loss": 1.5808,
      "step": 5468
    },
    {
      "epoch": 0.824824673855667,
      "grad_norm": 0.2564645935056465,
      "learning_rate": 8.319775482555325e-08,
      "loss": 1.6054,
      "step": 5469
    },
    {
      "epoch": 0.8249754920443405,
      "grad_norm": 0.26104144470292767,
      "learning_rate": 8.314203780264914e-08,
      "loss": 1.5334,
      "step": 5470
    },
    {
      "epoch": 0.8251263102330141,
      "grad_norm": 0.23933997322284464,
      "learning_rate": 8.3086363857445e-08,
      "loss": 1.5773,
      "step": 5471
    },
    {
      "epoch": 0.8252771284216877,
      "grad_norm": 0.24138886614256383,
      "learning_rate": 8.303073300244137e-08,
      "loss": 1.535,
      "step": 5472
    },
    {
      "epoch": 0.8254279466103612,
      "grad_norm": 0.24743779693958193,
      "learning_rate": 8.297514525012895e-08,
      "loss": 1.5456,
      "step": 5473
    },
    {
      "epoch": 0.8255787647990348,
      "grad_norm": 0.24615796285010186,
      "learning_rate": 8.29196006129888e-08,
      "loss": 1.5427,
      "step": 5474
    },
    {
      "epoch": 0.8257295829877083,
      "grad_norm": 0.2704201999162556,
      "learning_rate": 8.286409910349227e-08,
      "loss": 1.4741,
      "step": 5475
    },
    {
      "epoch": 0.8258804011763818,
      "grad_norm": 0.24226687705850347,
      "learning_rate": 8.280864073410116e-08,
      "loss": 1.5244,
      "step": 5476
    },
    {
      "epoch": 0.8260312193650554,
      "grad_norm": 0.2678719068712771,
      "learning_rate": 8.27532255172674e-08,
      "loss": 1.5282,
      "step": 5477
    },
    {
      "epoch": 0.826182037553729,
      "grad_norm": 0.2636572663143235,
      "learning_rate": 8.269785346543334e-08,
      "loss": 1.5513,
      "step": 5478
    },
    {
      "epoch": 0.8263328557424026,
      "grad_norm": 0.2798055508355367,
      "learning_rate": 8.264252459103169e-08,
      "loss": 1.5618,
      "step": 5479
    },
    {
      "epoch": 0.8264836739310761,
      "grad_norm": 0.6178532566448571,
      "learning_rate": 8.258723890648533e-08,
      "loss": 1.6157,
      "step": 5480
    },
    {
      "epoch": 0.8266344921197496,
      "grad_norm": 0.29986391638280213,
      "learning_rate": 8.253199642420747e-08,
      "loss": 1.5464,
      "step": 5481
    },
    {
      "epoch": 0.8267853103084232,
      "grad_norm": 0.2520571795820321,
      "learning_rate": 8.247679715660174e-08,
      "loss": 1.6092,
      "step": 5482
    },
    {
      "epoch": 0.8269361284970967,
      "grad_norm": 0.2367118617638586,
      "learning_rate": 8.242164111606193e-08,
      "loss": 1.5416,
      "step": 5483
    },
    {
      "epoch": 0.8270869466857703,
      "grad_norm": 0.2719291940920188,
      "learning_rate": 8.236652831497215e-08,
      "loss": 1.5367,
      "step": 5484
    },
    {
      "epoch": 0.8272377648744439,
      "grad_norm": 0.2502373723967542,
      "learning_rate": 8.231145876570691e-08,
      "loss": 1.5677,
      "step": 5485
    },
    {
      "epoch": 0.8273885830631175,
      "grad_norm": 0.2789187183183438,
      "learning_rate": 8.225643248063091e-08,
      "loss": 1.5479,
      "step": 5486
    },
    {
      "epoch": 0.8275394012517909,
      "grad_norm": 0.27432617296751016,
      "learning_rate": 8.220144947209914e-08,
      "loss": 1.5756,
      "step": 5487
    },
    {
      "epoch": 0.8276902194404645,
      "grad_norm": 0.5782504197737236,
      "learning_rate": 8.21465097524568e-08,
      "loss": 1.5519,
      "step": 5488
    },
    {
      "epoch": 0.8278410376291381,
      "grad_norm": 0.2427849386158717,
      "learning_rate": 8.209161333403961e-08,
      "loss": 1.5212,
      "step": 5489
    },
    {
      "epoch": 0.8279918558178117,
      "grad_norm": 0.25720121590749473,
      "learning_rate": 8.203676022917333e-08,
      "loss": 1.5388,
      "step": 5490
    },
    {
      "epoch": 0.8281426740064852,
      "grad_norm": 0.2641600883680885,
      "learning_rate": 8.198195045017403e-08,
      "loss": 1.5655,
      "step": 5491
    },
    {
      "epoch": 0.8282934921951587,
      "grad_norm": 0.27795188615700056,
      "learning_rate": 8.192718400934817e-08,
      "loss": 1.59,
      "step": 5492
    },
    {
      "epoch": 0.8284443103838323,
      "grad_norm": 0.24300716325259156,
      "learning_rate": 8.187246091899244e-08,
      "loss": 1.5495,
      "step": 5493
    },
    {
      "epoch": 0.8285951285725058,
      "grad_norm": 0.25179431711121714,
      "learning_rate": 8.181778119139376e-08,
      "loss": 1.6051,
      "step": 5494
    },
    {
      "epoch": 0.8287459467611794,
      "grad_norm": 0.24904010718627334,
      "learning_rate": 8.17631448388292e-08,
      "loss": 1.4861,
      "step": 5495
    },
    {
      "epoch": 0.828896764949853,
      "grad_norm": 0.2540518015165979,
      "learning_rate": 8.170855187356637e-08,
      "loss": 1.5089,
      "step": 5496
    },
    {
      "epoch": 0.8290475831385266,
      "grad_norm": 0.2665402508650307,
      "learning_rate": 8.165400230786292e-08,
      "loss": 1.5303,
      "step": 5497
    },
    {
      "epoch": 0.8291984013272,
      "grad_norm": 0.2436131353265455,
      "learning_rate": 8.159949615396677e-08,
      "loss": 1.5759,
      "step": 5498
    },
    {
      "epoch": 0.8293492195158736,
      "grad_norm": 0.24836390355686813,
      "learning_rate": 8.154503342411619e-08,
      "loss": 1.5564,
      "step": 5499
    },
    {
      "epoch": 0.8295000377045472,
      "grad_norm": 0.25474851828650724,
      "learning_rate": 8.149061413053966e-08,
      "loss": 1.5884,
      "step": 5500
    },
    {
      "epoch": 0.8296508558932207,
      "grad_norm": 0.2624267415973674,
      "learning_rate": 8.143623828545581e-08,
      "loss": 1.5885,
      "step": 5501
    },
    {
      "epoch": 0.8298016740818943,
      "grad_norm": 0.28913872445950517,
      "learning_rate": 8.138190590107375e-08,
      "loss": 1.5947,
      "step": 5502
    },
    {
      "epoch": 0.8299524922705678,
      "grad_norm": 0.26690457060983375,
      "learning_rate": 8.132761698959258e-08,
      "loss": 1.5308,
      "step": 5503
    },
    {
      "epoch": 0.8301033104592414,
      "grad_norm": 0.2511696551494377,
      "learning_rate": 8.127337156320177e-08,
      "loss": 1.514,
      "step": 5504
    },
    {
      "epoch": 0.8302541286479149,
      "grad_norm": 0.2466855227551212,
      "learning_rate": 8.121916963408092e-08,
      "loss": 1.5415,
      "step": 5505
    },
    {
      "epoch": 0.8304049468365885,
      "grad_norm": 0.24274934564025172,
      "learning_rate": 8.116501121440005e-08,
      "loss": 1.6399,
      "step": 5506
    },
    {
      "epoch": 0.8305557650252621,
      "grad_norm": 0.2729618232546073,
      "learning_rate": 8.111089631631929e-08,
      "loss": 1.6211,
      "step": 5507
    },
    {
      "epoch": 0.8307065832139356,
      "grad_norm": 0.24271182317106732,
      "learning_rate": 8.105682495198891e-08,
      "loss": 1.5797,
      "step": 5508
    },
    {
      "epoch": 0.8308574014026091,
      "grad_norm": 0.2507970848649864,
      "learning_rate": 8.100279713354963e-08,
      "loss": 1.4944,
      "step": 5509
    },
    {
      "epoch": 0.8310082195912827,
      "grad_norm": 0.2404971843812375,
      "learning_rate": 8.094881287313221e-08,
      "loss": 1.5518,
      "step": 5510
    },
    {
      "epoch": 0.8311590377799563,
      "grad_norm": 0.3189830776593624,
      "learning_rate": 8.089487218285767e-08,
      "loss": 1.5306,
      "step": 5511
    },
    {
      "epoch": 0.8313098559686298,
      "grad_norm": 0.2390377412502148,
      "learning_rate": 8.084097507483732e-08,
      "loss": 1.5635,
      "step": 5512
    },
    {
      "epoch": 0.8314606741573034,
      "grad_norm": 0.24737509670879096,
      "learning_rate": 8.078712156117261e-08,
      "loss": 1.5703,
      "step": 5513
    },
    {
      "epoch": 0.831611492345977,
      "grad_norm": 0.24042296074711592,
      "learning_rate": 8.073331165395516e-08,
      "loss": 1.5698,
      "step": 5514
    },
    {
      "epoch": 0.8317623105346504,
      "grad_norm": 0.2469180979004506,
      "learning_rate": 8.067954536526697e-08,
      "loss": 1.5306,
      "step": 5515
    },
    {
      "epoch": 0.831913128723324,
      "grad_norm": 0.25633531814682303,
      "learning_rate": 8.062582270718008e-08,
      "loss": 1.5689,
      "step": 5516
    },
    {
      "epoch": 0.8320639469119976,
      "grad_norm": 0.381908193777433,
      "learning_rate": 8.057214369175679e-08,
      "loss": 1.5383,
      "step": 5517
    },
    {
      "epoch": 0.8322147651006712,
      "grad_norm": 0.33798264426474456,
      "learning_rate": 8.051850833104955e-08,
      "loss": 1.6076,
      "step": 5518
    },
    {
      "epoch": 0.8323655832893447,
      "grad_norm": 0.25391239904314006,
      "learning_rate": 8.046491663710119e-08,
      "loss": 1.5714,
      "step": 5519
    },
    {
      "epoch": 0.8325164014780182,
      "grad_norm": 0.28153495556832586,
      "learning_rate": 8.04113686219445e-08,
      "loss": 1.5612,
      "step": 5520
    },
    {
      "epoch": 0.8326672196666918,
      "grad_norm": 0.256479463704896,
      "learning_rate": 8.035786429760258e-08,
      "loss": 1.5384,
      "step": 5521
    },
    {
      "epoch": 0.8328180378553653,
      "grad_norm": 0.26785797989635485,
      "learning_rate": 8.030440367608876e-08,
      "loss": 1.541,
      "step": 5522
    },
    {
      "epoch": 0.8329688560440389,
      "grad_norm": 1.2850846020371895,
      "learning_rate": 8.025098676940649e-08,
      "loss": 1.5482,
      "step": 5523
    },
    {
      "epoch": 0.8331196742327125,
      "grad_norm": 0.28770264232268766,
      "learning_rate": 8.019761358954936e-08,
      "loss": 1.5556,
      "step": 5524
    },
    {
      "epoch": 0.8332704924213861,
      "grad_norm": 0.316290337019335,
      "learning_rate": 8.014428414850129e-08,
      "loss": 1.5643,
      "step": 5525
    },
    {
      "epoch": 0.8334213106100595,
      "grad_norm": 0.2534859342404115,
      "learning_rate": 8.009099845823628e-08,
      "loss": 1.5442,
      "step": 5526
    },
    {
      "epoch": 0.8335721287987331,
      "grad_norm": 0.24181255104096053,
      "learning_rate": 8.003775653071841e-08,
      "loss": 1.5488,
      "step": 5527
    },
    {
      "epoch": 0.8337229469874067,
      "grad_norm": 0.25788127231317026,
      "learning_rate": 7.99845583779022e-08,
      "loss": 1.6367,
      "step": 5528
    },
    {
      "epoch": 0.8338737651760803,
      "grad_norm": 0.2612586270195239,
      "learning_rate": 7.993140401173211e-08,
      "loss": 1.5347,
      "step": 5529
    },
    {
      "epoch": 0.8340245833647538,
      "grad_norm": 0.3201414132852245,
      "learning_rate": 7.987829344414284e-08,
      "loss": 1.5919,
      "step": 5530
    },
    {
      "epoch": 0.8341754015534274,
      "grad_norm": 0.24560379475832575,
      "learning_rate": 7.982522668705922e-08,
      "loss": 1.5847,
      "step": 5531
    },
    {
      "epoch": 0.8343262197421009,
      "grad_norm": 0.2479258474056162,
      "learning_rate": 7.977220375239635e-08,
      "loss": 1.5666,
      "step": 5532
    },
    {
      "epoch": 0.8344770379307744,
      "grad_norm": 0.3420062562165439,
      "learning_rate": 7.971922465205944e-08,
      "loss": 1.5402,
      "step": 5533
    },
    {
      "epoch": 0.834627856119448,
      "grad_norm": 0.32407071948208743,
      "learning_rate": 7.966628939794371e-08,
      "loss": 1.5554,
      "step": 5534
    },
    {
      "epoch": 0.8347786743081216,
      "grad_norm": 0.25065365752743557,
      "learning_rate": 7.961339800193484e-08,
      "loss": 1.6027,
      "step": 5535
    },
    {
      "epoch": 0.8349294924967952,
      "grad_norm": 0.28240372022150034,
      "learning_rate": 7.956055047590839e-08,
      "loss": 1.5693,
      "step": 5536
    },
    {
      "epoch": 0.8350803106854686,
      "grad_norm": 0.25248800480380923,
      "learning_rate": 7.950774683173011e-08,
      "loss": 1.5187,
      "step": 5537
    },
    {
      "epoch": 0.8352311288741422,
      "grad_norm": 0.2730600523929056,
      "learning_rate": 7.945498708125612e-08,
      "loss": 1.5953,
      "step": 5538
    },
    {
      "epoch": 0.8353819470628158,
      "grad_norm": 0.2821780763486024,
      "learning_rate": 7.940227123633243e-08,
      "loss": 1.6235,
      "step": 5539
    },
    {
      "epoch": 0.8355327652514893,
      "grad_norm": 0.24347407182184386,
      "learning_rate": 7.934959930879521e-08,
      "loss": 1.5455,
      "step": 5540
    },
    {
      "epoch": 0.8356835834401629,
      "grad_norm": 0.25890397409168897,
      "learning_rate": 7.929697131047099e-08,
      "loss": 1.5123,
      "step": 5541
    },
    {
      "epoch": 0.8358344016288365,
      "grad_norm": 0.24575142926152252,
      "learning_rate": 7.924438725317618e-08,
      "loss": 1.5395,
      "step": 5542
    },
    {
      "epoch": 0.83598521981751,
      "grad_norm": 0.2890724391677038,
      "learning_rate": 7.919184714871749e-08,
      "loss": 1.5507,
      "step": 5543
    },
    {
      "epoch": 0.8361360380061835,
      "grad_norm": 0.2556031827743178,
      "learning_rate": 7.91393510088916e-08,
      "loss": 1.6032,
      "step": 5544
    },
    {
      "epoch": 0.8362868561948571,
      "grad_norm": 0.2393186238286218,
      "learning_rate": 7.90868988454856e-08,
      "loss": 1.5682,
      "step": 5545
    },
    {
      "epoch": 0.8364376743835307,
      "grad_norm": 0.2543272307152719,
      "learning_rate": 7.90344906702764e-08,
      "loss": 1.5486,
      "step": 5546
    },
    {
      "epoch": 0.8365884925722042,
      "grad_norm": 0.2593772272704211,
      "learning_rate": 7.898212649503113e-08,
      "loss": 1.5835,
      "step": 5547
    },
    {
      "epoch": 0.8367393107608777,
      "grad_norm": 1.1666915568949274,
      "learning_rate": 7.892980633150718e-08,
      "loss": 1.6054,
      "step": 5548
    },
    {
      "epoch": 0.8368901289495513,
      "grad_norm": 0.2557101868732255,
      "learning_rate": 7.887753019145192e-08,
      "loss": 1.5073,
      "step": 5549
    },
    {
      "epoch": 0.8370409471382249,
      "grad_norm": 0.27047911109897305,
      "learning_rate": 7.882529808660282e-08,
      "loss": 1.5251,
      "step": 5550
    },
    {
      "epoch": 0.8371917653268984,
      "grad_norm": 0.2590908915314607,
      "learning_rate": 7.877311002868755e-08,
      "loss": 1.608,
      "step": 5551
    },
    {
      "epoch": 0.837342583515572,
      "grad_norm": 0.23856843868428404,
      "learning_rate": 7.872096602942385e-08,
      "loss": 1.5582,
      "step": 5552
    },
    {
      "epoch": 0.8374934017042456,
      "grad_norm": 0.24239285227628923,
      "learning_rate": 7.866886610051951e-08,
      "loss": 1.5828,
      "step": 5553
    },
    {
      "epoch": 0.837644219892919,
      "grad_norm": 0.2559961614417599,
      "learning_rate": 7.861681025367257e-08,
      "loss": 1.5976,
      "step": 5554
    },
    {
      "epoch": 0.8377950380815926,
      "grad_norm": 0.24511248185342885,
      "learning_rate": 7.856479850057105e-08,
      "loss": 1.6036,
      "step": 5555
    },
    {
      "epoch": 0.8379458562702662,
      "grad_norm": 0.25345278706853985,
      "learning_rate": 7.851283085289308e-08,
      "loss": 1.5205,
      "step": 5556
    },
    {
      "epoch": 0.8380966744589398,
      "grad_norm": 0.2674505680921593,
      "learning_rate": 7.846090732230689e-08,
      "loss": 1.5836,
      "step": 5557
    },
    {
      "epoch": 0.8382474926476133,
      "grad_norm": 0.2675322282599704,
      "learning_rate": 7.840902792047087e-08,
      "loss": 1.5791,
      "step": 5558
    },
    {
      "epoch": 0.8383983108362869,
      "grad_norm": 0.6383639200515794,
      "learning_rate": 7.835719265903347e-08,
      "loss": 1.5299,
      "step": 5559
    },
    {
      "epoch": 0.8385491290249604,
      "grad_norm": 0.3140365428237704,
      "learning_rate": 7.830540154963315e-08,
      "loss": 1.6029,
      "step": 5560
    },
    {
      "epoch": 0.838699947213634,
      "grad_norm": 0.2496152360860865,
      "learning_rate": 7.82536546038986e-08,
      "loss": 1.5501,
      "step": 5561
    },
    {
      "epoch": 0.8388507654023075,
      "grad_norm": 0.25380911714867804,
      "learning_rate": 7.820195183344848e-08,
      "loss": 1.5474,
      "step": 5562
    },
    {
      "epoch": 0.8390015835909811,
      "grad_norm": 0.3370098993111597,
      "learning_rate": 7.81502932498915e-08,
      "loss": 1.5793,
      "step": 5563
    },
    {
      "epoch": 0.8391524017796547,
      "grad_norm": 0.24127129356228108,
      "learning_rate": 7.809867886482666e-08,
      "loss": 1.5889,
      "step": 5564
    },
    {
      "epoch": 0.8393032199683281,
      "grad_norm": 0.24667600687525584,
      "learning_rate": 7.804710868984279e-08,
      "loss": 1.5609,
      "step": 5565
    },
    {
      "epoch": 0.8394540381570017,
      "grad_norm": 0.2400473169198927,
      "learning_rate": 7.799558273651889e-08,
      "loss": 1.5117,
      "step": 5566
    },
    {
      "epoch": 0.8396048563456753,
      "grad_norm": 0.24668529167986164,
      "learning_rate": 7.794410101642411e-08,
      "loss": 1.5906,
      "step": 5567
    },
    {
      "epoch": 0.8397556745343489,
      "grad_norm": 0.25670417154480313,
      "learning_rate": 7.789266354111755e-08,
      "loss": 1.5443,
      "step": 5568
    },
    {
      "epoch": 0.8399064927230224,
      "grad_norm": 0.2523566172943502,
      "learning_rate": 7.784127032214843e-08,
      "loss": 1.5969,
      "step": 5569
    },
    {
      "epoch": 0.840057310911696,
      "grad_norm": 0.25452856223989523,
      "learning_rate": 7.778992137105597e-08,
      "loss": 1.5534,
      "step": 5570
    },
    {
      "epoch": 0.8402081291003695,
      "grad_norm": 0.2544835216453723,
      "learning_rate": 7.773861669936962e-08,
      "loss": 1.5626,
      "step": 5571
    },
    {
      "epoch": 0.840358947289043,
      "grad_norm": 0.2699831837110939,
      "learning_rate": 7.768735631860867e-08,
      "loss": 1.6507,
      "step": 5572
    },
    {
      "epoch": 0.8405097654777166,
      "grad_norm": 0.25028425880563293,
      "learning_rate": 7.763614024028259e-08,
      "loss": 1.5905,
      "step": 5573
    },
    {
      "epoch": 0.8406605836663902,
      "grad_norm": 0.2814255923009269,
      "learning_rate": 7.758496847589091e-08,
      "loss": 1.5698,
      "step": 5574
    },
    {
      "epoch": 0.8408114018550638,
      "grad_norm": 0.6791691978786829,
      "learning_rate": 7.75338410369232e-08,
      "loss": 1.516,
      "step": 5575
    },
    {
      "epoch": 0.8409622200437373,
      "grad_norm": 0.25940149426018794,
      "learning_rate": 7.748275793485896e-08,
      "loss": 1.5214,
      "step": 5576
    },
    {
      "epoch": 0.8411130382324108,
      "grad_norm": 0.24413826048920553,
      "learning_rate": 7.743171918116797e-08,
      "loss": 1.5391,
      "step": 5577
    },
    {
      "epoch": 0.8412638564210844,
      "grad_norm": 0.2461978788362587,
      "learning_rate": 7.738072478730983e-08,
      "loss": 1.5605,
      "step": 5578
    },
    {
      "epoch": 0.8414146746097579,
      "grad_norm": 0.26108702667629563,
      "learning_rate": 7.732977476473425e-08,
      "loss": 1.6231,
      "step": 5579
    },
    {
      "epoch": 0.8415654927984315,
      "grad_norm": 0.2595152265125195,
      "learning_rate": 7.727886912488106e-08,
      "loss": 1.5943,
      "step": 5580
    },
    {
      "epoch": 0.8417163109871051,
      "grad_norm": 0.2658759507588597,
      "learning_rate": 7.722800787918005e-08,
      "loss": 1.5507,
      "step": 5581
    },
    {
      "epoch": 0.8418671291757786,
      "grad_norm": 0.448379173388005,
      "learning_rate": 7.717719103905102e-08,
      "loss": 1.4971,
      "step": 5582
    },
    {
      "epoch": 0.8420179473644521,
      "grad_norm": 0.24230736636294858,
      "learning_rate": 7.71264186159038e-08,
      "loss": 1.5485,
      "step": 5583
    },
    {
      "epoch": 0.8421687655531257,
      "grad_norm": 0.2533778985670752,
      "learning_rate": 7.707569062113835e-08,
      "loss": 1.5598,
      "step": 5584
    },
    {
      "epoch": 0.8423195837417993,
      "grad_norm": 0.2798241000686758,
      "learning_rate": 7.702500706614454e-08,
      "loss": 1.6241,
      "step": 5585
    },
    {
      "epoch": 0.8424704019304728,
      "grad_norm": 0.25086697903795707,
      "learning_rate": 7.697436796230228e-08,
      "loss": 1.5514,
      "step": 5586
    },
    {
      "epoch": 0.8426212201191464,
      "grad_norm": 0.2887072752715407,
      "learning_rate": 7.69237733209816e-08,
      "loss": 1.522,
      "step": 5587
    },
    {
      "epoch": 0.8427720383078199,
      "grad_norm": 0.42417698732702486,
      "learning_rate": 7.687322315354242e-08,
      "loss": 1.6105,
      "step": 5588
    },
    {
      "epoch": 0.8429228564964935,
      "grad_norm": 0.25776113315138227,
      "learning_rate": 7.682271747133468e-08,
      "loss": 1.5826,
      "step": 5589
    },
    {
      "epoch": 0.843073674685167,
      "grad_norm": 0.244347068403562,
      "learning_rate": 7.67722562856985e-08,
      "loss": 1.6212,
      "step": 5590
    },
    {
      "epoch": 0.8432244928738406,
      "grad_norm": 0.28406537972836354,
      "learning_rate": 7.67218396079638e-08,
      "loss": 1.5369,
      "step": 5591
    },
    {
      "epoch": 0.8433753110625142,
      "grad_norm": 0.25018817775540264,
      "learning_rate": 7.66714674494506e-08,
      "loss": 1.6136,
      "step": 5592
    },
    {
      "epoch": 0.8435261292511876,
      "grad_norm": 0.2660496140642561,
      "learning_rate": 7.662113982146887e-08,
      "loss": 1.5692,
      "step": 5593
    },
    {
      "epoch": 0.8436769474398612,
      "grad_norm": 0.24446740140627457,
      "learning_rate": 7.657085673531874e-08,
      "loss": 1.5945,
      "step": 5594
    },
    {
      "epoch": 0.8438277656285348,
      "grad_norm": 0.2528406292945506,
      "learning_rate": 7.652061820229016e-08,
      "loss": 1.5499,
      "step": 5595
    },
    {
      "epoch": 0.8439785838172084,
      "grad_norm": 0.24120135201280207,
      "learning_rate": 7.647042423366312e-08,
      "loss": 1.502,
      "step": 5596
    },
    {
      "epoch": 0.8441294020058819,
      "grad_norm": 0.24289934292254134,
      "learning_rate": 7.642027484070772e-08,
      "loss": 1.5658,
      "step": 5597
    },
    {
      "epoch": 0.8442802201945555,
      "grad_norm": 0.24757642087809018,
      "learning_rate": 7.637017003468385e-08,
      "loss": 1.5622,
      "step": 5598
    },
    {
      "epoch": 0.844431038383229,
      "grad_norm": 0.2507387146409576,
      "learning_rate": 7.632010982684153e-08,
      "loss": 1.5469,
      "step": 5599
    },
    {
      "epoch": 0.8445818565719025,
      "grad_norm": 0.26293837564154415,
      "learning_rate": 7.627009422842081e-08,
      "loss": 1.5901,
      "step": 5600
    },
    {
      "epoch": 0.8447326747605761,
      "grad_norm": 0.24592612200606645,
      "learning_rate": 7.622012325065157e-08,
      "loss": 1.5197,
      "step": 5601
    },
    {
      "epoch": 0.8448834929492497,
      "grad_norm": 0.3179625061264734,
      "learning_rate": 7.617019690475374e-08,
      "loss": 1.5853,
      "step": 5602
    },
    {
      "epoch": 0.8450343111379233,
      "grad_norm": 0.31593090013279307,
      "learning_rate": 7.61203152019373e-08,
      "loss": 1.5618,
      "step": 5603
    },
    {
      "epoch": 0.8451851293265968,
      "grad_norm": 0.24646286521620397,
      "learning_rate": 7.607047815340212e-08,
      "loss": 1.511,
      "step": 5604
    },
    {
      "epoch": 0.8453359475152703,
      "grad_norm": 0.24985925653566338,
      "learning_rate": 7.602068577033807e-08,
      "loss": 1.5493,
      "step": 5605
    },
    {
      "epoch": 0.8454867657039439,
      "grad_norm": 0.23941690494100681,
      "learning_rate": 7.597093806392493e-08,
      "loss": 1.5622,
      "step": 5606
    },
    {
      "epoch": 0.8456375838926175,
      "grad_norm": 1.316138816902561,
      "learning_rate": 7.59212350453326e-08,
      "loss": 1.6131,
      "step": 5607
    },
    {
      "epoch": 0.845788402081291,
      "grad_norm": 0.239637249766247,
      "learning_rate": 7.587157672572083e-08,
      "loss": 1.5916,
      "step": 5608
    },
    {
      "epoch": 0.8459392202699646,
      "grad_norm": 0.3962825317708086,
      "learning_rate": 7.582196311623928e-08,
      "loss": 1.589,
      "step": 5609
    },
    {
      "epoch": 0.8460900384586381,
      "grad_norm": 0.2630091138207068,
      "learning_rate": 7.577239422802779e-08,
      "loss": 1.5847,
      "step": 5610
    },
    {
      "epoch": 0.8462408566473116,
      "grad_norm": 0.2599746401134038,
      "learning_rate": 7.572287007221593e-08,
      "loss": 1.5781,
      "step": 5611
    },
    {
      "epoch": 0.8463916748359852,
      "grad_norm": 0.25472182490482365,
      "learning_rate": 7.567339065992328e-08,
      "loss": 1.5655,
      "step": 5612
    },
    {
      "epoch": 0.8465424930246588,
      "grad_norm": 0.24206481005066374,
      "learning_rate": 7.56239560022595e-08,
      "loss": 1.4791,
      "step": 5613
    },
    {
      "epoch": 0.8466933112133324,
      "grad_norm": 0.24641177015967947,
      "learning_rate": 7.557456611032408e-08,
      "loss": 1.5529,
      "step": 5614
    },
    {
      "epoch": 0.8468441294020059,
      "grad_norm": 0.39188335002461877,
      "learning_rate": 7.552522099520645e-08,
      "loss": 1.5349,
      "step": 5615
    },
    {
      "epoch": 0.8469949475906794,
      "grad_norm": 0.28795551978446376,
      "learning_rate": 7.547592066798609e-08,
      "loss": 1.5694,
      "step": 5616
    },
    {
      "epoch": 0.847145765779353,
      "grad_norm": 0.2527501508656064,
      "learning_rate": 7.542666513973229e-08,
      "loss": 1.573,
      "step": 5617
    },
    {
      "epoch": 0.8472965839680265,
      "grad_norm": 0.32103011451645874,
      "learning_rate": 7.537745442150443e-08,
      "loss": 1.5757,
      "step": 5618
    },
    {
      "epoch": 0.8474474021567001,
      "grad_norm": 0.306449996138448,
      "learning_rate": 7.532828852435163e-08,
      "loss": 1.5639,
      "step": 5619
    },
    {
      "epoch": 0.8475982203453737,
      "grad_norm": 0.25749194558302557,
      "learning_rate": 7.52791674593132e-08,
      "loss": 1.5643,
      "step": 5620
    },
    {
      "epoch": 0.8477490385340473,
      "grad_norm": 0.2840668647636897,
      "learning_rate": 7.523009123741819e-08,
      "loss": 1.5831,
      "step": 5621
    },
    {
      "epoch": 0.8478998567227207,
      "grad_norm": 0.24338742656455553,
      "learning_rate": 7.518105986968558e-08,
      "loss": 1.5688,
      "step": 5622
    },
    {
      "epoch": 0.8480506749113943,
      "grad_norm": 0.24009537974004305,
      "learning_rate": 7.513207336712446e-08,
      "loss": 1.5483,
      "step": 5623
    },
    {
      "epoch": 0.8482014931000679,
      "grad_norm": 0.25237577432805486,
      "learning_rate": 7.508313174073364e-08,
      "loss": 1.5215,
      "step": 5624
    },
    {
      "epoch": 0.8483523112887414,
      "grad_norm": 0.24748066675250419,
      "learning_rate": 7.503423500150194e-08,
      "loss": 1.6166,
      "step": 5625
    },
    {
      "epoch": 0.848503129477415,
      "grad_norm": 0.25702548854248036,
      "learning_rate": 7.498538316040818e-08,
      "loss": 1.4807,
      "step": 5626
    },
    {
      "epoch": 0.8486539476660885,
      "grad_norm": 0.25364583712625355,
      "learning_rate": 7.493657622842098e-08,
      "loss": 1.5994,
      "step": 5627
    },
    {
      "epoch": 0.8488047658547621,
      "grad_norm": 0.2600433736421623,
      "learning_rate": 7.488781421649882e-08,
      "loss": 1.5958,
      "step": 5628
    },
    {
      "epoch": 0.8489555840434356,
      "grad_norm": 0.32046931742402496,
      "learning_rate": 7.483909713559035e-08,
      "loss": 1.6043,
      "step": 5629
    },
    {
      "epoch": 0.8491064022321092,
      "grad_norm": 0.4017573004659793,
      "learning_rate": 7.479042499663392e-08,
      "loss": 1.5679,
      "step": 5630
    },
    {
      "epoch": 0.8492572204207828,
      "grad_norm": 0.26866946450764245,
      "learning_rate": 7.474179781055782e-08,
      "loss": 1.592,
      "step": 5631
    },
    {
      "epoch": 0.8494080386094564,
      "grad_norm": 0.24874450046086075,
      "learning_rate": 7.46932155882802e-08,
      "loss": 1.5707,
      "step": 5632
    },
    {
      "epoch": 0.8495588567981298,
      "grad_norm": 0.660046811125414,
      "learning_rate": 7.464467834070934e-08,
      "loss": 1.5288,
      "step": 5633
    },
    {
      "epoch": 0.8497096749868034,
      "grad_norm": 0.252021037615029,
      "learning_rate": 7.459618607874316e-08,
      "loss": 1.5588,
      "step": 5634
    },
    {
      "epoch": 0.849860493175477,
      "grad_norm": 0.25898808254480193,
      "learning_rate": 7.454773881326957e-08,
      "loss": 1.5758,
      "step": 5635
    },
    {
      "epoch": 0.8500113113641505,
      "grad_norm": 0.24795417150648452,
      "learning_rate": 7.449933655516646e-08,
      "loss": 1.5693,
      "step": 5636
    },
    {
      "epoch": 0.8501621295528241,
      "grad_norm": 0.25017592633937047,
      "learning_rate": 7.445097931530156e-08,
      "loss": 1.5651,
      "step": 5637
    },
    {
      "epoch": 0.8503129477414976,
      "grad_norm": 0.24218888929716142,
      "learning_rate": 7.440266710453235e-08,
      "loss": 1.5262,
      "step": 5638
    },
    {
      "epoch": 0.8504637659301711,
      "grad_norm": 0.24610871936433715,
      "learning_rate": 7.435439993370649e-08,
      "loss": 1.5375,
      "step": 5639
    },
    {
      "epoch": 0.8506145841188447,
      "grad_norm": 0.25806357334048957,
      "learning_rate": 7.430617781366128e-08,
      "loss": 1.5717,
      "step": 5640
    },
    {
      "epoch": 0.8507654023075183,
      "grad_norm": 0.25833198862097434,
      "learning_rate": 7.425800075522395e-08,
      "loss": 1.6277,
      "step": 5641
    },
    {
      "epoch": 0.8509162204961919,
      "grad_norm": 0.39064568521698845,
      "learning_rate": 7.420986876921174e-08,
      "loss": 1.5529,
      "step": 5642
    },
    {
      "epoch": 0.8510670386848654,
      "grad_norm": 0.25120786856002797,
      "learning_rate": 7.416178186643168e-08,
      "loss": 1.5673,
      "step": 5643
    },
    {
      "epoch": 0.8512178568735389,
      "grad_norm": 0.25188296940139504,
      "learning_rate": 7.411374005768061e-08,
      "loss": 1.5236,
      "step": 5644
    },
    {
      "epoch": 0.8513686750622125,
      "grad_norm": 0.2389433206874455,
      "learning_rate": 7.406574335374534e-08,
      "loss": 1.5731,
      "step": 5645
    },
    {
      "epoch": 0.851519493250886,
      "grad_norm": 0.2497285089354711,
      "learning_rate": 7.401779176540256e-08,
      "loss": 1.5333,
      "step": 5646
    },
    {
      "epoch": 0.8516703114395596,
      "grad_norm": 0.25756152163144364,
      "learning_rate": 7.396988530341877e-08,
      "loss": 1.5509,
      "step": 5647
    },
    {
      "epoch": 0.8518211296282332,
      "grad_norm": 0.2539468520583549,
      "learning_rate": 7.392202397855036e-08,
      "loss": 1.5229,
      "step": 5648
    },
    {
      "epoch": 0.8519719478169068,
      "grad_norm": 0.26121415983020824,
      "learning_rate": 7.387420780154363e-08,
      "loss": 1.4851,
      "step": 5649
    },
    {
      "epoch": 0.8521227660055802,
      "grad_norm": 0.261710287768565,
      "learning_rate": 7.382643678313467e-08,
      "loss": 1.5977,
      "step": 5650
    },
    {
      "epoch": 0.8522735841942538,
      "grad_norm": 0.240550327088055,
      "learning_rate": 7.377871093404946e-08,
      "loss": 1.52,
      "step": 5651
    },
    {
      "epoch": 0.8524244023829274,
      "grad_norm": 0.2570836869612234,
      "learning_rate": 7.373103026500387e-08,
      "loss": 1.4896,
      "step": 5652
    },
    {
      "epoch": 0.852575220571601,
      "grad_norm": 0.24238387060556735,
      "learning_rate": 7.36833947867036e-08,
      "loss": 1.5692,
      "step": 5653
    },
    {
      "epoch": 0.8527260387602745,
      "grad_norm": 0.2377588804521395,
      "learning_rate": 7.363580450984417e-08,
      "loss": 1.4929,
      "step": 5654
    },
    {
      "epoch": 0.852876856948948,
      "grad_norm": 0.2501993070047394,
      "learning_rate": 7.3588259445111e-08,
      "loss": 1.6074,
      "step": 5655
    },
    {
      "epoch": 0.8530276751376216,
      "grad_norm": 0.2906505749795475,
      "learning_rate": 7.354075960317934e-08,
      "loss": 1.5748,
      "step": 5656
    },
    {
      "epoch": 0.8531784933262951,
      "grad_norm": 0.39481323868466706,
      "learning_rate": 7.349330499471432e-08,
      "loss": 1.5164,
      "step": 5657
    },
    {
      "epoch": 0.8533293115149687,
      "grad_norm": 0.2640431809310381,
      "learning_rate": 7.344589563037081e-08,
      "loss": 1.5883,
      "step": 5658
    },
    {
      "epoch": 0.8534801297036423,
      "grad_norm": 0.24985927957112009,
      "learning_rate": 7.339853152079366e-08,
      "loss": 1.5835,
      "step": 5659
    },
    {
      "epoch": 0.8536309478923159,
      "grad_norm": 0.24758996546327033,
      "learning_rate": 7.335121267661747e-08,
      "loss": 1.6124,
      "step": 5660
    },
    {
      "epoch": 0.8537817660809893,
      "grad_norm": 0.2858069415079648,
      "learning_rate": 7.330393910846664e-08,
      "loss": 1.5454,
      "step": 5661
    },
    {
      "epoch": 0.8539325842696629,
      "grad_norm": 0.2541569244416825,
      "learning_rate": 7.32567108269556e-08,
      "loss": 1.5531,
      "step": 5662
    },
    {
      "epoch": 0.8540834024583365,
      "grad_norm": 0.2504834782357204,
      "learning_rate": 7.320952784268838e-08,
      "loss": 1.5662,
      "step": 5663
    },
    {
      "epoch": 0.85423422064701,
      "grad_norm": 0.24326816683178065,
      "learning_rate": 7.316239016625888e-08,
      "loss": 1.549,
      "step": 5664
    },
    {
      "epoch": 0.8543850388356836,
      "grad_norm": 0.24333803414480876,
      "learning_rate": 7.311529780825101e-08,
      "loss": 1.6189,
      "step": 5665
    },
    {
      "epoch": 0.8545358570243572,
      "grad_norm": 0.24632431855036313,
      "learning_rate": 7.306825077923832e-08,
      "loss": 1.5989,
      "step": 5666
    },
    {
      "epoch": 0.8546866752130307,
      "grad_norm": 0.24275227716740327,
      "learning_rate": 7.302124908978419e-08,
      "loss": 1.539,
      "step": 5667
    },
    {
      "epoch": 0.8548374934017042,
      "grad_norm": 0.27317071756739514,
      "learning_rate": 7.297429275044193e-08,
      "loss": 1.5005,
      "step": 5668
    },
    {
      "epoch": 0.8549883115903778,
      "grad_norm": 0.5863882394070457,
      "learning_rate": 7.292738177175462e-08,
      "loss": 1.6435,
      "step": 5669
    },
    {
      "epoch": 0.8551391297790514,
      "grad_norm": 0.2546533554328017,
      "learning_rate": 7.28805161642551e-08,
      "loss": 1.5396,
      "step": 5670
    },
    {
      "epoch": 0.855289947967725,
      "grad_norm": 0.2611516498872777,
      "learning_rate": 7.283369593846602e-08,
      "loss": 1.5391,
      "step": 5671
    },
    {
      "epoch": 0.8554407661563984,
      "grad_norm": 0.25575431344121646,
      "learning_rate": 7.278692110489998e-08,
      "loss": 1.66,
      "step": 5672
    },
    {
      "epoch": 0.855591584345072,
      "grad_norm": 0.25337217431271386,
      "learning_rate": 7.274019167405922e-08,
      "loss": 1.5817,
      "step": 5673
    },
    {
      "epoch": 0.8557424025337456,
      "grad_norm": 0.25231846480237374,
      "learning_rate": 7.269350765643588e-08,
      "loss": 1.5301,
      "step": 5674
    },
    {
      "epoch": 0.8558932207224191,
      "grad_norm": 0.2430603528695757,
      "learning_rate": 7.264686906251191e-08,
      "loss": 1.5735,
      "step": 5675
    },
    {
      "epoch": 0.8560440389110927,
      "grad_norm": 0.25613443783027007,
      "learning_rate": 7.260027590275903e-08,
      "loss": 1.5524,
      "step": 5676
    },
    {
      "epoch": 0.8561948570997663,
      "grad_norm": 0.41771966232826235,
      "learning_rate": 7.255372818763868e-08,
      "loss": 1.5148,
      "step": 5677
    },
    {
      "epoch": 0.8563456752884397,
      "grad_norm": 0.26630543909180215,
      "learning_rate": 7.25072259276023e-08,
      "loss": 1.6217,
      "step": 5678
    },
    {
      "epoch": 0.8564964934771133,
      "grad_norm": 0.27526533874062464,
      "learning_rate": 7.246076913309094e-08,
      "loss": 1.5862,
      "step": 5679
    },
    {
      "epoch": 0.8566473116657869,
      "grad_norm": 0.2429329669365647,
      "learning_rate": 7.241435781453543e-08,
      "loss": 1.6406,
      "step": 5680
    },
    {
      "epoch": 0.8567981298544605,
      "grad_norm": 0.2668874326283706,
      "learning_rate": 7.23679919823566e-08,
      "loss": 1.5187,
      "step": 5681
    },
    {
      "epoch": 0.856948948043134,
      "grad_norm": 0.24466236372841946,
      "learning_rate": 7.232167164696487e-08,
      "loss": 1.5276,
      "step": 5682
    },
    {
      "epoch": 0.8570997662318075,
      "grad_norm": 0.30415978131778804,
      "learning_rate": 7.22753968187605e-08,
      "loss": 1.5922,
      "step": 5683
    },
    {
      "epoch": 0.8572505844204811,
      "grad_norm": 0.26139733105451657,
      "learning_rate": 7.222916750813349e-08,
      "loss": 1.5761,
      "step": 5684
    },
    {
      "epoch": 0.8574014026091547,
      "grad_norm": 0.6127118554212033,
      "learning_rate": 7.218298372546374e-08,
      "loss": 1.5888,
      "step": 5685
    },
    {
      "epoch": 0.8575522207978282,
      "grad_norm": 0.2401331880168586,
      "learning_rate": 7.213684548112085e-08,
      "loss": 1.5736,
      "step": 5686
    },
    {
      "epoch": 0.8577030389865018,
      "grad_norm": 0.2444064979383252,
      "learning_rate": 7.209075278546411e-08,
      "loss": 1.5905,
      "step": 5687
    },
    {
      "epoch": 0.8578538571751754,
      "grad_norm": 0.2532890179769075,
      "learning_rate": 7.204470564884281e-08,
      "loss": 1.4892,
      "step": 5688
    },
    {
      "epoch": 0.8580046753638488,
      "grad_norm": 0.23250439984319513,
      "learning_rate": 7.19987040815958e-08,
      "loss": 1.5769,
      "step": 5689
    },
    {
      "epoch": 0.8581554935525224,
      "grad_norm": 0.2862891322557891,
      "learning_rate": 7.195274809405173e-08,
      "loss": 1.5883,
      "step": 5690
    },
    {
      "epoch": 0.858306311741196,
      "grad_norm": 0.26399834309076403,
      "learning_rate": 7.190683769652914e-08,
      "loss": 1.5601,
      "step": 5691
    },
    {
      "epoch": 0.8584571299298696,
      "grad_norm": 0.24562698941058486,
      "learning_rate": 7.186097289933623e-08,
      "loss": 1.5354,
      "step": 5692
    },
    {
      "epoch": 0.8586079481185431,
      "grad_norm": 0.2715514934169429,
      "learning_rate": 7.181515371277094e-08,
      "loss": 1.592,
      "step": 5693
    },
    {
      "epoch": 0.8587587663072167,
      "grad_norm": 0.2523047760437185,
      "learning_rate": 7.176938014712101e-08,
      "loss": 1.5668,
      "step": 5694
    },
    {
      "epoch": 0.8589095844958902,
      "grad_norm": 2.259263223543891,
      "learning_rate": 7.172365221266402e-08,
      "loss": 1.6373,
      "step": 5695
    },
    {
      "epoch": 0.8590604026845637,
      "grad_norm": 0.26405129336613986,
      "learning_rate": 7.167796991966716e-08,
      "loss": 1.6148,
      "step": 5696
    },
    {
      "epoch": 0.8592112208732373,
      "grad_norm": 0.24976180957847277,
      "learning_rate": 7.163233327838739e-08,
      "loss": 1.58,
      "step": 5697
    },
    {
      "epoch": 0.8593620390619109,
      "grad_norm": 0.2584168725948234,
      "learning_rate": 7.158674229907154e-08,
      "loss": 1.5139,
      "step": 5698
    },
    {
      "epoch": 0.8595128572505845,
      "grad_norm": 0.2590229882263229,
      "learning_rate": 7.154119699195613e-08,
      "loss": 1.5317,
      "step": 5699
    },
    {
      "epoch": 0.8596636754392579,
      "grad_norm": 0.2861304517231533,
      "learning_rate": 7.149569736726736e-08,
      "loss": 1.6012,
      "step": 5700
    },
    {
      "epoch": 0.8598144936279315,
      "grad_norm": 0.2847592989614429,
      "learning_rate": 7.145024343522119e-08,
      "loss": 1.5457,
      "step": 5701
    },
    {
      "epoch": 0.8599653118166051,
      "grad_norm": 0.8991918096872035,
      "learning_rate": 7.140483520602345e-08,
      "loss": 1.6882,
      "step": 5702
    },
    {
      "epoch": 0.8601161300052786,
      "grad_norm": 0.2564977267425987,
      "learning_rate": 7.135947268986955e-08,
      "loss": 1.5235,
      "step": 5703
    },
    {
      "epoch": 0.8602669481939522,
      "grad_norm": 0.2441356832573264,
      "learning_rate": 7.131415589694465e-08,
      "loss": 1.5601,
      "step": 5704
    },
    {
      "epoch": 0.8604177663826258,
      "grad_norm": 0.24779244117565505,
      "learning_rate": 7.126888483742377e-08,
      "loss": 1.4907,
      "step": 5705
    },
    {
      "epoch": 0.8605685845712993,
      "grad_norm": 0.24055359786155137,
      "learning_rate": 7.122365952147156e-08,
      "loss": 1.6108,
      "step": 5706
    },
    {
      "epoch": 0.8607194027599728,
      "grad_norm": 0.2662572871437054,
      "learning_rate": 7.117847995924237e-08,
      "loss": 1.543,
      "step": 5707
    },
    {
      "epoch": 0.8608702209486464,
      "grad_norm": 0.2548125020106732,
      "learning_rate": 7.113334616088041e-08,
      "loss": 1.6025,
      "step": 5708
    },
    {
      "epoch": 0.86102103913732,
      "grad_norm": 0.25052042672988706,
      "learning_rate": 7.108825813651951e-08,
      "loss": 1.5637,
      "step": 5709
    },
    {
      "epoch": 0.8611718573259936,
      "grad_norm": 0.2727387078531334,
      "learning_rate": 7.104321589628318e-08,
      "loss": 1.6394,
      "step": 5710
    },
    {
      "epoch": 0.8613226755146671,
      "grad_norm": 0.27553186094645604,
      "learning_rate": 7.099821945028474e-08,
      "loss": 1.5856,
      "step": 5711
    },
    {
      "epoch": 0.8614734937033406,
      "grad_norm": 0.24357870548678773,
      "learning_rate": 7.095326880862729e-08,
      "loss": 1.555,
      "step": 5712
    },
    {
      "epoch": 0.8616243118920142,
      "grad_norm": 0.3300369116661402,
      "learning_rate": 7.090836398140346e-08,
      "loss": 1.5183,
      "step": 5713
    },
    {
      "epoch": 0.8617751300806877,
      "grad_norm": 0.2534416640536777,
      "learning_rate": 7.086350497869567e-08,
      "loss": 1.5871,
      "step": 5714
    },
    {
      "epoch": 0.8619259482693613,
      "grad_norm": 0.24730149269518625,
      "learning_rate": 7.081869181057618e-08,
      "loss": 1.5443,
      "step": 5715
    },
    {
      "epoch": 0.8620767664580349,
      "grad_norm": 0.2487861282647905,
      "learning_rate": 7.077392448710681e-08,
      "loss": 1.5806,
      "step": 5716
    },
    {
      "epoch": 0.8622275846467083,
      "grad_norm": 0.2554637618776619,
      "learning_rate": 7.072920301833905e-08,
      "loss": 1.5154,
      "step": 5717
    },
    {
      "epoch": 0.8623784028353819,
      "grad_norm": 0.2516345689248287,
      "learning_rate": 7.068452741431429e-08,
      "loss": 1.5671,
      "step": 5718
    },
    {
      "epoch": 0.8625292210240555,
      "grad_norm": 0.24574308695529595,
      "learning_rate": 7.063989768506345e-08,
      "loss": 1.5913,
      "step": 5719
    },
    {
      "epoch": 0.8626800392127291,
      "grad_norm": 0.3014547838909663,
      "learning_rate": 7.059531384060716e-08,
      "loss": 1.5404,
      "step": 5720
    },
    {
      "epoch": 0.8628308574014026,
      "grad_norm": 0.25323018494644195,
      "learning_rate": 7.055077589095587e-08,
      "loss": 1.5407,
      "step": 5721
    },
    {
      "epoch": 0.8629816755900762,
      "grad_norm": 0.31986724307662023,
      "learning_rate": 7.050628384610962e-08,
      "loss": 1.5921,
      "step": 5722
    },
    {
      "epoch": 0.8631324937787497,
      "grad_norm": 0.2924292113515159,
      "learning_rate": 7.046183771605815e-08,
      "loss": 1.4993,
      "step": 5723
    },
    {
      "epoch": 0.8632833119674233,
      "grad_norm": 0.2472371685103149,
      "learning_rate": 7.041743751078089e-08,
      "loss": 1.5444,
      "step": 5724
    },
    {
      "epoch": 0.8634341301560968,
      "grad_norm": 0.245464840150854,
      "learning_rate": 7.037308324024704e-08,
      "loss": 1.5841,
      "step": 5725
    },
    {
      "epoch": 0.8635849483447704,
      "grad_norm": 0.24565565993670377,
      "learning_rate": 7.03287749144154e-08,
      "loss": 1.5878,
      "step": 5726
    },
    {
      "epoch": 0.863735766533444,
      "grad_norm": 0.2387712430985956,
      "learning_rate": 7.028451254323447e-08,
      "loss": 1.5179,
      "step": 5727
    },
    {
      "epoch": 0.8638865847221174,
      "grad_norm": 0.24125454898533927,
      "learning_rate": 7.024029613664247e-08,
      "loss": 1.4979,
      "step": 5728
    },
    {
      "epoch": 0.864037402910791,
      "grad_norm": 0.2606603120641778,
      "learning_rate": 7.019612570456723e-08,
      "loss": 1.612,
      "step": 5729
    },
    {
      "epoch": 0.8641882210994646,
      "grad_norm": 0.2708461358028818,
      "learning_rate": 7.015200125692629e-08,
      "loss": 1.4908,
      "step": 5730
    },
    {
      "epoch": 0.8643390392881382,
      "grad_norm": 0.2495650801146675,
      "learning_rate": 7.010792280362695e-08,
      "loss": 1.6097,
      "step": 5731
    },
    {
      "epoch": 0.8644898574768117,
      "grad_norm": 0.2630385060355195,
      "learning_rate": 7.006389035456604e-08,
      "loss": 1.607,
      "step": 5732
    },
    {
      "epoch": 0.8646406756654853,
      "grad_norm": 0.2541416968748563,
      "learning_rate": 7.00199039196301e-08,
      "loss": 1.5801,
      "step": 5733
    },
    {
      "epoch": 0.8647914938541588,
      "grad_norm": 0.26597707933889025,
      "learning_rate": 6.997596350869547e-08,
      "loss": 1.5898,
      "step": 5734
    },
    {
      "epoch": 0.8649423120428323,
      "grad_norm": 0.2506369269093555,
      "learning_rate": 6.993206913162799e-08,
      "loss": 1.5741,
      "step": 5735
    },
    {
      "epoch": 0.8650931302315059,
      "grad_norm": 0.2634889182152881,
      "learning_rate": 6.988822079828322e-08,
      "loss": 1.5036,
      "step": 5736
    },
    {
      "epoch": 0.8652439484201795,
      "grad_norm": 0.4923504453984544,
      "learning_rate": 6.984441851850637e-08,
      "loss": 1.5609,
      "step": 5737
    },
    {
      "epoch": 0.8653947666088531,
      "grad_norm": 0.2521371255702623,
      "learning_rate": 6.980066230213236e-08,
      "loss": 1.5463,
      "step": 5738
    },
    {
      "epoch": 0.8655455847975266,
      "grad_norm": 0.24552411522231396,
      "learning_rate": 6.975695215898576e-08,
      "loss": 1.4921,
      "step": 5739
    },
    {
      "epoch": 0.8656964029862001,
      "grad_norm": 0.978859631129009,
      "learning_rate": 6.97132880988807e-08,
      "loss": 1.5929,
      "step": 5740
    },
    {
      "epoch": 0.8658472211748737,
      "grad_norm": 0.24414622961417418,
      "learning_rate": 6.966967013162108e-08,
      "loss": 1.5397,
      "step": 5741
    },
    {
      "epoch": 0.8659980393635472,
      "grad_norm": 0.24789664979346512,
      "learning_rate": 6.962609826700042e-08,
      "loss": 1.5825,
      "step": 5742
    },
    {
      "epoch": 0.8661488575522208,
      "grad_norm": 0.23748110644282722,
      "learning_rate": 6.95825725148018e-08,
      "loss": 1.5822,
      "step": 5743
    },
    {
      "epoch": 0.8662996757408944,
      "grad_norm": 0.3831547332485798,
      "learning_rate": 6.953909288479811e-08,
      "loss": 1.5613,
      "step": 5744
    },
    {
      "epoch": 0.8664504939295679,
      "grad_norm": 0.23485383567984625,
      "learning_rate": 6.949565938675177e-08,
      "loss": 1.5425,
      "step": 5745
    },
    {
      "epoch": 0.8666013121182414,
      "grad_norm": 0.2429333921619344,
      "learning_rate": 6.945227203041478e-08,
      "loss": 1.5293,
      "step": 5746
    },
    {
      "epoch": 0.866752130306915,
      "grad_norm": 0.2368445493413857,
      "learning_rate": 6.940893082552898e-08,
      "loss": 1.4339,
      "step": 5747
    },
    {
      "epoch": 0.8669029484955886,
      "grad_norm": 0.26343309796527276,
      "learning_rate": 6.936563578182568e-08,
      "loss": 1.585,
      "step": 5748
    },
    {
      "epoch": 0.8670537666842622,
      "grad_norm": 0.26592788291536285,
      "learning_rate": 6.932238690902587e-08,
      "loss": 1.5261,
      "step": 5749
    },
    {
      "epoch": 0.8672045848729357,
      "grad_norm": 0.2662401816901963,
      "learning_rate": 6.927918421684018e-08,
      "loss": 1.5693,
      "step": 5750
    },
    {
      "epoch": 0.8673554030616092,
      "grad_norm": 0.2549205869783248,
      "learning_rate": 6.923602771496892e-08,
      "loss": 1.5528,
      "step": 5751
    },
    {
      "epoch": 0.8675062212502828,
      "grad_norm": 0.2572567660331949,
      "learning_rate": 6.919291741310195e-08,
      "loss": 1.5549,
      "step": 5752
    },
    {
      "epoch": 0.8676570394389563,
      "grad_norm": 0.2505870450308744,
      "learning_rate": 6.914985332091872e-08,
      "loss": 1.5467,
      "step": 5753
    },
    {
      "epoch": 0.8678078576276299,
      "grad_norm": 0.26579179135339104,
      "learning_rate": 6.91068354480885e-08,
      "loss": 1.5146,
      "step": 5754
    },
    {
      "epoch": 0.8679586758163035,
      "grad_norm": 0.24709517636008974,
      "learning_rate": 6.906386380426997e-08,
      "loss": 1.5421,
      "step": 5755
    },
    {
      "epoch": 0.868109494004977,
      "grad_norm": 0.24904109958646975,
      "learning_rate": 6.902093839911153e-08,
      "loss": 1.549,
      "step": 5756
    },
    {
      "epoch": 0.8682603121936505,
      "grad_norm": 0.24555385869845897,
      "learning_rate": 6.897805924225121e-08,
      "loss": 1.5294,
      "step": 5757
    },
    {
      "epoch": 0.8684111303823241,
      "grad_norm": 0.2531488037293015,
      "learning_rate": 6.893522634331664e-08,
      "loss": 1.5193,
      "step": 5758
    },
    {
      "epoch": 0.8685619485709977,
      "grad_norm": 0.24920935663852561,
      "learning_rate": 6.889243971192496e-08,
      "loss": 1.569,
      "step": 5759
    },
    {
      "epoch": 0.8687127667596712,
      "grad_norm": 0.2518169601440843,
      "learning_rate": 6.884969935768312e-08,
      "loss": 1.5928,
      "step": 5760
    },
    {
      "epoch": 0.8688635849483448,
      "grad_norm": 0.2563759875025315,
      "learning_rate": 6.880700529018754e-08,
      "loss": 1.5736,
      "step": 5761
    },
    {
      "epoch": 0.8690144031370183,
      "grad_norm": 0.24551476144593656,
      "learning_rate": 6.876435751902429e-08,
      "loss": 1.5914,
      "step": 5762
    },
    {
      "epoch": 0.8691652213256919,
      "grad_norm": 0.2555129218654098,
      "learning_rate": 6.872175605376895e-08,
      "loss": 1.552,
      "step": 5763
    },
    {
      "epoch": 0.8693160395143654,
      "grad_norm": 0.24750220376331367,
      "learning_rate": 6.867920090398691e-08,
      "loss": 1.5317,
      "step": 5764
    },
    {
      "epoch": 0.869466857703039,
      "grad_norm": 0.25743591918926084,
      "learning_rate": 6.863669207923304e-08,
      "loss": 1.5233,
      "step": 5765
    },
    {
      "epoch": 0.8696176758917126,
      "grad_norm": 0.2628750595949352,
      "learning_rate": 6.859422958905168e-08,
      "loss": 1.5431,
      "step": 5766
    },
    {
      "epoch": 0.8697684940803861,
      "grad_norm": 0.2526559671701926,
      "learning_rate": 6.8551813442977e-08,
      "loss": 1.5568,
      "step": 5767
    },
    {
      "epoch": 0.8699193122690596,
      "grad_norm": 0.2608591221468531,
      "learning_rate": 6.850944365053267e-08,
      "loss": 1.577,
      "step": 5768
    },
    {
      "epoch": 0.8700701304577332,
      "grad_norm": 0.24115166603471838,
      "learning_rate": 6.846712022123188e-08,
      "loss": 1.5662,
      "step": 5769
    },
    {
      "epoch": 0.8702209486464068,
      "grad_norm": 0.25783684286768616,
      "learning_rate": 6.84248431645775e-08,
      "loss": 1.5063,
      "step": 5770
    },
    {
      "epoch": 0.8703717668350803,
      "grad_norm": 0.25905139910313585,
      "learning_rate": 6.838261249006201e-08,
      "loss": 1.5644,
      "step": 5771
    },
    {
      "epoch": 0.8705225850237539,
      "grad_norm": 0.3336424296277969,
      "learning_rate": 6.834042820716731e-08,
      "loss": 1.5495,
      "step": 5772
    },
    {
      "epoch": 0.8706734032124274,
      "grad_norm": 0.25415718407660476,
      "learning_rate": 6.829829032536512e-08,
      "loss": 1.5935,
      "step": 5773
    },
    {
      "epoch": 0.8708242214011009,
      "grad_norm": 0.2627446059365829,
      "learning_rate": 6.825619885411657e-08,
      "loss": 1.5677,
      "step": 5774
    },
    {
      "epoch": 0.8709750395897745,
      "grad_norm": 0.27321392533057337,
      "learning_rate": 6.82141538028724e-08,
      "loss": 1.603,
      "step": 5775
    },
    {
      "epoch": 0.8711258577784481,
      "grad_norm": 0.49729836523555004,
      "learning_rate": 6.817215518107296e-08,
      "loss": 1.5617,
      "step": 5776
    },
    {
      "epoch": 0.8712766759671217,
      "grad_norm": 0.25681974644368083,
      "learning_rate": 6.813020299814821e-08,
      "loss": 1.5916,
      "step": 5777
    },
    {
      "epoch": 0.8714274941557952,
      "grad_norm": 0.2599369277347524,
      "learning_rate": 6.80882972635176e-08,
      "loss": 1.5781,
      "step": 5778
    },
    {
      "epoch": 0.8715783123444687,
      "grad_norm": 0.7437037849338393,
      "learning_rate": 6.804643798659014e-08,
      "loss": 1.5506,
      "step": 5779
    },
    {
      "epoch": 0.8717291305331423,
      "grad_norm": 0.24341262204991174,
      "learning_rate": 6.800462517676456e-08,
      "loss": 1.5698,
      "step": 5780
    },
    {
      "epoch": 0.8718799487218158,
      "grad_norm": 0.28515803115037064,
      "learning_rate": 6.796285884342899e-08,
      "loss": 1.5737,
      "step": 5781
    },
    {
      "epoch": 0.8720307669104894,
      "grad_norm": 0.24612164199905281,
      "learning_rate": 6.792113899596117e-08,
      "loss": 1.5673,
      "step": 5782
    },
    {
      "epoch": 0.872181585099163,
      "grad_norm": 0.25811193708811464,
      "learning_rate": 6.787946564372847e-08,
      "loss": 1.5352,
      "step": 5783
    },
    {
      "epoch": 0.8723324032878366,
      "grad_norm": 0.24997846469445648,
      "learning_rate": 6.783783879608776e-08,
      "loss": 1.6021,
      "step": 5784
    },
    {
      "epoch": 0.87248322147651,
      "grad_norm": 0.25787697298968776,
      "learning_rate": 6.779625846238543e-08,
      "loss": 1.542,
      "step": 5785
    },
    {
      "epoch": 0.8726340396651836,
      "grad_norm": 0.24797558138876954,
      "learning_rate": 6.775472465195755e-08,
      "loss": 1.5959,
      "step": 5786
    },
    {
      "epoch": 0.8727848578538572,
      "grad_norm": 0.24589188754588415,
      "learning_rate": 6.771323737412964e-08,
      "loss": 1.6038,
      "step": 5787
    },
    {
      "epoch": 0.8729356760425308,
      "grad_norm": 0.24759771757639476,
      "learning_rate": 6.767179663821677e-08,
      "loss": 1.5319,
      "step": 5788
    },
    {
      "epoch": 0.8730864942312043,
      "grad_norm": 0.24471607955377447,
      "learning_rate": 6.763040245352363e-08,
      "loss": 1.5651,
      "step": 5789
    },
    {
      "epoch": 0.8732373124198778,
      "grad_norm": 0.27370792316214204,
      "learning_rate": 6.75890548293444e-08,
      "loss": 1.5699,
      "step": 5790
    },
    {
      "epoch": 0.8733881306085514,
      "grad_norm": 0.24812544496721092,
      "learning_rate": 6.754775377496287e-08,
      "loss": 1.5811,
      "step": 5791
    },
    {
      "epoch": 0.8735389487972249,
      "grad_norm": 0.24741161438852477,
      "learning_rate": 6.750649929965226e-08,
      "loss": 1.552,
      "step": 5792
    },
    {
      "epoch": 0.8736897669858985,
      "grad_norm": 0.24811467249527291,
      "learning_rate": 6.746529141267546e-08,
      "loss": 1.5063,
      "step": 5793
    },
    {
      "epoch": 0.8738405851745721,
      "grad_norm": 0.29930632459785994,
      "learning_rate": 6.742413012328479e-08,
      "loss": 1.5434,
      "step": 5794
    },
    {
      "epoch": 0.8739914033632457,
      "grad_norm": 0.276737500035982,
      "learning_rate": 6.738301544072216e-08,
      "loss": 1.5531,
      "step": 5795
    },
    {
      "epoch": 0.8741422215519191,
      "grad_norm": 0.2597364745992147,
      "learning_rate": 6.734194737421909e-08,
      "loss": 1.5429,
      "step": 5796
    },
    {
      "epoch": 0.8742930397405927,
      "grad_norm": 0.26515093591468036,
      "learning_rate": 6.730092593299652e-08,
      "loss": 1.5143,
      "step": 5797
    },
    {
      "epoch": 0.8744438579292663,
      "grad_norm": 0.3663922869437447,
      "learning_rate": 6.725995112626489e-08,
      "loss": 1.606,
      "step": 5798
    },
    {
      "epoch": 0.8745946761179398,
      "grad_norm": 0.3252535706789248,
      "learning_rate": 6.721902296322432e-08,
      "loss": 1.437,
      "step": 5799
    },
    {
      "epoch": 0.8747454943066134,
      "grad_norm": 0.2568741329223731,
      "learning_rate": 6.717814145306436e-08,
      "loss": 1.5478,
      "step": 5800
    },
    {
      "epoch": 0.8748963124952869,
      "grad_norm": 0.24996302895460534,
      "learning_rate": 6.713730660496409e-08,
      "loss": 1.5551,
      "step": 5801
    },
    {
      "epoch": 0.8750471306839605,
      "grad_norm": 0.2545392985313996,
      "learning_rate": 6.709651842809205e-08,
      "loss": 1.6085,
      "step": 5802
    },
    {
      "epoch": 0.875197948872634,
      "grad_norm": 0.25044393006807736,
      "learning_rate": 6.70557769316065e-08,
      "loss": 1.503,
      "step": 5803
    },
    {
      "epoch": 0.8753487670613076,
      "grad_norm": 0.24949983065925968,
      "learning_rate": 6.701508212465503e-08,
      "loss": 1.5185,
      "step": 5804
    },
    {
      "epoch": 0.8754995852499812,
      "grad_norm": 0.24634745287309373,
      "learning_rate": 6.697443401637479e-08,
      "loss": 1.53,
      "step": 5805
    },
    {
      "epoch": 0.8756504034386547,
      "grad_norm": 0.2546824800620017,
      "learning_rate": 6.69338326158925e-08,
      "loss": 1.5123,
      "step": 5806
    },
    {
      "epoch": 0.8758012216273282,
      "grad_norm": 0.36811860514044387,
      "learning_rate": 6.689327793232436e-08,
      "loss": 1.5241,
      "step": 5807
    },
    {
      "epoch": 0.8759520398160018,
      "grad_norm": 0.2606483365818859,
      "learning_rate": 6.685276997477603e-08,
      "loss": 1.5727,
      "step": 5808
    },
    {
      "epoch": 0.8761028580046754,
      "grad_norm": 0.2734831217011574,
      "learning_rate": 6.681230875234277e-08,
      "loss": 1.5894,
      "step": 5809
    },
    {
      "epoch": 0.8762536761933489,
      "grad_norm": 0.3024542050398058,
      "learning_rate": 6.677189427410932e-08,
      "loss": 1.6002,
      "step": 5810
    },
    {
      "epoch": 0.8764044943820225,
      "grad_norm": 0.2668761682946003,
      "learning_rate": 6.673152654914985e-08,
      "loss": 1.5581,
      "step": 5811
    },
    {
      "epoch": 0.8765553125706961,
      "grad_norm": 0.2625276781394117,
      "learning_rate": 6.669120558652811e-08,
      "loss": 1.6457,
      "step": 5812
    },
    {
      "epoch": 0.8767061307593695,
      "grad_norm": 0.24232071568159685,
      "learning_rate": 6.665093139529735e-08,
      "loss": 1.5521,
      "step": 5813
    },
    {
      "epoch": 0.8768569489480431,
      "grad_norm": 0.2624439848241586,
      "learning_rate": 6.661070398450031e-08,
      "loss": 1.5716,
      "step": 5814
    },
    {
      "epoch": 0.8770077671367167,
      "grad_norm": 0.25094127590812293,
      "learning_rate": 6.657052336316915e-08,
      "loss": 1.6076,
      "step": 5815
    },
    {
      "epoch": 0.8771585853253903,
      "grad_norm": 0.25558888200209967,
      "learning_rate": 6.653038954032567e-08,
      "loss": 1.5274,
      "step": 5816
    },
    {
      "epoch": 0.8773094035140638,
      "grad_norm": 0.28301639685082025,
      "learning_rate": 6.649030252498103e-08,
      "loss": 1.6492,
      "step": 5817
    },
    {
      "epoch": 0.8774602217027373,
      "grad_norm": 0.2522194226059869,
      "learning_rate": 6.645026232613591e-08,
      "loss": 1.49,
      "step": 5818
    },
    {
      "epoch": 0.8776110398914109,
      "grad_norm": 0.25572083691157027,
      "learning_rate": 6.64102689527806e-08,
      "loss": 1.5293,
      "step": 5819
    },
    {
      "epoch": 0.8777618580800844,
      "grad_norm": 0.2575052070493476,
      "learning_rate": 6.637032241389469e-08,
      "loss": 1.6022,
      "step": 5820
    },
    {
      "epoch": 0.877912676268758,
      "grad_norm": 0.2473719042944877,
      "learning_rate": 6.633042271844735e-08,
      "loss": 1.52,
      "step": 5821
    },
    {
      "epoch": 0.8780634944574316,
      "grad_norm": 0.26112198892787164,
      "learning_rate": 6.629056987539724e-08,
      "loss": 1.5811,
      "step": 5822
    },
    {
      "epoch": 0.8782143126461052,
      "grad_norm": 0.26605878544749945,
      "learning_rate": 6.625076389369253e-08,
      "loss": 1.532,
      "step": 5823
    },
    {
      "epoch": 0.8783651308347786,
      "grad_norm": 0.2705578409821992,
      "learning_rate": 6.621100478227074e-08,
      "loss": 1.5703,
      "step": 5824
    },
    {
      "epoch": 0.8785159490234522,
      "grad_norm": 0.3073932781527974,
      "learning_rate": 6.617129255005897e-08,
      "loss": 1.5783,
      "step": 5825
    },
    {
      "epoch": 0.8786667672121258,
      "grad_norm": 0.24982712134015259,
      "learning_rate": 6.613162720597379e-08,
      "loss": 1.539,
      "step": 5826
    },
    {
      "epoch": 0.8788175854007994,
      "grad_norm": 0.2527065045533689,
      "learning_rate": 6.609200875892125e-08,
      "loss": 1.5615,
      "step": 5827
    },
    {
      "epoch": 0.8789684035894729,
      "grad_norm": 0.2579373139291263,
      "learning_rate": 6.605243721779679e-08,
      "loss": 1.499,
      "step": 5828
    },
    {
      "epoch": 0.8791192217781465,
      "grad_norm": 0.2577578580334874,
      "learning_rate": 6.601291259148542e-08,
      "loss": 1.6307,
      "step": 5829
    },
    {
      "epoch": 0.87927003996682,
      "grad_norm": 0.4646593161063779,
      "learning_rate": 6.597343488886152e-08,
      "loss": 1.592,
      "step": 5830
    },
    {
      "epoch": 0.8794208581554935,
      "grad_norm": 0.24590121517129615,
      "learning_rate": 6.593400411878904e-08,
      "loss": 1.5284,
      "step": 5831
    },
    {
      "epoch": 0.8795716763441671,
      "grad_norm": 0.2546596404400722,
      "learning_rate": 6.58946202901213e-08,
      "loss": 1.5823,
      "step": 5832
    },
    {
      "epoch": 0.8797224945328407,
      "grad_norm": 0.24785867246068702,
      "learning_rate": 6.585528341170113e-08,
      "loss": 1.6144,
      "step": 5833
    },
    {
      "epoch": 0.8798733127215143,
      "grad_norm": 0.257903523205089,
      "learning_rate": 6.58159934923608e-08,
      "loss": 1.6212,
      "step": 5834
    },
    {
      "epoch": 0.8800241309101877,
      "grad_norm": 0.28376176827500943,
      "learning_rate": 6.577675054092204e-08,
      "loss": 1.5282,
      "step": 5835
    },
    {
      "epoch": 0.8801749490988613,
      "grad_norm": 0.2413382149580761,
      "learning_rate": 6.573755456619605e-08,
      "loss": 1.5698,
      "step": 5836
    },
    {
      "epoch": 0.8803257672875349,
      "grad_norm": 0.841519277052757,
      "learning_rate": 6.569840557698348e-08,
      "loss": 1.56,
      "step": 5837
    },
    {
      "epoch": 0.8804765854762084,
      "grad_norm": 0.24583507349649472,
      "learning_rate": 6.565930358207432e-08,
      "loss": 1.5974,
      "step": 5838
    },
    {
      "epoch": 0.880627403664882,
      "grad_norm": 0.25423698271210005,
      "learning_rate": 6.562024859024823e-08,
      "loss": 1.6486,
      "step": 5839
    },
    {
      "epoch": 0.8807782218535556,
      "grad_norm": 0.2737973890941905,
      "learning_rate": 6.558124061027416e-08,
      "loss": 1.5341,
      "step": 5840
    },
    {
      "epoch": 0.880929040042229,
      "grad_norm": 0.23529592907654184,
      "learning_rate": 6.554227965091046e-08,
      "loss": 1.5336,
      "step": 5841
    },
    {
      "epoch": 0.8810798582309026,
      "grad_norm": 0.2706876439468758,
      "learning_rate": 6.550336572090511e-08,
      "loss": 1.5282,
      "step": 5842
    },
    {
      "epoch": 0.8812306764195762,
      "grad_norm": 0.25116041596406796,
      "learning_rate": 6.546449882899538e-08,
      "loss": 1.5206,
      "step": 5843
    },
    {
      "epoch": 0.8813814946082498,
      "grad_norm": 0.2541844979530065,
      "learning_rate": 6.542567898390796e-08,
      "loss": 1.5461,
      "step": 5844
    },
    {
      "epoch": 0.8815323127969233,
      "grad_norm": 0.2602737220053533,
      "learning_rate": 6.538690619435913e-08,
      "loss": 1.5829,
      "step": 5845
    },
    {
      "epoch": 0.8816831309855968,
      "grad_norm": 0.5266348515071663,
      "learning_rate": 6.534818046905448e-08,
      "loss": 1.6131,
      "step": 5846
    },
    {
      "epoch": 0.8818339491742704,
      "grad_norm": 0.23945077495057282,
      "learning_rate": 6.5309501816689e-08,
      "loss": 1.5225,
      "step": 5847
    },
    {
      "epoch": 0.881984767362944,
      "grad_norm": 0.24306577197166404,
      "learning_rate": 6.527087024594727e-08,
      "loss": 1.6451,
      "step": 5848
    },
    {
      "epoch": 0.8821355855516175,
      "grad_norm": 0.2511505414286209,
      "learning_rate": 6.523228576550313e-08,
      "loss": 1.5301,
      "step": 5849
    },
    {
      "epoch": 0.8822864037402911,
      "grad_norm": 0.3157981994234282,
      "learning_rate": 6.519374838401996e-08,
      "loss": 1.547,
      "step": 5850
    },
    {
      "epoch": 0.8824372219289647,
      "grad_norm": 0.24197185158085258,
      "learning_rate": 6.51552581101505e-08,
      "loss": 1.5305,
      "step": 5851
    },
    {
      "epoch": 0.8825880401176381,
      "grad_norm": 0.2780996962107519,
      "learning_rate": 6.511681495253696e-08,
      "loss": 1.5642,
      "step": 5852
    },
    {
      "epoch": 0.8827388583063117,
      "grad_norm": 0.25352910001836493,
      "learning_rate": 6.507841891981095e-08,
      "loss": 1.534,
      "step": 5853
    },
    {
      "epoch": 0.8828896764949853,
      "grad_norm": 0.25044494272911494,
      "learning_rate": 6.504007002059343e-08,
      "loss": 1.5529,
      "step": 5854
    },
    {
      "epoch": 0.8830404946836589,
      "grad_norm": 0.3117023973963776,
      "learning_rate": 6.500176826349495e-08,
      "loss": 1.5422,
      "step": 5855
    },
    {
      "epoch": 0.8831913128723324,
      "grad_norm": 0.2442724461387855,
      "learning_rate": 6.496351365711531e-08,
      "loss": 1.5985,
      "step": 5856
    },
    {
      "epoch": 0.883342131061006,
      "grad_norm": 0.270081945555682,
      "learning_rate": 6.492530621004377e-08,
      "loss": 1.494,
      "step": 5857
    },
    {
      "epoch": 0.8834929492496795,
      "grad_norm": 0.2752048956448727,
      "learning_rate": 6.488714593085907e-08,
      "loss": 1.5454,
      "step": 5858
    },
    {
      "epoch": 0.883643767438353,
      "grad_norm": 0.25020273816589,
      "learning_rate": 6.484903282812927e-08,
      "loss": 1.5419,
      "step": 5859
    },
    {
      "epoch": 0.8837945856270266,
      "grad_norm": 0.2552255559663703,
      "learning_rate": 6.481096691041184e-08,
      "loss": 1.5507,
      "step": 5860
    },
    {
      "epoch": 0.8839454038157002,
      "grad_norm": 0.24406927446969745,
      "learning_rate": 6.477294818625375e-08,
      "loss": 1.5815,
      "step": 5861
    },
    {
      "epoch": 0.8840962220043738,
      "grad_norm": 0.2691424640015323,
      "learning_rate": 6.473497666419127e-08,
      "loss": 1.5694,
      "step": 5862
    },
    {
      "epoch": 0.8842470401930472,
      "grad_norm": 0.27891157110759063,
      "learning_rate": 6.469705235275016e-08,
      "loss": 1.5966,
      "step": 5863
    },
    {
      "epoch": 0.8843978583817208,
      "grad_norm": 0.4048163210053456,
      "learning_rate": 6.465917526044541e-08,
      "loss": 1.6641,
      "step": 5864
    },
    {
      "epoch": 0.8845486765703944,
      "grad_norm": 0.2517174592052388,
      "learning_rate": 6.462134539578169e-08,
      "loss": 1.6265,
      "step": 5865
    },
    {
      "epoch": 0.884699494759068,
      "grad_norm": 0.2535782327644837,
      "learning_rate": 6.458356276725281e-08,
      "loss": 1.5737,
      "step": 5866
    },
    {
      "epoch": 0.8848503129477415,
      "grad_norm": 0.24205301714222377,
      "learning_rate": 6.454582738334208e-08,
      "loss": 1.589,
      "step": 5867
    },
    {
      "epoch": 0.8850011311364151,
      "grad_norm": 0.2564795886088914,
      "learning_rate": 6.450813925252221e-08,
      "loss": 1.5431,
      "step": 5868
    },
    {
      "epoch": 0.8851519493250886,
      "grad_norm": 0.4133754335843493,
      "learning_rate": 6.447049838325529e-08,
      "loss": 1.5507,
      "step": 5869
    },
    {
      "epoch": 0.8853027675137621,
      "grad_norm": 0.28005206464773075,
      "learning_rate": 6.443290478399276e-08,
      "loss": 1.5409,
      "step": 5870
    },
    {
      "epoch": 0.8854535857024357,
      "grad_norm": 0.5400376931836294,
      "learning_rate": 6.439535846317553e-08,
      "loss": 1.5197,
      "step": 5871
    },
    {
      "epoch": 0.8856044038911093,
      "grad_norm": 0.2735622427700534,
      "learning_rate": 6.435785942923379e-08,
      "loss": 1.6257,
      "step": 5872
    },
    {
      "epoch": 0.8857552220797829,
      "grad_norm": 0.26715549035195923,
      "learning_rate": 6.432040769058716e-08,
      "loss": 1.541,
      "step": 5873
    },
    {
      "epoch": 0.8859060402684564,
      "grad_norm": 0.2550453636354242,
      "learning_rate": 6.42830032556447e-08,
      "loss": 1.6178,
      "step": 5874
    },
    {
      "epoch": 0.8860568584571299,
      "grad_norm": 0.24667162238930262,
      "learning_rate": 6.424564613280478e-08,
      "loss": 1.566,
      "step": 5875
    },
    {
      "epoch": 0.8862076766458035,
      "grad_norm": 0.25575405213987346,
      "learning_rate": 6.420833633045514e-08,
      "loss": 1.5986,
      "step": 5876
    },
    {
      "epoch": 0.886358494834477,
      "grad_norm": 0.2567939555373869,
      "learning_rate": 6.41710738569729e-08,
      "loss": 1.6104,
      "step": 5877
    },
    {
      "epoch": 0.8865093130231506,
      "grad_norm": 0.27743083094419724,
      "learning_rate": 6.413385872072462e-08,
      "loss": 1.5155,
      "step": 5878
    },
    {
      "epoch": 0.8866601312118242,
      "grad_norm": 0.25766363740830706,
      "learning_rate": 6.409669093006618e-08,
      "loss": 1.509,
      "step": 5879
    },
    {
      "epoch": 0.8868109494004977,
      "grad_norm": 0.29953726419607446,
      "learning_rate": 6.405957049334275e-08,
      "loss": 1.5885,
      "step": 5880
    },
    {
      "epoch": 0.8869617675891712,
      "grad_norm": 0.24196780546874763,
      "learning_rate": 6.402249741888905e-08,
      "loss": 1.5239,
      "step": 5881
    },
    {
      "epoch": 0.8871125857778448,
      "grad_norm": 0.24233031263300328,
      "learning_rate": 6.398547171502903e-08,
      "loss": 1.464,
      "step": 5882
    },
    {
      "epoch": 0.8872634039665184,
      "grad_norm": 0.24816442235875175,
      "learning_rate": 6.3948493390076e-08,
      "loss": 1.5234,
      "step": 5883
    },
    {
      "epoch": 0.887414222155192,
      "grad_norm": 0.2480019620615259,
      "learning_rate": 6.391156245233272e-08,
      "loss": 1.5351,
      "step": 5884
    },
    {
      "epoch": 0.8875650403438655,
      "grad_norm": 0.2560935364839055,
      "learning_rate": 6.387467891009125e-08,
      "loss": 1.6397,
      "step": 5885
    },
    {
      "epoch": 0.887715858532539,
      "grad_norm": 0.269378814178076,
      "learning_rate": 6.383784277163299e-08,
      "loss": 1.5915,
      "step": 5886
    },
    {
      "epoch": 0.8878666767212126,
      "grad_norm": 0.2667927821700778,
      "learning_rate": 6.380105404522875e-08,
      "loss": 1.6394,
      "step": 5887
    },
    {
      "epoch": 0.8880174949098861,
      "grad_norm": 0.24908878957242192,
      "learning_rate": 6.37643127391387e-08,
      "loss": 1.6094,
      "step": 5888
    },
    {
      "epoch": 0.8881683130985597,
      "grad_norm": 0.4882189504331334,
      "learning_rate": 6.372761886161231e-08,
      "loss": 1.5773,
      "step": 5889
    },
    {
      "epoch": 0.8883191312872333,
      "grad_norm": 0.25313376930267606,
      "learning_rate": 6.369097242088837e-08,
      "loss": 1.5588,
      "step": 5890
    },
    {
      "epoch": 0.8884699494759067,
      "grad_norm": 0.2658884687807718,
      "learning_rate": 6.365437342519517e-08,
      "loss": 1.523,
      "step": 5891
    },
    {
      "epoch": 0.8886207676645803,
      "grad_norm": 0.2529329522885053,
      "learning_rate": 6.361782188275018e-08,
      "loss": 1.5984,
      "step": 5892
    },
    {
      "epoch": 0.8887715858532539,
      "grad_norm": 0.2774872541482239,
      "learning_rate": 6.358131780176028e-08,
      "loss": 1.5754,
      "step": 5893
    },
    {
      "epoch": 0.8889224040419275,
      "grad_norm": 0.2421467165011948,
      "learning_rate": 6.354486119042178e-08,
      "loss": 1.485,
      "step": 5894
    },
    {
      "epoch": 0.889073222230601,
      "grad_norm": 0.2715653990408799,
      "learning_rate": 6.350845205692016e-08,
      "loss": 1.4851,
      "step": 5895
    },
    {
      "epoch": 0.8892240404192746,
      "grad_norm": 0.32996512731212596,
      "learning_rate": 6.347209040943034e-08,
      "loss": 1.5237,
      "step": 5896
    },
    {
      "epoch": 0.8893748586079481,
      "grad_norm": 0.2811997761623907,
      "learning_rate": 6.343577625611666e-08,
      "loss": 1.5832,
      "step": 5897
    },
    {
      "epoch": 0.8895256767966216,
      "grad_norm": 0.25437689079194636,
      "learning_rate": 6.339950960513261e-08,
      "loss": 1.5377,
      "step": 5898
    },
    {
      "epoch": 0.8896764949852952,
      "grad_norm": 0.24800163959629032,
      "learning_rate": 6.336329046462115e-08,
      "loss": 1.4737,
      "step": 5899
    },
    {
      "epoch": 0.8898273131739688,
      "grad_norm": 0.28929413175546753,
      "learning_rate": 6.332711884271452e-08,
      "loss": 1.5879,
      "step": 5900
    },
    {
      "epoch": 0.8899781313626424,
      "grad_norm": 0.2530746149930727,
      "learning_rate": 6.32909947475343e-08,
      "loss": 1.5705,
      "step": 5901
    },
    {
      "epoch": 0.8901289495513159,
      "grad_norm": 0.262257833317664,
      "learning_rate": 6.325491818719144e-08,
      "loss": 1.5289,
      "step": 5902
    },
    {
      "epoch": 0.8902797677399894,
      "grad_norm": 0.2548884487560699,
      "learning_rate": 6.321888916978609e-08,
      "loss": 1.5929,
      "step": 5903
    },
    {
      "epoch": 0.890430585928663,
      "grad_norm": 0.25979559417848014,
      "learning_rate": 6.31829077034079e-08,
      "loss": 1.6308,
      "step": 5904
    },
    {
      "epoch": 0.8905814041173366,
      "grad_norm": 0.24462400630588177,
      "learning_rate": 6.314697379613569e-08,
      "loss": 1.5462,
      "step": 5905
    },
    {
      "epoch": 0.8907322223060101,
      "grad_norm": 0.25649500321946517,
      "learning_rate": 6.311108745603775e-08,
      "loss": 1.6261,
      "step": 5906
    },
    {
      "epoch": 0.8908830404946837,
      "grad_norm": 0.3051374242499988,
      "learning_rate": 6.307524869117154e-08,
      "loss": 1.5937,
      "step": 5907
    },
    {
      "epoch": 0.8910338586833572,
      "grad_norm": 0.26413628832939406,
      "learning_rate": 6.303945750958393e-08,
      "loss": 1.5781,
      "step": 5908
    },
    {
      "epoch": 0.8911846768720307,
      "grad_norm": 0.2699633938984024,
      "learning_rate": 6.300371391931113e-08,
      "loss": 1.5348,
      "step": 5909
    },
    {
      "epoch": 0.8913354950607043,
      "grad_norm": 0.28555157419570487,
      "learning_rate": 6.296801792837848e-08,
      "loss": 1.4669,
      "step": 5910
    },
    {
      "epoch": 0.8914863132493779,
      "grad_norm": 0.48857873667811774,
      "learning_rate": 6.293236954480092e-08,
      "loss": 1.5307,
      "step": 5911
    },
    {
      "epoch": 0.8916371314380515,
      "grad_norm": 0.23834644553922518,
      "learning_rate": 6.28967687765825e-08,
      "loss": 1.5111,
      "step": 5912
    },
    {
      "epoch": 0.891787949626725,
      "grad_norm": 0.26603501578231303,
      "learning_rate": 6.286121563171657e-08,
      "loss": 1.5718,
      "step": 5913
    },
    {
      "epoch": 0.8919387678153985,
      "grad_norm": 0.23842106316087344,
      "learning_rate": 6.282571011818594e-08,
      "loss": 1.4927,
      "step": 5914
    },
    {
      "epoch": 0.8920895860040721,
      "grad_norm": 0.29860749122389607,
      "learning_rate": 6.279025224396257e-08,
      "loss": 1.5237,
      "step": 5915
    },
    {
      "epoch": 0.8922404041927456,
      "grad_norm": 0.36973073453293803,
      "learning_rate": 6.275484201700779e-08,
      "loss": 1.5553,
      "step": 5916
    },
    {
      "epoch": 0.8923912223814192,
      "grad_norm": 0.26957002690688664,
      "learning_rate": 6.271947944527226e-08,
      "loss": 1.5176,
      "step": 5917
    },
    {
      "epoch": 0.8925420405700928,
      "grad_norm": 0.24833599270790296,
      "learning_rate": 6.268416453669588e-08,
      "loss": 1.6262,
      "step": 5918
    },
    {
      "epoch": 0.8926928587587664,
      "grad_norm": 0.26736278994547186,
      "learning_rate": 6.264889729920788e-08,
      "loss": 1.4807,
      "step": 5919
    },
    {
      "epoch": 0.8928436769474398,
      "grad_norm": 0.2528725015623935,
      "learning_rate": 6.261367774072677e-08,
      "loss": 1.5732,
      "step": 5920
    },
    {
      "epoch": 0.8929944951361134,
      "grad_norm": 0.25935914936506105,
      "learning_rate": 6.25785058691604e-08,
      "loss": 1.4699,
      "step": 5921
    },
    {
      "epoch": 0.893145313324787,
      "grad_norm": 0.25700196910473727,
      "learning_rate": 6.254338169240588e-08,
      "loss": 1.5718,
      "step": 5922
    },
    {
      "epoch": 0.8932961315134605,
      "grad_norm": 0.2612362352666459,
      "learning_rate": 6.250830521834953e-08,
      "loss": 1.4954,
      "step": 5923
    },
    {
      "epoch": 0.8934469497021341,
      "grad_norm": 0.2398806807867513,
      "learning_rate": 6.247327645486715e-08,
      "loss": 1.5542,
      "step": 5924
    },
    {
      "epoch": 0.8935977678908076,
      "grad_norm": 0.24741778039983103,
      "learning_rate": 6.243829540982365e-08,
      "loss": 1.5585,
      "step": 5925
    },
    {
      "epoch": 0.8937485860794812,
      "grad_norm": 0.2533496331004564,
      "learning_rate": 6.24033620910733e-08,
      "loss": 1.5317,
      "step": 5926
    },
    {
      "epoch": 0.8938994042681547,
      "grad_norm": 0.24478317756933246,
      "learning_rate": 6.236847650645969e-08,
      "loss": 1.5864,
      "step": 5927
    },
    {
      "epoch": 0.8940502224568283,
      "grad_norm": 0.25150332003977416,
      "learning_rate": 6.233363866381562e-08,
      "loss": 1.5466,
      "step": 5928
    },
    {
      "epoch": 0.8942010406455019,
      "grad_norm": 0.2718123301282069,
      "learning_rate": 6.229884857096318e-08,
      "loss": 1.5986,
      "step": 5929
    },
    {
      "epoch": 0.8943518588341755,
      "grad_norm": 0.4184994268005829,
      "learning_rate": 6.226410623571375e-08,
      "loss": 1.6133,
      "step": 5930
    },
    {
      "epoch": 0.8945026770228489,
      "grad_norm": 0.2862373432969868,
      "learning_rate": 6.222941166586807e-08,
      "loss": 1.569,
      "step": 5931
    },
    {
      "epoch": 0.8946534952115225,
      "grad_norm": 0.28116572971179365,
      "learning_rate": 6.219476486921604e-08,
      "loss": 1.5196,
      "step": 5932
    },
    {
      "epoch": 0.8948043134001961,
      "grad_norm": 0.27920990887066083,
      "learning_rate": 6.216016585353684e-08,
      "loss": 1.5565,
      "step": 5933
    },
    {
      "epoch": 0.8949551315888696,
      "grad_norm": 0.9468327930177274,
      "learning_rate": 6.212561462659899e-08,
      "loss": 1.4884,
      "step": 5934
    },
    {
      "epoch": 0.8951059497775432,
      "grad_norm": 0.24085009761212245,
      "learning_rate": 6.209111119616026e-08,
      "loss": 1.5676,
      "step": 5935
    },
    {
      "epoch": 0.8952567679662167,
      "grad_norm": 0.24640198307092454,
      "learning_rate": 6.205665556996763e-08,
      "loss": 1.5443,
      "step": 5936
    },
    {
      "epoch": 0.8954075861548902,
      "grad_norm": 0.244946177856438,
      "learning_rate": 6.202224775575743e-08,
      "loss": 1.5786,
      "step": 5937
    },
    {
      "epoch": 0.8955584043435638,
      "grad_norm": 0.23629801333322237,
      "learning_rate": 6.198788776125523e-08,
      "loss": 1.5061,
      "step": 5938
    },
    {
      "epoch": 0.8957092225322374,
      "grad_norm": 0.24437752940312316,
      "learning_rate": 6.195357559417579e-08,
      "loss": 1.5267,
      "step": 5939
    },
    {
      "epoch": 0.895860040720911,
      "grad_norm": 1.1115769329679674,
      "learning_rate": 6.191931126222327e-08,
      "loss": 1.5187,
      "step": 5940
    },
    {
      "epoch": 0.8960108589095845,
      "grad_norm": 0.25349389272698614,
      "learning_rate": 6.188509477309095e-08,
      "loss": 1.5622,
      "step": 5941
    },
    {
      "epoch": 0.896161677098258,
      "grad_norm": 0.45245419843398993,
      "learning_rate": 6.185092613446144e-08,
      "loss": 1.5551,
      "step": 5942
    },
    {
      "epoch": 0.8963124952869316,
      "grad_norm": 0.27820637443519197,
      "learning_rate": 6.181680535400658e-08,
      "loss": 1.6205,
      "step": 5943
    },
    {
      "epoch": 0.8964633134756052,
      "grad_norm": 0.26015948602270705,
      "learning_rate": 6.178273243938754e-08,
      "loss": 1.5722,
      "step": 5944
    },
    {
      "epoch": 0.8966141316642787,
      "grad_norm": 0.2531641812074329,
      "learning_rate": 6.174870739825462e-08,
      "loss": 1.5779,
      "step": 5945
    },
    {
      "epoch": 0.8967649498529523,
      "grad_norm": 0.28896923356037835,
      "learning_rate": 6.171473023824745e-08,
      "loss": 1.5222,
      "step": 5946
    },
    {
      "epoch": 0.8969157680416259,
      "grad_norm": 0.257840125587561,
      "learning_rate": 6.168080096699491e-08,
      "loss": 1.5753,
      "step": 5947
    },
    {
      "epoch": 0.8970665862302993,
      "grad_norm": 0.2601253932992487,
      "learning_rate": 6.164691959211508e-08,
      "loss": 1.5733,
      "step": 5948
    },
    {
      "epoch": 0.8972174044189729,
      "grad_norm": 0.2693931957959471,
      "learning_rate": 6.161308612121533e-08,
      "loss": 1.5236,
      "step": 5949
    },
    {
      "epoch": 0.8973682226076465,
      "grad_norm": 0.4050835473138554,
      "learning_rate": 6.157930056189226e-08,
      "loss": 1.565,
      "step": 5950
    },
    {
      "epoch": 0.8975190407963201,
      "grad_norm": 0.24394173590491194,
      "learning_rate": 6.154556292173172e-08,
      "loss": 1.4795,
      "step": 5951
    },
    {
      "epoch": 0.8976698589849936,
      "grad_norm": 0.26346869726216715,
      "learning_rate": 6.151187320830875e-08,
      "loss": 1.541,
      "step": 5952
    },
    {
      "epoch": 0.8978206771736671,
      "grad_norm": 0.25805070681206316,
      "learning_rate": 6.147823142918775e-08,
      "loss": 1.531,
      "step": 5953
    },
    {
      "epoch": 0.8979714953623407,
      "grad_norm": 0.24566885665807253,
      "learning_rate": 6.144463759192224e-08,
      "loss": 1.5557,
      "step": 5954
    },
    {
      "epoch": 0.8981223135510142,
      "grad_norm": 0.2520165272941859,
      "learning_rate": 6.1411091704055e-08,
      "loss": 1.6103,
      "step": 5955
    },
    {
      "epoch": 0.8982731317396878,
      "grad_norm": 0.24379749398124398,
      "learning_rate": 6.137759377311803e-08,
      "loss": 1.4831,
      "step": 5956
    },
    {
      "epoch": 0.8984239499283614,
      "grad_norm": 0.2479661131005401,
      "learning_rate": 6.134414380663268e-08,
      "loss": 1.5335,
      "step": 5957
    },
    {
      "epoch": 0.898574768117035,
      "grad_norm": 0.3180718221807806,
      "learning_rate": 6.131074181210937e-08,
      "loss": 1.5028,
      "step": 5958
    },
    {
      "epoch": 0.8987255863057084,
      "grad_norm": 0.3582739368580261,
      "learning_rate": 6.127738779704781e-08,
      "loss": 1.5343,
      "step": 5959
    },
    {
      "epoch": 0.898876404494382,
      "grad_norm": 0.24072622519868206,
      "learning_rate": 6.124408176893704e-08,
      "loss": 1.4762,
      "step": 5960
    },
    {
      "epoch": 0.8990272226830556,
      "grad_norm": 0.24225971651789271,
      "learning_rate": 6.121082373525516e-08,
      "loss": 1.5625,
      "step": 5961
    },
    {
      "epoch": 0.8991780408717291,
      "grad_norm": 0.2878400417245367,
      "learning_rate": 6.117761370346952e-08,
      "loss": 1.5547,
      "step": 5962
    },
    {
      "epoch": 0.8993288590604027,
      "grad_norm": 0.25902823185013885,
      "learning_rate": 6.114445168103686e-08,
      "loss": 1.5859,
      "step": 5963
    },
    {
      "epoch": 0.8994796772490763,
      "grad_norm": 0.2565759083430694,
      "learning_rate": 6.111133767540296e-08,
      "loss": 1.527,
      "step": 5964
    },
    {
      "epoch": 0.8996304954377498,
      "grad_norm": 0.25825035292048387,
      "learning_rate": 6.107827169400286e-08,
      "loss": 1.6749,
      "step": 5965
    },
    {
      "epoch": 0.8997813136264233,
      "grad_norm": 0.24646332618767774,
      "learning_rate": 6.104525374426087e-08,
      "loss": 1.5141,
      "step": 5966
    },
    {
      "epoch": 0.8999321318150969,
      "grad_norm": 0.28275712404453535,
      "learning_rate": 6.101228383359045e-08,
      "loss": 1.5549,
      "step": 5967
    },
    {
      "epoch": 0.9000829500037705,
      "grad_norm": 0.24842922555258082,
      "learning_rate": 6.097936196939434e-08,
      "loss": 1.5833,
      "step": 5968
    },
    {
      "epoch": 0.900233768192444,
      "grad_norm": 0.3588249034386644,
      "learning_rate": 6.094648815906439e-08,
      "loss": 1.5332,
      "step": 5969
    },
    {
      "epoch": 0.9003845863811175,
      "grad_norm": 0.2641650885366285,
      "learning_rate": 6.091366240998182e-08,
      "loss": 1.6196,
      "step": 5970
    },
    {
      "epoch": 0.9005354045697911,
      "grad_norm": 0.24696161829150579,
      "learning_rate": 6.088088472951694e-08,
      "loss": 1.5711,
      "step": 5971
    },
    {
      "epoch": 0.9006862227584647,
      "grad_norm": 0.24249422449486577,
      "learning_rate": 6.084815512502922e-08,
      "loss": 1.5507,
      "step": 5972
    },
    {
      "epoch": 0.9008370409471382,
      "grad_norm": 0.24074282720599494,
      "learning_rate": 6.081547360386748e-08,
      "loss": 1.5016,
      "step": 5973
    },
    {
      "epoch": 0.9009878591358118,
      "grad_norm": 0.27379028533114474,
      "learning_rate": 6.07828401733697e-08,
      "loss": 1.5807,
      "step": 5974
    },
    {
      "epoch": 0.9011386773244854,
      "grad_norm": 0.24662719532027477,
      "learning_rate": 6.075025484086294e-08,
      "loss": 1.526,
      "step": 5975
    },
    {
      "epoch": 0.9012894955131588,
      "grad_norm": 0.26352249578185005,
      "learning_rate": 6.071771761366364e-08,
      "loss": 1.5176,
      "step": 5976
    },
    {
      "epoch": 0.9014403137018324,
      "grad_norm": 0.25389674701605053,
      "learning_rate": 6.068522849907734e-08,
      "loss": 1.5477,
      "step": 5977
    },
    {
      "epoch": 0.901591131890506,
      "grad_norm": 0.2621310667880105,
      "learning_rate": 6.065278750439873e-08,
      "loss": 1.5737,
      "step": 5978
    },
    {
      "epoch": 0.9017419500791796,
      "grad_norm": 0.2616883665866148,
      "learning_rate": 6.062039463691185e-08,
      "loss": 1.5583,
      "step": 5979
    },
    {
      "epoch": 0.9018927682678531,
      "grad_norm": 0.24728104360191377,
      "learning_rate": 6.058804990388979e-08,
      "loss": 1.526,
      "step": 5980
    },
    {
      "epoch": 0.9020435864565266,
      "grad_norm": 0.2559460115090124,
      "learning_rate": 6.055575331259488e-08,
      "loss": 1.5865,
      "step": 5981
    },
    {
      "epoch": 0.9021944046452002,
      "grad_norm": 0.25760239630524945,
      "learning_rate": 6.052350487027865e-08,
      "loss": 1.57,
      "step": 5982
    },
    {
      "epoch": 0.9023452228338738,
      "grad_norm": 0.24738618795501038,
      "learning_rate": 6.049130458418183e-08,
      "loss": 1.6156,
      "step": 5983
    },
    {
      "epoch": 0.9024960410225473,
      "grad_norm": 0.26018903050986375,
      "learning_rate": 6.045915246153435e-08,
      "loss": 1.561,
      "step": 5984
    },
    {
      "epoch": 0.9026468592112209,
      "grad_norm": 0.2568670949435877,
      "learning_rate": 6.042704850955521e-08,
      "loss": 1.541,
      "step": 5985
    },
    {
      "epoch": 0.9027976773998945,
      "grad_norm": 0.25089162499423023,
      "learning_rate": 6.039499273545277e-08,
      "loss": 1.5407,
      "step": 5986
    },
    {
      "epoch": 0.9029484955885679,
      "grad_norm": 0.24381195120656876,
      "learning_rate": 6.036298514642445e-08,
      "loss": 1.5889,
      "step": 5987
    },
    {
      "epoch": 0.9030993137772415,
      "grad_norm": 0.25755710349757727,
      "learning_rate": 6.033102574965685e-08,
      "loss": 1.5078,
      "step": 5988
    },
    {
      "epoch": 0.9032501319659151,
      "grad_norm": 0.23761719968154357,
      "learning_rate": 6.029911455232585e-08,
      "loss": 1.4897,
      "step": 5989
    },
    {
      "epoch": 0.9034009501545887,
      "grad_norm": 0.24978975522047717,
      "learning_rate": 6.026725156159643e-08,
      "loss": 1.6201,
      "step": 5990
    },
    {
      "epoch": 0.9035517683432622,
      "grad_norm": 0.2730426495004039,
      "learning_rate": 6.023543678462268e-08,
      "loss": 1.5745,
      "step": 5991
    },
    {
      "epoch": 0.9037025865319358,
      "grad_norm": 0.2529225158454245,
      "learning_rate": 6.020367022854805e-08,
      "loss": 1.5771,
      "step": 5992
    },
    {
      "epoch": 0.9038534047206093,
      "grad_norm": 0.2624770484136695,
      "learning_rate": 6.017195190050502e-08,
      "loss": 1.526,
      "step": 5993
    },
    {
      "epoch": 0.9040042229092828,
      "grad_norm": 0.24752289297805732,
      "learning_rate": 6.014028180761527e-08,
      "loss": 1.5887,
      "step": 5994
    },
    {
      "epoch": 0.9041550410979564,
      "grad_norm": 0.25359951881522097,
      "learning_rate": 6.010865995698965e-08,
      "loss": 1.5637,
      "step": 5995
    },
    {
      "epoch": 0.90430585928663,
      "grad_norm": 0.24977479265531782,
      "learning_rate": 6.007708635572821e-08,
      "loss": 1.675,
      "step": 5996
    },
    {
      "epoch": 0.9044566774753036,
      "grad_norm": 0.31466477706062923,
      "learning_rate": 6.004556101092012e-08,
      "loss": 1.5225,
      "step": 5997
    },
    {
      "epoch": 0.904607495663977,
      "grad_norm": 0.24303541638011292,
      "learning_rate": 6.001408392964375e-08,
      "loss": 1.5208,
      "step": 5998
    },
    {
      "epoch": 0.9047583138526506,
      "grad_norm": 0.6739544452758239,
      "learning_rate": 5.998265511896662e-08,
      "loss": 1.5446,
      "step": 5999
    },
    {
      "epoch": 0.9049091320413242,
      "grad_norm": 0.25687482238901066,
      "learning_rate": 5.995127458594542e-08,
      "loss": 1.5913,
      "step": 6000
    },
    {
      "epoch": 0.9050599502299977,
      "grad_norm": 0.2550897498231508,
      "learning_rate": 5.991994233762597e-08,
      "loss": 1.5659,
      "step": 6001
    },
    {
      "epoch": 0.9052107684186713,
      "grad_norm": 0.24268037160138126,
      "learning_rate": 5.988865838104332e-08,
      "loss": 1.5651,
      "step": 6002
    },
    {
      "epoch": 0.9053615866073449,
      "grad_norm": 0.2731439566265382,
      "learning_rate": 5.985742272322157e-08,
      "loss": 1.5762,
      "step": 6003
    },
    {
      "epoch": 0.9055124047960184,
      "grad_norm": 0.26038850236812383,
      "learning_rate": 5.982623537117405e-08,
      "loss": 1.5946,
      "step": 6004
    },
    {
      "epoch": 0.9056632229846919,
      "grad_norm": 0.8607046633015673,
      "learning_rate": 5.979509633190327e-08,
      "loss": 1.5304,
      "step": 6005
    },
    {
      "epoch": 0.9058140411733655,
      "grad_norm": 0.25976877570104584,
      "learning_rate": 5.976400561240083e-08,
      "loss": 1.575,
      "step": 6006
    },
    {
      "epoch": 0.9059648593620391,
      "grad_norm": 0.30493413653552726,
      "learning_rate": 5.97329632196475e-08,
      "loss": 1.4739,
      "step": 6007
    },
    {
      "epoch": 0.9061156775507127,
      "grad_norm": 0.26836670517825983,
      "learning_rate": 5.970196916061314e-08,
      "loss": 1.6054,
      "step": 6008
    },
    {
      "epoch": 0.9062664957393862,
      "grad_norm": 0.23988289191523413,
      "learning_rate": 5.967102344225691e-08,
      "loss": 1.5178,
      "step": 6009
    },
    {
      "epoch": 0.9064173139280597,
      "grad_norm": 0.30141474342371144,
      "learning_rate": 5.964012607152701e-08,
      "loss": 1.5522,
      "step": 6010
    },
    {
      "epoch": 0.9065681321167333,
      "grad_norm": 0.2545175356890916,
      "learning_rate": 5.960927705536076e-08,
      "loss": 1.5099,
      "step": 6011
    },
    {
      "epoch": 0.9067189503054068,
      "grad_norm": 0.26200645955094654,
      "learning_rate": 5.957847640068467e-08,
      "loss": 1.5973,
      "step": 6012
    },
    {
      "epoch": 0.9068697684940804,
      "grad_norm": 0.25436559423044286,
      "learning_rate": 5.954772411441442e-08,
      "loss": 1.5871,
      "step": 6013
    },
    {
      "epoch": 0.907020586682754,
      "grad_norm": 0.2684570896237277,
      "learning_rate": 5.9517020203454756e-08,
      "loss": 1.5906,
      "step": 6014
    },
    {
      "epoch": 0.9071714048714274,
      "grad_norm": 0.44423987142588256,
      "learning_rate": 5.948636467469963e-08,
      "loss": 1.5618,
      "step": 6015
    },
    {
      "epoch": 0.907322223060101,
      "grad_norm": 0.2465686813729949,
      "learning_rate": 5.945575753503207e-08,
      "loss": 1.5585,
      "step": 6016
    },
    {
      "epoch": 0.9074730412487746,
      "grad_norm": 0.25285875225744253,
      "learning_rate": 5.942519879132427e-08,
      "loss": 1.566,
      "step": 6017
    },
    {
      "epoch": 0.9076238594374482,
      "grad_norm": 0.24574371258077005,
      "learning_rate": 5.93946884504376e-08,
      "loss": 1.6016,
      "step": 6018
    },
    {
      "epoch": 0.9077746776261217,
      "grad_norm": 0.24671212121303002,
      "learning_rate": 5.93642265192225e-08,
      "loss": 1.5013,
      "step": 6019
    },
    {
      "epoch": 0.9079254958147953,
      "grad_norm": 0.2734618189493248,
      "learning_rate": 5.933381300451855e-08,
      "loss": 1.5508,
      "step": 6020
    },
    {
      "epoch": 0.9080763140034688,
      "grad_norm": 0.24606026666908068,
      "learning_rate": 5.9303447913154435e-08,
      "loss": 1.5824,
      "step": 6021
    },
    {
      "epoch": 0.9082271321921424,
      "grad_norm": 0.27087054216315615,
      "learning_rate": 5.927313125194809e-08,
      "loss": 1.5711,
      "step": 6022
    },
    {
      "epoch": 0.9083779503808159,
      "grad_norm": 1.7962946270160636,
      "learning_rate": 5.924286302770642e-08,
      "loss": 1.523,
      "step": 6023
    },
    {
      "epoch": 0.9085287685694895,
      "grad_norm": 0.2423753250984084,
      "learning_rate": 5.921264324722553e-08,
      "loss": 1.5427,
      "step": 6024
    },
    {
      "epoch": 0.9086795867581631,
      "grad_norm": 0.26772680116993963,
      "learning_rate": 5.918247191729067e-08,
      "loss": 1.4842,
      "step": 6025
    },
    {
      "epoch": 0.9088304049468365,
      "grad_norm": 0.24826463739719304,
      "learning_rate": 5.915234904467617e-08,
      "loss": 1.5398,
      "step": 6026
    },
    {
      "epoch": 0.9089812231355101,
      "grad_norm": 0.29586232205972984,
      "learning_rate": 5.912227463614545e-08,
      "loss": 1.5409,
      "step": 6027
    },
    {
      "epoch": 0.9091320413241837,
      "grad_norm": 0.2599860801557601,
      "learning_rate": 5.9092248698451196e-08,
      "loss": 1.5961,
      "step": 6028
    },
    {
      "epoch": 0.9092828595128573,
      "grad_norm": 0.35821856761478,
      "learning_rate": 5.906227123833501e-08,
      "loss": 1.5799,
      "step": 6029
    },
    {
      "epoch": 0.9094336777015308,
      "grad_norm": 0.25207841952714977,
      "learning_rate": 5.9032342262527716e-08,
      "loss": 1.6431,
      "step": 6030
    },
    {
      "epoch": 0.9095844958902044,
      "grad_norm": 0.28798786232057827,
      "learning_rate": 5.900246177774926e-08,
      "loss": 1.4955,
      "step": 6031
    },
    {
      "epoch": 0.9097353140788779,
      "grad_norm": 0.24850564745268564,
      "learning_rate": 5.897262979070868e-08,
      "loss": 1.5698,
      "step": 6032
    },
    {
      "epoch": 0.9098861322675514,
      "grad_norm": 0.2941221192303597,
      "learning_rate": 5.894284630810415e-08,
      "loss": 1.5519,
      "step": 6033
    },
    {
      "epoch": 0.910036950456225,
      "grad_norm": 0.24259196567563732,
      "learning_rate": 5.8913111336622856e-08,
      "loss": 1.6027,
      "step": 6034
    },
    {
      "epoch": 0.9101877686448986,
      "grad_norm": 0.25153935046572595,
      "learning_rate": 5.888342488294125e-08,
      "loss": 1.5018,
      "step": 6035
    },
    {
      "epoch": 0.9103385868335722,
      "grad_norm": 0.24229883035317848,
      "learning_rate": 5.885378695372476e-08,
      "loss": 1.4775,
      "step": 6036
    },
    {
      "epoch": 0.9104894050222457,
      "grad_norm": 0.24241717043334063,
      "learning_rate": 5.8824197555627936e-08,
      "loss": 1.5488,
      "step": 6037
    },
    {
      "epoch": 0.9106402232109192,
      "grad_norm": 0.24568885905359372,
      "learning_rate": 5.8794656695294546e-08,
      "loss": 1.5211,
      "step": 6038
    },
    {
      "epoch": 0.9107910413995928,
      "grad_norm": 0.23744917362961723,
      "learning_rate": 5.8765164379357314e-08,
      "loss": 1.6052,
      "step": 6039
    },
    {
      "epoch": 0.9109418595882663,
      "grad_norm": 0.36932078031473947,
      "learning_rate": 5.873572061443811e-08,
      "loss": 1.5795,
      "step": 6040
    },
    {
      "epoch": 0.9110926777769399,
      "grad_norm": 0.2542230004886954,
      "learning_rate": 5.8706325407147956e-08,
      "loss": 1.6133,
      "step": 6041
    },
    {
      "epoch": 0.9112434959656135,
      "grad_norm": 0.255384496145479,
      "learning_rate": 5.867697876408694e-08,
      "loss": 1.5637,
      "step": 6042
    },
    {
      "epoch": 0.911394314154287,
      "grad_norm": 0.25485060766998097,
      "learning_rate": 5.8647680691844184e-08,
      "loss": 1.6338,
      "step": 6043
    },
    {
      "epoch": 0.9115451323429605,
      "grad_norm": 0.25237897567980183,
      "learning_rate": 5.8618431196997984e-08,
      "loss": 1.5547,
      "step": 6044
    },
    {
      "epoch": 0.9116959505316341,
      "grad_norm": 0.24409432431768954,
      "learning_rate": 5.8589230286115716e-08,
      "loss": 1.5827,
      "step": 6045
    },
    {
      "epoch": 0.9118467687203077,
      "grad_norm": 0.2570925994521639,
      "learning_rate": 5.856007796575385e-08,
      "loss": 1.5491,
      "step": 6046
    },
    {
      "epoch": 0.9119975869089813,
      "grad_norm": 0.24830464335674102,
      "learning_rate": 5.8530974242457856e-08,
      "loss": 1.4652,
      "step": 6047
    },
    {
      "epoch": 0.9121484050976548,
      "grad_norm": 0.24781908325718147,
      "learning_rate": 5.8501919122762464e-08,
      "loss": 1.6006,
      "step": 6048
    },
    {
      "epoch": 0.9122992232863283,
      "grad_norm": 0.26083921342285155,
      "learning_rate": 5.847291261319134e-08,
      "loss": 1.5819,
      "step": 6049
    },
    {
      "epoch": 0.9124500414750019,
      "grad_norm": 0.26309633869145815,
      "learning_rate": 5.8443954720257283e-08,
      "loss": 1.553,
      "step": 6050
    },
    {
      "epoch": 0.9126008596636754,
      "grad_norm": 0.26281025692831333,
      "learning_rate": 5.841504545046222e-08,
      "loss": 1.5374,
      "step": 6051
    },
    {
      "epoch": 0.912751677852349,
      "grad_norm": 0.27750393946945345,
      "learning_rate": 5.83861848102971e-08,
      "loss": 1.5446,
      "step": 6052
    },
    {
      "epoch": 0.9129024960410226,
      "grad_norm": 0.2799515575723963,
      "learning_rate": 5.835737280624197e-08,
      "loss": 1.5626,
      "step": 6053
    },
    {
      "epoch": 0.9130533142296962,
      "grad_norm": 0.24894486614996936,
      "learning_rate": 5.8328609444766006e-08,
      "loss": 1.5599,
      "step": 6054
    },
    {
      "epoch": 0.9132041324183696,
      "grad_norm": 0.26019381642413064,
      "learning_rate": 5.829989473232737e-08,
      "loss": 1.5944,
      "step": 6055
    },
    {
      "epoch": 0.9133549506070432,
      "grad_norm": 0.2759237853661695,
      "learning_rate": 5.827122867537337e-08,
      "loss": 1.5644,
      "step": 6056
    },
    {
      "epoch": 0.9135057687957168,
      "grad_norm": 0.24814657412343966,
      "learning_rate": 5.8242611280340356e-08,
      "loss": 1.5764,
      "step": 6057
    },
    {
      "epoch": 0.9136565869843903,
      "grad_norm": 0.25487268213850733,
      "learning_rate": 5.821404255365381e-08,
      "loss": 1.5767,
      "step": 6058
    },
    {
      "epoch": 0.9138074051730639,
      "grad_norm": 0.2446630422502796,
      "learning_rate": 5.818552250172822e-08,
      "loss": 1.545,
      "step": 6059
    },
    {
      "epoch": 0.9139582233617374,
      "grad_norm": 0.26215823714021624,
      "learning_rate": 5.815705113096712e-08,
      "loss": 1.6177,
      "step": 6060
    },
    {
      "epoch": 0.914109041550411,
      "grad_norm": 0.35947069849148555,
      "learning_rate": 5.812862844776324e-08,
      "loss": 1.5917,
      "step": 6061
    },
    {
      "epoch": 0.9142598597390845,
      "grad_norm": 0.308709803926604,
      "learning_rate": 5.8100254458498285e-08,
      "loss": 1.5329,
      "step": 6062
    },
    {
      "epoch": 0.9144106779277581,
      "grad_norm": 0.24201885723949065,
      "learning_rate": 5.8071929169542974e-08,
      "loss": 1.6049,
      "step": 6063
    },
    {
      "epoch": 0.9145614961164317,
      "grad_norm": 0.2542398848923103,
      "learning_rate": 5.804365258725724e-08,
      "loss": 1.5361,
      "step": 6064
    },
    {
      "epoch": 0.9147123143051052,
      "grad_norm": 0.24749220381791429,
      "learning_rate": 5.801542471798998e-08,
      "loss": 1.5734,
      "step": 6065
    },
    {
      "epoch": 0.9148631324937787,
      "grad_norm": 0.2399673199320219,
      "learning_rate": 5.7987245568079114e-08,
      "loss": 1.5079,
      "step": 6066
    },
    {
      "epoch": 0.9150139506824523,
      "grad_norm": 0.25543942985532075,
      "learning_rate": 5.795911514385176e-08,
      "loss": 1.5887,
      "step": 6067
    },
    {
      "epoch": 0.9151647688711259,
      "grad_norm": 0.25005576353674086,
      "learning_rate": 5.793103345162397e-08,
      "loss": 1.5817,
      "step": 6068
    },
    {
      "epoch": 0.9153155870597994,
      "grad_norm": 0.2469560749520833,
      "learning_rate": 5.790300049770094e-08,
      "loss": 1.5127,
      "step": 6069
    },
    {
      "epoch": 0.915466405248473,
      "grad_norm": 0.2501041361963282,
      "learning_rate": 5.7875016288376826e-08,
      "loss": 1.616,
      "step": 6070
    },
    {
      "epoch": 0.9156172234371465,
      "grad_norm": 0.24059136281338953,
      "learning_rate": 5.7847080829934966e-08,
      "loss": 1.6286,
      "step": 6071
    },
    {
      "epoch": 0.91576804162582,
      "grad_norm": 0.24840412852701096,
      "learning_rate": 5.781919412864765e-08,
      "loss": 1.5391,
      "step": 6072
    },
    {
      "epoch": 0.9159188598144936,
      "grad_norm": 0.2560422391892665,
      "learning_rate": 5.779135619077623e-08,
      "loss": 1.6219,
      "step": 6073
    },
    {
      "epoch": 0.9160696780031672,
      "grad_norm": 0.6623667693400862,
      "learning_rate": 5.7763567022571196e-08,
      "loss": 1.4886,
      "step": 6074
    },
    {
      "epoch": 0.9162204961918408,
      "grad_norm": 0.2664057578498538,
      "learning_rate": 5.7735826630271974e-08,
      "loss": 1.5858,
      "step": 6075
    },
    {
      "epoch": 0.9163713143805143,
      "grad_norm": 0.3096052738479129,
      "learning_rate": 5.770813502010708e-08,
      "loss": 1.5794,
      "step": 6076
    },
    {
      "epoch": 0.9165221325691878,
      "grad_norm": 0.31916402504883584,
      "learning_rate": 5.768049219829415e-08,
      "loss": 1.5336,
      "step": 6077
    },
    {
      "epoch": 0.9166729507578614,
      "grad_norm": 0.2563832597165492,
      "learning_rate": 5.765289817103976e-08,
      "loss": 1.554,
      "step": 6078
    },
    {
      "epoch": 0.916823768946535,
      "grad_norm": 0.2352107286443002,
      "learning_rate": 5.7625352944539537e-08,
      "loss": 1.5427,
      "step": 6079
    },
    {
      "epoch": 0.9169745871352085,
      "grad_norm": 0.24655627148991396,
      "learning_rate": 5.759785652497827e-08,
      "loss": 1.5758,
      "step": 6080
    },
    {
      "epoch": 0.9171254053238821,
      "grad_norm": 0.25237182893359833,
      "learning_rate": 5.757040891852967e-08,
      "loss": 1.6457,
      "step": 6081
    },
    {
      "epoch": 0.9172762235125557,
      "grad_norm": 0.2429829153276654,
      "learning_rate": 5.754301013135651e-08,
      "loss": 1.5361,
      "step": 6082
    },
    {
      "epoch": 0.9174270417012291,
      "grad_norm": 0.25485848289645746,
      "learning_rate": 5.75156601696106e-08,
      "loss": 1.5314,
      "step": 6083
    },
    {
      "epoch": 0.9175778598899027,
      "grad_norm": 0.23571301769763325,
      "learning_rate": 5.7488359039432835e-08,
      "loss": 1.4689,
      "step": 6084
    },
    {
      "epoch": 0.9177286780785763,
      "grad_norm": 0.24958446516321273,
      "learning_rate": 5.746110674695312e-08,
      "loss": 1.5411,
      "step": 6085
    },
    {
      "epoch": 0.9178794962672499,
      "grad_norm": 0.25595237502687773,
      "learning_rate": 5.743390329829035e-08,
      "loss": 1.5244,
      "step": 6086
    },
    {
      "epoch": 0.9180303144559234,
      "grad_norm": 0.24543998320526575,
      "learning_rate": 5.740674869955254e-08,
      "loss": 1.5927,
      "step": 6087
    },
    {
      "epoch": 0.9181811326445969,
      "grad_norm": 0.2595006942873366,
      "learning_rate": 5.737964295683665e-08,
      "loss": 1.5793,
      "step": 6088
    },
    {
      "epoch": 0.9183319508332705,
      "grad_norm": 0.634422558558343,
      "learning_rate": 5.73525860762287e-08,
      "loss": 1.5654,
      "step": 6089
    },
    {
      "epoch": 0.918482769021944,
      "grad_norm": 0.2561577816220454,
      "learning_rate": 5.73255780638038e-08,
      "loss": 1.5743,
      "step": 6090
    },
    {
      "epoch": 0.9186335872106176,
      "grad_norm": 0.256582214734705,
      "learning_rate": 5.7298618925626e-08,
      "loss": 1.5056,
      "step": 6091
    },
    {
      "epoch": 0.9187844053992912,
      "grad_norm": 0.29479209097127596,
      "learning_rate": 5.7271708667748377e-08,
      "loss": 1.4916,
      "step": 6092
    },
    {
      "epoch": 0.9189352235879648,
      "grad_norm": 0.3003603630013369,
      "learning_rate": 5.724484729621317e-08,
      "loss": 1.558,
      "step": 6093
    },
    {
      "epoch": 0.9190860417766382,
      "grad_norm": 0.26160301110829026,
      "learning_rate": 5.7218034817051436e-08,
      "loss": 1.4789,
      "step": 6094
    },
    {
      "epoch": 0.9192368599653118,
      "grad_norm": 0.24774684920350054,
      "learning_rate": 5.719127123628343e-08,
      "loss": 1.5946,
      "step": 6095
    },
    {
      "epoch": 0.9193876781539854,
      "grad_norm": 0.28108399934166406,
      "learning_rate": 5.716455655991827e-08,
      "loss": 1.5268,
      "step": 6096
    },
    {
      "epoch": 0.9195384963426589,
      "grad_norm": 0.5761951650568155,
      "learning_rate": 5.713789079395427e-08,
      "loss": 1.5801,
      "step": 6097
    },
    {
      "epoch": 0.9196893145313325,
      "grad_norm": 0.25754475229133966,
      "learning_rate": 5.711127394437863e-08,
      "loss": 1.5637,
      "step": 6098
    },
    {
      "epoch": 0.9198401327200061,
      "grad_norm": 0.25139716251584615,
      "learning_rate": 5.7084706017167595e-08,
      "loss": 1.5676,
      "step": 6099
    },
    {
      "epoch": 0.9199909509086796,
      "grad_norm": 0.28185865101427054,
      "learning_rate": 5.705818701828646e-08,
      "loss": 1.6325,
      "step": 6100
    },
    {
      "epoch": 0.9201417690973531,
      "grad_norm": 0.26628385922175346,
      "learning_rate": 5.7031716953689516e-08,
      "loss": 1.5633,
      "step": 6101
    },
    {
      "epoch": 0.9202925872860267,
      "grad_norm": 0.3687397497971009,
      "learning_rate": 5.7005295829320044e-08,
      "loss": 1.5356,
      "step": 6102
    },
    {
      "epoch": 0.9204434054747003,
      "grad_norm": 0.28602796011570303,
      "learning_rate": 5.697892365111037e-08,
      "loss": 1.5567,
      "step": 6103
    },
    {
      "epoch": 0.9205942236633738,
      "grad_norm": 0.2538349209624199,
      "learning_rate": 5.695260042498182e-08,
      "loss": 1.533,
      "step": 6104
    },
    {
      "epoch": 0.9207450418520473,
      "grad_norm": 0.26657271123144255,
      "learning_rate": 5.6926326156844695e-08,
      "loss": 1.6001,
      "step": 6105
    },
    {
      "epoch": 0.9208958600407209,
      "grad_norm": 0.24861917653118323,
      "learning_rate": 5.690010085259839e-08,
      "loss": 1.5704,
      "step": 6106
    },
    {
      "epoch": 0.9210466782293945,
      "grad_norm": 0.24530394043827866,
      "learning_rate": 5.6873924518131185e-08,
      "loss": 1.6085,
      "step": 6107
    },
    {
      "epoch": 0.921197496418068,
      "grad_norm": 0.24313647967223873,
      "learning_rate": 5.684779715932047e-08,
      "loss": 1.5621,
      "step": 6108
    },
    {
      "epoch": 0.9213483146067416,
      "grad_norm": 0.2504009150214136,
      "learning_rate": 5.6821718782032566e-08,
      "loss": 1.4987,
      "step": 6109
    },
    {
      "epoch": 0.9214991327954152,
      "grad_norm": 0.6787606527205317,
      "learning_rate": 5.679568939212289e-08,
      "loss": 1.5836,
      "step": 6110
    },
    {
      "epoch": 0.9216499509840886,
      "grad_norm": 0.2876594125286204,
      "learning_rate": 5.676970899543574e-08,
      "loss": 1.5046,
      "step": 6111
    },
    {
      "epoch": 0.9218007691727622,
      "grad_norm": 2.551721847230167,
      "learning_rate": 5.674377759780446e-08,
      "loss": 1.5481,
      "step": 6112
    },
    {
      "epoch": 0.9219515873614358,
      "grad_norm": 0.3347547508301126,
      "learning_rate": 5.671789520505145e-08,
      "loss": 1.5646,
      "step": 6113
    },
    {
      "epoch": 0.9221024055501094,
      "grad_norm": 0.2469328025595555,
      "learning_rate": 5.669206182298805e-08,
      "loss": 1.542,
      "step": 6114
    },
    {
      "epoch": 0.9222532237387829,
      "grad_norm": 0.24934031772475193,
      "learning_rate": 5.6666277457414625e-08,
      "loss": 1.5075,
      "step": 6115
    },
    {
      "epoch": 0.9224040419274564,
      "grad_norm": 0.28294027549133693,
      "learning_rate": 5.6640542114120455e-08,
      "loss": 1.5542,
      "step": 6116
    },
    {
      "epoch": 0.92255486011613,
      "grad_norm": 0.2564436610759413,
      "learning_rate": 5.661485579888392e-08,
      "loss": 1.5324,
      "step": 6117
    },
    {
      "epoch": 0.9227056783048035,
      "grad_norm": 0.2584982097210025,
      "learning_rate": 5.658921851747234e-08,
      "loss": 1.5662,
      "step": 6118
    },
    {
      "epoch": 0.9228564964934771,
      "grad_norm": 0.2670956480863373,
      "learning_rate": 5.656363027564201e-08,
      "loss": 1.596,
      "step": 6119
    },
    {
      "epoch": 0.9230073146821507,
      "grad_norm": 0.30653116963810917,
      "learning_rate": 5.6538091079138294e-08,
      "loss": 1.5695,
      "step": 6120
    },
    {
      "epoch": 0.9231581328708243,
      "grad_norm": 0.30199507780830126,
      "learning_rate": 5.6512600933695425e-08,
      "loss": 1.5549,
      "step": 6121
    },
    {
      "epoch": 0.9233089510594977,
      "grad_norm": 0.33741167082112555,
      "learning_rate": 5.6487159845036685e-08,
      "loss": 1.5277,
      "step": 6122
    },
    {
      "epoch": 0.9234597692481713,
      "grad_norm": 0.275068483440525,
      "learning_rate": 5.6461767818874364e-08,
      "loss": 1.5245,
      "step": 6123
    },
    {
      "epoch": 0.9236105874368449,
      "grad_norm": 0.24574270325486824,
      "learning_rate": 5.643642486090973e-08,
      "loss": 1.5421,
      "step": 6124
    },
    {
      "epoch": 0.9237614056255185,
      "grad_norm": 0.2477649885538568,
      "learning_rate": 5.6411130976832994e-08,
      "loss": 1.5655,
      "step": 6125
    },
    {
      "epoch": 0.923912223814192,
      "grad_norm": 0.24572262725711227,
      "learning_rate": 5.638588617232331e-08,
      "loss": 1.5446,
      "step": 6126
    },
    {
      "epoch": 0.9240630420028656,
      "grad_norm": 0.24912195102215598,
      "learning_rate": 5.6360690453048983e-08,
      "loss": 1.5682,
      "step": 6127
    },
    {
      "epoch": 0.9242138601915391,
      "grad_norm": 0.25792672085752477,
      "learning_rate": 5.6335543824667114e-08,
      "loss": 1.491,
      "step": 6128
    },
    {
      "epoch": 0.9243646783802126,
      "grad_norm": 0.3338943665700372,
      "learning_rate": 5.6310446292823845e-08,
      "loss": 1.5448,
      "step": 6129
    },
    {
      "epoch": 0.9245154965688862,
      "grad_norm": 0.30488090147190133,
      "learning_rate": 5.6285397863154355e-08,
      "loss": 1.5588,
      "step": 6130
    },
    {
      "epoch": 0.9246663147575598,
      "grad_norm": 0.24595625574750304,
      "learning_rate": 5.6260398541282723e-08,
      "loss": 1.5927,
      "step": 6131
    },
    {
      "epoch": 0.9248171329462334,
      "grad_norm": 0.23974432099050957,
      "learning_rate": 5.623544833282197e-08,
      "loss": 1.5094,
      "step": 6132
    },
    {
      "epoch": 0.9249679511349068,
      "grad_norm": 0.2549180339482979,
      "learning_rate": 5.6210547243374235e-08,
      "loss": 1.6182,
      "step": 6133
    },
    {
      "epoch": 0.9251187693235804,
      "grad_norm": 0.2537976703144186,
      "learning_rate": 5.61856952785305e-08,
      "loss": 1.6009,
      "step": 6134
    },
    {
      "epoch": 0.925269587512254,
      "grad_norm": 0.24465920281166262,
      "learning_rate": 5.6160892443870716e-08,
      "loss": 1.5955,
      "step": 6135
    },
    {
      "epoch": 0.9254204057009275,
      "grad_norm": 0.25195229045779016,
      "learning_rate": 5.613613874496392e-08,
      "loss": 1.5537,
      "step": 6136
    },
    {
      "epoch": 0.9255712238896011,
      "grad_norm": 0.24869817633114924,
      "learning_rate": 5.6111434187368e-08,
      "loss": 1.5412,
      "step": 6137
    },
    {
      "epoch": 0.9257220420782747,
      "grad_norm": 0.271284202605858,
      "learning_rate": 5.608677877662981e-08,
      "loss": 1.545,
      "step": 6138
    },
    {
      "epoch": 0.9258728602669481,
      "grad_norm": 0.27163866870959363,
      "learning_rate": 5.606217251828525e-08,
      "loss": 1.5452,
      "step": 6139
    },
    {
      "epoch": 0.9260236784556217,
      "grad_norm": 0.46977552446701015,
      "learning_rate": 5.6037615417859155e-08,
      "loss": 1.5388,
      "step": 6140
    },
    {
      "epoch": 0.9261744966442953,
      "grad_norm": 0.36444870166684856,
      "learning_rate": 5.601310748086528e-08,
      "loss": 1.5172,
      "step": 6141
    },
    {
      "epoch": 0.9263253148329689,
      "grad_norm": 0.24999061175916848,
      "learning_rate": 5.5988648712806355e-08,
      "loss": 1.5342,
      "step": 6142
    },
    {
      "epoch": 0.9264761330216424,
      "grad_norm": 0.24147710000217487,
      "learning_rate": 5.596423911917414e-08,
      "loss": 1.5201,
      "step": 6143
    },
    {
      "epoch": 0.926626951210316,
      "grad_norm": 0.26529601088070126,
      "learning_rate": 5.5939878705449263e-08,
      "loss": 1.5234,
      "step": 6144
    },
    {
      "epoch": 0.9267777693989895,
      "grad_norm": 0.2450064676081888,
      "learning_rate": 5.591556747710131e-08,
      "loss": 1.5023,
      "step": 6145
    },
    {
      "epoch": 0.926928587587663,
      "grad_norm": 0.5562495178275069,
      "learning_rate": 5.589130543958897e-08,
      "loss": 1.5791,
      "step": 6146
    },
    {
      "epoch": 0.9270794057763366,
      "grad_norm": 0.25271078865290386,
      "learning_rate": 5.5867092598359674e-08,
      "loss": 1.5522,
      "step": 6147
    },
    {
      "epoch": 0.9272302239650102,
      "grad_norm": 0.2510375488276909,
      "learning_rate": 5.584292895884995e-08,
      "loss": 1.5708,
      "step": 6148
    },
    {
      "epoch": 0.9273810421536838,
      "grad_norm": 0.30069052420596476,
      "learning_rate": 5.581881452648523e-08,
      "loss": 1.5017,
      "step": 6149
    },
    {
      "epoch": 0.9275318603423572,
      "grad_norm": 0.25405929708733727,
      "learning_rate": 5.579474930667992e-08,
      "loss": 1.5602,
      "step": 6150
    },
    {
      "epoch": 0.9276826785310308,
      "grad_norm": 0.24886204828205324,
      "learning_rate": 5.577073330483736e-08,
      "loss": 1.556,
      "step": 6151
    },
    {
      "epoch": 0.9278334967197044,
      "grad_norm": 0.24788735210464843,
      "learning_rate": 5.574676652634979e-08,
      "loss": 1.5254,
      "step": 6152
    },
    {
      "epoch": 0.927984314908378,
      "grad_norm": 0.2626661539030793,
      "learning_rate": 5.572284897659854e-08,
      "loss": 1.5193,
      "step": 6153
    },
    {
      "epoch": 0.9281351330970515,
      "grad_norm": 0.26222135209071656,
      "learning_rate": 5.5698980660953747e-08,
      "loss": 1.5583,
      "step": 6154
    },
    {
      "epoch": 0.9282859512857251,
      "grad_norm": 0.316263266609963,
      "learning_rate": 5.5675161584774516e-08,
      "loss": 1.5786,
      "step": 6155
    },
    {
      "epoch": 0.9284367694743986,
      "grad_norm": 0.2543644512752816,
      "learning_rate": 5.565139175340899e-08,
      "loss": 1.6077,
      "step": 6156
    },
    {
      "epoch": 0.9285875876630721,
      "grad_norm": 0.26142964558477094,
      "learning_rate": 5.562767117219414e-08,
      "loss": 1.4931,
      "step": 6157
    },
    {
      "epoch": 0.9287384058517457,
      "grad_norm": 0.25780942757545783,
      "learning_rate": 5.5603999846455915e-08,
      "loss": 1.4687,
      "step": 6158
    },
    {
      "epoch": 0.9288892240404193,
      "grad_norm": 0.24910206966222853,
      "learning_rate": 5.558037778150929e-08,
      "loss": 1.5521,
      "step": 6159
    },
    {
      "epoch": 0.9290400422290929,
      "grad_norm": 0.2610805885124634,
      "learning_rate": 5.555680498265804e-08,
      "loss": 1.586,
      "step": 6160
    },
    {
      "epoch": 0.9291908604177663,
      "grad_norm": 0.24786750742412972,
      "learning_rate": 5.5533281455194955e-08,
      "loss": 1.6002,
      "step": 6161
    },
    {
      "epoch": 0.9293416786064399,
      "grad_norm": 0.27917077529733814,
      "learning_rate": 5.550980720440176e-08,
      "loss": 1.5492,
      "step": 6162
    },
    {
      "epoch": 0.9294924967951135,
      "grad_norm": 0.6410057671884332,
      "learning_rate": 5.5486382235549103e-08,
      "loss": 1.5816,
      "step": 6163
    },
    {
      "epoch": 0.929643314983787,
      "grad_norm": 0.2511566309452938,
      "learning_rate": 5.5463006553896596e-08,
      "loss": 1.5545,
      "step": 6164
    },
    {
      "epoch": 0.9297941331724606,
      "grad_norm": 0.2737661737146725,
      "learning_rate": 5.543968016469273e-08,
      "loss": 1.5745,
      "step": 6165
    },
    {
      "epoch": 0.9299449513611342,
      "grad_norm": 0.3797270236127085,
      "learning_rate": 5.5416403073174975e-08,
      "loss": 1.6144,
      "step": 6166
    },
    {
      "epoch": 0.9300957695498077,
      "grad_norm": 0.31918928141620784,
      "learning_rate": 5.5393175284569716e-08,
      "loss": 1.6324,
      "step": 6167
    },
    {
      "epoch": 0.9302465877384812,
      "grad_norm": 0.24958868757458863,
      "learning_rate": 5.5369996804092246e-08,
      "loss": 1.5633,
      "step": 6168
    },
    {
      "epoch": 0.9303974059271548,
      "grad_norm": 0.25361093030951837,
      "learning_rate": 5.534686763694684e-08,
      "loss": 1.552,
      "step": 6169
    },
    {
      "epoch": 0.9305482241158284,
      "grad_norm": 0.26060521044377927,
      "learning_rate": 5.532378778832665e-08,
      "loss": 1.5283,
      "step": 6170
    },
    {
      "epoch": 0.930699042304502,
      "grad_norm": 0.441136825411557,
      "learning_rate": 5.530075726341377e-08,
      "loss": 1.5484,
      "step": 6171
    },
    {
      "epoch": 0.9308498604931755,
      "grad_norm": 0.2464745890853672,
      "learning_rate": 5.527777606737926e-08,
      "loss": 1.5102,
      "step": 6172
    },
    {
      "epoch": 0.931000678681849,
      "grad_norm": 0.24944079204088596,
      "learning_rate": 5.525484420538303e-08,
      "loss": 1.523,
      "step": 6173
    },
    {
      "epoch": 0.9311514968705226,
      "grad_norm": 0.2875126059730783,
      "learning_rate": 5.523196168257397e-08,
      "loss": 1.6301,
      "step": 6174
    },
    {
      "epoch": 0.9313023150591961,
      "grad_norm": 0.26474669232214115,
      "learning_rate": 5.520912850408985e-08,
      "loss": 1.589,
      "step": 6175
    },
    {
      "epoch": 0.9314531332478697,
      "grad_norm": 0.26217816594758014,
      "learning_rate": 5.518634467505739e-08,
      "loss": 1.553,
      "step": 6176
    },
    {
      "epoch": 0.9316039514365433,
      "grad_norm": 0.2718608358446988,
      "learning_rate": 5.5163610200592256e-08,
      "loss": 1.5838,
      "step": 6177
    },
    {
      "epoch": 0.9317547696252167,
      "grad_norm": 0.24415865400433565,
      "learning_rate": 5.514092508579894e-08,
      "loss": 1.5133,
      "step": 6178
    },
    {
      "epoch": 0.9319055878138903,
      "grad_norm": 0.2411641486293077,
      "learning_rate": 5.5118289335771e-08,
      "loss": 1.5839,
      "step": 6179
    },
    {
      "epoch": 0.9320564060025639,
      "grad_norm": 0.2514668725290416,
      "learning_rate": 5.509570295559074e-08,
      "loss": 1.5972,
      "step": 6180
    },
    {
      "epoch": 0.9322072241912375,
      "grad_norm": 0.2759880586716973,
      "learning_rate": 5.5073165950329465e-08,
      "loss": 1.5304,
      "step": 6181
    },
    {
      "epoch": 0.932358042379911,
      "grad_norm": 0.25605097901499024,
      "learning_rate": 5.505067832504744e-08,
      "loss": 1.5246,
      "step": 6182
    },
    {
      "epoch": 0.9325088605685846,
      "grad_norm": 0.29444578427717816,
      "learning_rate": 5.502824008479376e-08,
      "loss": 1.5881,
      "step": 6183
    },
    {
      "epoch": 0.9326596787572581,
      "grad_norm": 0.24273982291992666,
      "learning_rate": 5.5005851234606446e-08,
      "loss": 1.5598,
      "step": 6184
    },
    {
      "epoch": 0.9328104969459317,
      "grad_norm": 0.25299211156493445,
      "learning_rate": 5.4983511779512494e-08,
      "loss": 1.595,
      "step": 6185
    },
    {
      "epoch": 0.9329613151346052,
      "grad_norm": 0.25732622340646677,
      "learning_rate": 5.4961221724527716e-08,
      "loss": 1.6029,
      "step": 6186
    },
    {
      "epoch": 0.9331121333232788,
      "grad_norm": 0.8128019546387675,
      "learning_rate": 5.493898107465691e-08,
      "loss": 1.5349,
      "step": 6187
    },
    {
      "epoch": 0.9332629515119524,
      "grad_norm": 0.27992600807371115,
      "learning_rate": 5.4916789834893725e-08,
      "loss": 1.6154,
      "step": 6188
    },
    {
      "epoch": 0.933413769700626,
      "grad_norm": 0.2542356440560896,
      "learning_rate": 5.489464801022077e-08,
      "loss": 1.5421,
      "step": 6189
    },
    {
      "epoch": 0.9335645878892994,
      "grad_norm": 0.26740460030730834,
      "learning_rate": 5.487255560560952e-08,
      "loss": 1.6063,
      "step": 6190
    },
    {
      "epoch": 0.933715406077973,
      "grad_norm": 0.272674547101346,
      "learning_rate": 5.4850512626020335e-08,
      "loss": 1.6055,
      "step": 6191
    },
    {
      "epoch": 0.9338662242666466,
      "grad_norm": 0.24536914204213503,
      "learning_rate": 5.482851907640254e-08,
      "loss": 1.572,
      "step": 6192
    },
    {
      "epoch": 0.9340170424553201,
      "grad_norm": 0.2449252209481925,
      "learning_rate": 5.4806574961694314e-08,
      "loss": 1.5289,
      "step": 6193
    },
    {
      "epoch": 0.9341678606439937,
      "grad_norm": 0.24055151366290678,
      "learning_rate": 5.478468028682272e-08,
      "loss": 1.5498,
      "step": 6194
    },
    {
      "epoch": 0.9343186788326672,
      "grad_norm": 0.2527859810550758,
      "learning_rate": 5.476283505670381e-08,
      "loss": 1.5134,
      "step": 6195
    },
    {
      "epoch": 0.9344694970213407,
      "grad_norm": 0.3577246499396631,
      "learning_rate": 5.474103927624247e-08,
      "loss": 1.5736,
      "step": 6196
    },
    {
      "epoch": 0.9346203152100143,
      "grad_norm": 0.2766852958914957,
      "learning_rate": 5.471929295033241e-08,
      "loss": 1.6239,
      "step": 6197
    },
    {
      "epoch": 0.9347711333986879,
      "grad_norm": 0.24795491996332725,
      "learning_rate": 5.4697596083856396e-08,
      "loss": 1.5726,
      "step": 6198
    },
    {
      "epoch": 0.9349219515873615,
      "grad_norm": 0.26782304778646604,
      "learning_rate": 5.4675948681685984e-08,
      "loss": 1.5763,
      "step": 6199
    },
    {
      "epoch": 0.935072769776035,
      "grad_norm": 0.26596988390834714,
      "learning_rate": 5.465435074868161e-08,
      "loss": 1.5687,
      "step": 6200
    },
    {
      "epoch": 0.9352235879647085,
      "grad_norm": 0.2580184881996988,
      "learning_rate": 5.463280228969268e-08,
      "loss": 1.5726,
      "step": 6201
    },
    {
      "epoch": 0.9353744061533821,
      "grad_norm": 0.24960174972988566,
      "learning_rate": 5.4611303309557454e-08,
      "loss": 1.5639,
      "step": 6202
    },
    {
      "epoch": 0.9355252243420557,
      "grad_norm": 0.24331613996923387,
      "learning_rate": 5.4589853813103055e-08,
      "loss": 1.5338,
      "step": 6203
    },
    {
      "epoch": 0.9356760425307292,
      "grad_norm": 0.2879551135187979,
      "learning_rate": 5.456845380514551e-08,
      "loss": 1.5685,
      "step": 6204
    },
    {
      "epoch": 0.9358268607194028,
      "grad_norm": 0.25161922629399475,
      "learning_rate": 5.454710329048977e-08,
      "loss": 1.4967,
      "step": 6205
    },
    {
      "epoch": 0.9359776789080763,
      "grad_norm": 0.26340960169106836,
      "learning_rate": 5.452580227392966e-08,
      "loss": 1.5289,
      "step": 6206
    },
    {
      "epoch": 0.9361284970967498,
      "grad_norm": 0.2670862257291208,
      "learning_rate": 5.450455076024781e-08,
      "loss": 1.6131,
      "step": 6207
    },
    {
      "epoch": 0.9362793152854234,
      "grad_norm": 0.24782221497804563,
      "learning_rate": 5.4483348754215895e-08,
      "loss": 1.5653,
      "step": 6208
    },
    {
      "epoch": 0.936430133474097,
      "grad_norm": 0.2808602684717364,
      "learning_rate": 5.446219626059433e-08,
      "loss": 1.5309,
      "step": 6209
    },
    {
      "epoch": 0.9365809516627706,
      "grad_norm": 0.24441080991582628,
      "learning_rate": 5.444109328413246e-08,
      "loss": 1.5623,
      "step": 6210
    },
    {
      "epoch": 0.9367317698514441,
      "grad_norm": 0.29055086813288683,
      "learning_rate": 5.442003982956855e-08,
      "loss": 1.5119,
      "step": 6211
    },
    {
      "epoch": 0.9368825880401176,
      "grad_norm": 0.2522579754654987,
      "learning_rate": 5.439903590162971e-08,
      "loss": 1.5241,
      "step": 6212
    },
    {
      "epoch": 0.9370334062287912,
      "grad_norm": 0.25334183102849595,
      "learning_rate": 5.43780815050319e-08,
      "loss": 1.5799,
      "step": 6213
    },
    {
      "epoch": 0.9371842244174647,
      "grad_norm": 2.436419004177727,
      "learning_rate": 5.435717664448001e-08,
      "loss": 1.5813,
      "step": 6214
    },
    {
      "epoch": 0.9373350426061383,
      "grad_norm": 0.2828523731284309,
      "learning_rate": 5.4336321324667806e-08,
      "loss": 1.5146,
      "step": 6215
    },
    {
      "epoch": 0.9374858607948119,
      "grad_norm": 0.27876418593965085,
      "learning_rate": 5.431551555027789e-08,
      "loss": 1.6188,
      "step": 6216
    },
    {
      "epoch": 0.9376366789834855,
      "grad_norm": 0.258647266375581,
      "learning_rate": 5.429475932598177e-08,
      "loss": 1.5489,
      "step": 6217
    },
    {
      "epoch": 0.9377874971721589,
      "grad_norm": 0.4395462622142405,
      "learning_rate": 5.427405265643983e-08,
      "loss": 1.5212,
      "step": 6218
    },
    {
      "epoch": 0.9379383153608325,
      "grad_norm": 0.25527207182841355,
      "learning_rate": 5.4253395546301344e-08,
      "loss": 1.5375,
      "step": 6219
    },
    {
      "epoch": 0.9380891335495061,
      "grad_norm": 0.25692760225637207,
      "learning_rate": 5.4232788000204385e-08,
      "loss": 1.5743,
      "step": 6220
    },
    {
      "epoch": 0.9382399517381796,
      "grad_norm": 0.2527388733388531,
      "learning_rate": 5.4212230022775965e-08,
      "loss": 1.5302,
      "step": 6221
    },
    {
      "epoch": 0.9383907699268532,
      "grad_norm": 0.24386827293627816,
      "learning_rate": 5.4191721618631976e-08,
      "loss": 1.603,
      "step": 6222
    },
    {
      "epoch": 0.9385415881155267,
      "grad_norm": 0.2502629611784765,
      "learning_rate": 5.4171262792377134e-08,
      "loss": 1.5227,
      "step": 6223
    },
    {
      "epoch": 0.9386924063042003,
      "grad_norm": 0.2495200719724515,
      "learning_rate": 5.415085354860503e-08,
      "loss": 1.5439,
      "step": 6224
    },
    {
      "epoch": 0.9388432244928738,
      "grad_norm": 0.24395521270584844,
      "learning_rate": 5.413049389189813e-08,
      "loss": 1.5743,
      "step": 6225
    },
    {
      "epoch": 0.9389940426815474,
      "grad_norm": 0.2528655609824242,
      "learning_rate": 5.4110183826827805e-08,
      "loss": 1.5886,
      "step": 6226
    },
    {
      "epoch": 0.939144860870221,
      "grad_norm": 0.2535177596142676,
      "learning_rate": 5.4089923357954206e-08,
      "loss": 1.5847,
      "step": 6227
    },
    {
      "epoch": 0.9392956790588946,
      "grad_norm": 0.30616125366794783,
      "learning_rate": 5.406971248982644e-08,
      "loss": 1.5544,
      "step": 6228
    },
    {
      "epoch": 0.939446497247568,
      "grad_norm": 0.24170407278087794,
      "learning_rate": 5.404955122698243e-08,
      "loss": 1.5934,
      "step": 6229
    },
    {
      "epoch": 0.9395973154362416,
      "grad_norm": 0.25291719608092217,
      "learning_rate": 5.4029439573948954e-08,
      "loss": 1.5598,
      "step": 6230
    },
    {
      "epoch": 0.9397481336249152,
      "grad_norm": 0.2725490134475271,
      "learning_rate": 5.400937753524165e-08,
      "loss": 1.5739,
      "step": 6231
    },
    {
      "epoch": 0.9398989518135887,
      "grad_norm": 0.30230082036390676,
      "learning_rate": 5.3989365115365075e-08,
      "loss": 1.5173,
      "step": 6232
    },
    {
      "epoch": 0.9400497700022623,
      "grad_norm": 0.2721094377001865,
      "learning_rate": 5.396940231881256e-08,
      "loss": 1.5834,
      "step": 6233
    },
    {
      "epoch": 0.9402005881909359,
      "grad_norm": 0.30026085064325764,
      "learning_rate": 5.394948915006634e-08,
      "loss": 1.6117,
      "step": 6234
    },
    {
      "epoch": 0.9403514063796093,
      "grad_norm": 0.2593433685604889,
      "learning_rate": 5.392962561359751e-08,
      "loss": 1.5859,
      "step": 6235
    },
    {
      "epoch": 0.9405022245682829,
      "grad_norm": 0.27293940297424424,
      "learning_rate": 5.390981171386603e-08,
      "loss": 1.5537,
      "step": 6236
    },
    {
      "epoch": 0.9406530427569565,
      "grad_norm": 0.6238862258928658,
      "learning_rate": 5.389004745532065e-08,
      "loss": 1.5627,
      "step": 6237
    },
    {
      "epoch": 0.9408038609456301,
      "grad_norm": 0.29140578643943493,
      "learning_rate": 5.387033284239907e-08,
      "loss": 1.6062,
      "step": 6238
    },
    {
      "epoch": 0.9409546791343036,
      "grad_norm": 0.2581601694967583,
      "learning_rate": 5.385066787952777e-08,
      "loss": 1.5281,
      "step": 6239
    },
    {
      "epoch": 0.9411054973229771,
      "grad_norm": 0.27090139343038255,
      "learning_rate": 5.3831052571122093e-08,
      "loss": 1.5505,
      "step": 6240
    },
    {
      "epoch": 0.9412563155116507,
      "grad_norm": 0.25486575308346454,
      "learning_rate": 5.3811486921586287e-08,
      "loss": 1.5933,
      "step": 6241
    },
    {
      "epoch": 0.9414071337003243,
      "grad_norm": 0.3990119272110088,
      "learning_rate": 5.37919709353134e-08,
      "loss": 1.5544,
      "step": 6242
    },
    {
      "epoch": 0.9415579518889978,
      "grad_norm": 0.31640864216127074,
      "learning_rate": 5.37725046166853e-08,
      "loss": 1.5405,
      "step": 6243
    },
    {
      "epoch": 0.9417087700776714,
      "grad_norm": 0.36875348906357763,
      "learning_rate": 5.3753087970072784e-08,
      "loss": 1.6093,
      "step": 6244
    },
    {
      "epoch": 0.941859588266345,
      "grad_norm": 0.2535016803140429,
      "learning_rate": 5.3733720999835434e-08,
      "loss": 1.5158,
      "step": 6245
    },
    {
      "epoch": 0.9420104064550184,
      "grad_norm": 0.2656723034956448,
      "learning_rate": 5.37144037103217e-08,
      "loss": 1.5191,
      "step": 6246
    },
    {
      "epoch": 0.942161224643692,
      "grad_norm": 0.25601902569295426,
      "learning_rate": 5.36951361058689e-08,
      "loss": 1.6099,
      "step": 6247
    },
    {
      "epoch": 0.9423120428323656,
      "grad_norm": 0.24039116671233865,
      "learning_rate": 5.367591819080316e-08,
      "loss": 1.5488,
      "step": 6248
    },
    {
      "epoch": 0.9424628610210392,
      "grad_norm": 0.2705119376049466,
      "learning_rate": 5.3656749969439443e-08,
      "loss": 1.5515,
      "step": 6249
    },
    {
      "epoch": 0.9426136792097127,
      "grad_norm": 0.2826894483285141,
      "learning_rate": 5.363763144608157e-08,
      "loss": 1.5691,
      "step": 6250
    },
    {
      "epoch": 0.9427644973983862,
      "grad_norm": 0.25073840366910916,
      "learning_rate": 5.361856262502224e-08,
      "loss": 1.5692,
      "step": 6251
    },
    {
      "epoch": 0.9429153155870598,
      "grad_norm": 0.36303998778681995,
      "learning_rate": 5.359954351054295e-08,
      "loss": 1.5497,
      "step": 6252
    },
    {
      "epoch": 0.9430661337757333,
      "grad_norm": 0.276823763660622,
      "learning_rate": 5.3580574106914006e-08,
      "loss": 1.5727,
      "step": 6253
    },
    {
      "epoch": 0.9432169519644069,
      "grad_norm": 0.3197162806628283,
      "learning_rate": 5.356165441839465e-08,
      "loss": 1.6099,
      "step": 6254
    },
    {
      "epoch": 0.9433677701530805,
      "grad_norm": 0.4140003917037914,
      "learning_rate": 5.3542784449232866e-08,
      "loss": 1.5471,
      "step": 6255
    },
    {
      "epoch": 0.9435185883417541,
      "grad_norm": 0.23637027304146924,
      "learning_rate": 5.3523964203665504e-08,
      "loss": 1.5509,
      "step": 6256
    },
    {
      "epoch": 0.9436694065304275,
      "grad_norm": 0.25411725979810995,
      "learning_rate": 5.350519368591828e-08,
      "loss": 1.5274,
      "step": 6257
    },
    {
      "epoch": 0.9438202247191011,
      "grad_norm": 0.2571056548657803,
      "learning_rate": 5.348647290020575e-08,
      "loss": 1.5255,
      "step": 6258
    },
    {
      "epoch": 0.9439710429077747,
      "grad_norm": 0.891663348681135,
      "learning_rate": 5.346780185073119e-08,
      "loss": 1.5706,
      "step": 6259
    },
    {
      "epoch": 0.9441218610964482,
      "grad_norm": 0.24728698455839027,
      "learning_rate": 5.344918054168689e-08,
      "loss": 1.6282,
      "step": 6260
    },
    {
      "epoch": 0.9442726792851218,
      "grad_norm": 0.26295350236768716,
      "learning_rate": 5.343060897725381e-08,
      "loss": 1.544,
      "step": 6261
    },
    {
      "epoch": 0.9444234974737954,
      "grad_norm": 0.4401728429434684,
      "learning_rate": 5.341208716160183e-08,
      "loss": 1.4944,
      "step": 6262
    },
    {
      "epoch": 0.9445743156624689,
      "grad_norm": 0.3103928113123362,
      "learning_rate": 5.339361509888962e-08,
      "loss": 1.5241,
      "step": 6263
    },
    {
      "epoch": 0.9447251338511424,
      "grad_norm": 0.2529679821462462,
      "learning_rate": 5.337519279326473e-08,
      "loss": 1.4817,
      "step": 6264
    },
    {
      "epoch": 0.944875952039816,
      "grad_norm": 0.2446506531939954,
      "learning_rate": 5.335682024886348e-08,
      "loss": 1.5688,
      "step": 6265
    },
    {
      "epoch": 0.9450267702284896,
      "grad_norm": 0.26458469173429827,
      "learning_rate": 5.3338497469811033e-08,
      "loss": 1.5002,
      "step": 6266
    },
    {
      "epoch": 0.9451775884171632,
      "grad_norm": 0.25643614837676953,
      "learning_rate": 5.3320224460221424e-08,
      "loss": 1.5558,
      "step": 6267
    },
    {
      "epoch": 0.9453284066058366,
      "grad_norm": 0.2539040408097631,
      "learning_rate": 5.3302001224197445e-08,
      "loss": 1.6157,
      "step": 6268
    },
    {
      "epoch": 0.9454792247945102,
      "grad_norm": 0.26683606113407243,
      "learning_rate": 5.3283827765830754e-08,
      "loss": 1.6008,
      "step": 6269
    },
    {
      "epoch": 0.9456300429831838,
      "grad_norm": 0.26095663501899063,
      "learning_rate": 5.32657040892018e-08,
      "loss": 1.581,
      "step": 6270
    },
    {
      "epoch": 0.9457808611718573,
      "grad_norm": 0.2856783576919806,
      "learning_rate": 5.3247630198379916e-08,
      "loss": 1.5888,
      "step": 6271
    },
    {
      "epoch": 0.9459316793605309,
      "grad_norm": 0.24943877739150502,
      "learning_rate": 5.322960609742317e-08,
      "loss": 1.4971,
      "step": 6272
    },
    {
      "epoch": 0.9460824975492045,
      "grad_norm": 0.25485051413068144,
      "learning_rate": 5.321163179037857e-08,
      "loss": 1.5837,
      "step": 6273
    },
    {
      "epoch": 0.946233315737878,
      "grad_norm": 0.2666569322243275,
      "learning_rate": 5.319370728128178e-08,
      "loss": 1.5049,
      "step": 6274
    },
    {
      "epoch": 0.9463841339265515,
      "grad_norm": 0.2442776208036276,
      "learning_rate": 5.317583257415743e-08,
      "loss": 1.5863,
      "step": 6275
    },
    {
      "epoch": 0.9465349521152251,
      "grad_norm": 0.254146941158463,
      "learning_rate": 5.31580076730189e-08,
      "loss": 1.5502,
      "step": 6276
    },
    {
      "epoch": 0.9466857703038987,
      "grad_norm": 0.2517459606134991,
      "learning_rate": 5.314023258186843e-08,
      "loss": 1.575,
      "step": 6277
    },
    {
      "epoch": 0.9468365884925722,
      "grad_norm": 0.2983302017671054,
      "learning_rate": 5.312250730469697e-08,
      "loss": 1.539,
      "step": 6278
    },
    {
      "epoch": 0.9469874066812458,
      "grad_norm": 0.26748833623162827,
      "learning_rate": 5.310483184548442e-08,
      "loss": 1.5791,
      "step": 6279
    },
    {
      "epoch": 0.9471382248699193,
      "grad_norm": 0.2768742628332462,
      "learning_rate": 5.308720620819943e-08,
      "loss": 1.5507,
      "step": 6280
    },
    {
      "epoch": 0.9472890430585929,
      "grad_norm": 0.38464340768160965,
      "learning_rate": 5.3069630396799485e-08,
      "loss": 1.5766,
      "step": 6281
    },
    {
      "epoch": 0.9474398612472664,
      "grad_norm": 0.23792739536809543,
      "learning_rate": 5.3052104415230795e-08,
      "loss": 1.5254,
      "step": 6282
    },
    {
      "epoch": 0.94759067943594,
      "grad_norm": 0.26708907903881945,
      "learning_rate": 5.303462826742852e-08,
      "loss": 1.6363,
      "step": 6283
    },
    {
      "epoch": 0.9477414976246136,
      "grad_norm": 0.25006308858650866,
      "learning_rate": 5.3017201957316576e-08,
      "loss": 1.5392,
      "step": 6284
    },
    {
      "epoch": 0.947892315813287,
      "grad_norm": 0.2500468930788792,
      "learning_rate": 5.29998254888076e-08,
      "loss": 1.5147,
      "step": 6285
    },
    {
      "epoch": 0.9480431340019606,
      "grad_norm": 0.2394152106733528,
      "learning_rate": 5.2982498865803195e-08,
      "loss": 1.5139,
      "step": 6286
    },
    {
      "epoch": 0.9481939521906342,
      "grad_norm": 0.32357770060991164,
      "learning_rate": 5.296522209219362e-08,
      "loss": 1.5668,
      "step": 6287
    },
    {
      "epoch": 0.9483447703793078,
      "grad_norm": 0.33410105145228264,
      "learning_rate": 5.2947995171858095e-08,
      "loss": 1.5436,
      "step": 6288
    },
    {
      "epoch": 0.9484955885679813,
      "grad_norm": 0.2512225546358416,
      "learning_rate": 5.2930818108664494e-08,
      "loss": 1.533,
      "step": 6289
    },
    {
      "epoch": 0.9486464067566549,
      "grad_norm": 0.28751801968688,
      "learning_rate": 5.2913690906469613e-08,
      "loss": 1.5494,
      "step": 6290
    },
    {
      "epoch": 0.9487972249453284,
      "grad_norm": 0.2515027708999194,
      "learning_rate": 5.2896613569119e-08,
      "loss": 1.4793,
      "step": 6291
    },
    {
      "epoch": 0.9489480431340019,
      "grad_norm": 0.26909642061522543,
      "learning_rate": 5.287958610044697e-08,
      "loss": 1.5931,
      "step": 6292
    },
    {
      "epoch": 0.9490988613226755,
      "grad_norm": 0.2835281984574101,
      "learning_rate": 5.286260850427675e-08,
      "loss": 1.5816,
      "step": 6293
    },
    {
      "epoch": 0.9492496795113491,
      "grad_norm": 0.251470368186942,
      "learning_rate": 5.284568078442029e-08,
      "loss": 1.6253,
      "step": 6294
    },
    {
      "epoch": 0.9494004977000227,
      "grad_norm": 0.2515840504511767,
      "learning_rate": 5.2828802944678316e-08,
      "loss": 1.5558,
      "step": 6295
    },
    {
      "epoch": 0.9495513158886961,
      "grad_norm": 0.24579037686156516,
      "learning_rate": 5.281197498884042e-08,
      "loss": 1.5177,
      "step": 6296
    },
    {
      "epoch": 0.9497021340773697,
      "grad_norm": 0.48143361296846493,
      "learning_rate": 5.2795196920684983e-08,
      "loss": 1.5907,
      "step": 6297
    },
    {
      "epoch": 0.9498529522660433,
      "grad_norm": 0.36402142735239346,
      "learning_rate": 5.277846874397915e-08,
      "loss": 1.5661,
      "step": 6298
    },
    {
      "epoch": 0.9500037704547168,
      "grad_norm": 0.8513052715793634,
      "learning_rate": 5.27617904624789e-08,
      "loss": 1.5928,
      "step": 6299
    },
    {
      "epoch": 0.9501545886433904,
      "grad_norm": 0.2492730562768886,
      "learning_rate": 5.274516207992898e-08,
      "loss": 1.5783,
      "step": 6300
    },
    {
      "epoch": 0.950305406832064,
      "grad_norm": 0.2483441833704336,
      "learning_rate": 5.272858360006294e-08,
      "loss": 1.5933,
      "step": 6301
    },
    {
      "epoch": 0.9504562250207375,
      "grad_norm": 0.4504628252228296,
      "learning_rate": 5.271205502660316e-08,
      "loss": 1.5785,
      "step": 6302
    },
    {
      "epoch": 0.950607043209411,
      "grad_norm": 0.2497635116932171,
      "learning_rate": 5.269557636326074e-08,
      "loss": 1.5573,
      "step": 6303
    },
    {
      "epoch": 0.9507578613980846,
      "grad_norm": 0.6568417764575003,
      "learning_rate": 5.267914761373567e-08,
      "loss": 1.5554,
      "step": 6304
    },
    {
      "epoch": 0.9509086795867582,
      "grad_norm": 0.2397379571816978,
      "learning_rate": 5.266276878171664e-08,
      "loss": 1.5802,
      "step": 6305
    },
    {
      "epoch": 0.9510594977754318,
      "grad_norm": 0.26934854200352987,
      "learning_rate": 5.264643987088122e-08,
      "loss": 1.4862,
      "step": 6306
    },
    {
      "epoch": 0.9512103159641053,
      "grad_norm": 0.25181008089940937,
      "learning_rate": 5.263016088489568e-08,
      "loss": 1.5685,
      "step": 6307
    },
    {
      "epoch": 0.9513611341527788,
      "grad_norm": 0.2421353533048425,
      "learning_rate": 5.261393182741514e-08,
      "loss": 1.556,
      "step": 6308
    },
    {
      "epoch": 0.9515119523414524,
      "grad_norm": 0.251096429171215,
      "learning_rate": 5.2597752702083514e-08,
      "loss": 1.4922,
      "step": 6309
    },
    {
      "epoch": 0.9516627705301259,
      "grad_norm": 0.37880347467434033,
      "learning_rate": 5.258162351253348e-08,
      "loss": 1.5302,
      "step": 6310
    },
    {
      "epoch": 0.9518135887187995,
      "grad_norm": 0.255749362520709,
      "learning_rate": 5.2565544262386504e-08,
      "loss": 1.5452,
      "step": 6311
    },
    {
      "epoch": 0.9519644069074731,
      "grad_norm": 0.24572970039643124,
      "learning_rate": 5.254951495525285e-08,
      "loss": 1.5086,
      "step": 6312
    },
    {
      "epoch": 0.9521152250961465,
      "grad_norm": 0.2671105386365278,
      "learning_rate": 5.2533535594731544e-08,
      "loss": 1.5678,
      "step": 6313
    },
    {
      "epoch": 0.9522660432848201,
      "grad_norm": 0.26509869032789696,
      "learning_rate": 5.251760618441047e-08,
      "loss": 1.5863,
      "step": 6314
    },
    {
      "epoch": 0.9524168614734937,
      "grad_norm": 0.24674607540460267,
      "learning_rate": 5.250172672786618e-08,
      "loss": 1.5591,
      "step": 6315
    },
    {
      "epoch": 0.9525676796621673,
      "grad_norm": 0.2810966899454014,
      "learning_rate": 5.248589722866413e-08,
      "loss": 1.5854,
      "step": 6316
    },
    {
      "epoch": 0.9527184978508408,
      "grad_norm": 0.2552064134059601,
      "learning_rate": 5.247011769035848e-08,
      "loss": 1.5454,
      "step": 6317
    },
    {
      "epoch": 0.9528693160395144,
      "grad_norm": 0.2616743545212081,
      "learning_rate": 5.2454388116492156e-08,
      "loss": 1.5682,
      "step": 6318
    },
    {
      "epoch": 0.9530201342281879,
      "grad_norm": 0.24562816546877292,
      "learning_rate": 5.2438708510596976e-08,
      "loss": 1.5482,
      "step": 6319
    },
    {
      "epoch": 0.9531709524168615,
      "grad_norm": 0.25582222827139806,
      "learning_rate": 5.242307887619344e-08,
      "loss": 1.5549,
      "step": 6320
    },
    {
      "epoch": 0.953321770605535,
      "grad_norm": 0.25225601203377085,
      "learning_rate": 5.2407499216790834e-08,
      "loss": 1.5699,
      "step": 6321
    },
    {
      "epoch": 0.9534725887942086,
      "grad_norm": 0.24564441907729032,
      "learning_rate": 5.2391969535887276e-08,
      "loss": 1.643,
      "step": 6322
    },
    {
      "epoch": 0.9536234069828822,
      "grad_norm": 0.24492164355094914,
      "learning_rate": 5.237648983696961e-08,
      "loss": 1.5292,
      "step": 6323
    },
    {
      "epoch": 0.9537742251715557,
      "grad_norm": 0.2706662227904134,
      "learning_rate": 5.23610601235135e-08,
      "loss": 1.5519,
      "step": 6324
    },
    {
      "epoch": 0.9539250433602292,
      "grad_norm": 0.25127336107480863,
      "learning_rate": 5.234568039898336e-08,
      "loss": 1.6202,
      "step": 6325
    },
    {
      "epoch": 0.9540758615489028,
      "grad_norm": 0.24323633765490138,
      "learning_rate": 5.233035066683238e-08,
      "loss": 1.564,
      "step": 6326
    },
    {
      "epoch": 0.9542266797375764,
      "grad_norm": 0.3904116632792208,
      "learning_rate": 5.2315070930502523e-08,
      "loss": 1.5072,
      "step": 6327
    },
    {
      "epoch": 0.9543774979262499,
      "grad_norm": 0.24535791701400553,
      "learning_rate": 5.229984119342452e-08,
      "loss": 1.6105,
      "step": 6328
    },
    {
      "epoch": 0.9545283161149235,
      "grad_norm": 0.24983407654198528,
      "learning_rate": 5.2284661459017956e-08,
      "loss": 1.6082,
      "step": 6329
    },
    {
      "epoch": 0.954679134303597,
      "grad_norm": 0.2735020691691619,
      "learning_rate": 5.2269531730691056e-08,
      "loss": 1.61,
      "step": 6330
    },
    {
      "epoch": 0.9548299524922705,
      "grad_norm": 0.3316571241912269,
      "learning_rate": 5.2254452011840936e-08,
      "loss": 1.6008,
      "step": 6331
    },
    {
      "epoch": 0.9549807706809441,
      "grad_norm": 0.24588425535404385,
      "learning_rate": 5.223942230585339e-08,
      "loss": 1.5382,
      "step": 6332
    },
    {
      "epoch": 0.9551315888696177,
      "grad_norm": 0.2410493398534231,
      "learning_rate": 5.222444261610305e-08,
      "loss": 1.5555,
      "step": 6333
    },
    {
      "epoch": 0.9552824070582913,
      "grad_norm": 0.27248429010886993,
      "learning_rate": 5.220951294595327e-08,
      "loss": 1.5324,
      "step": 6334
    },
    {
      "epoch": 0.9554332252469648,
      "grad_norm": 0.2531146404786899,
      "learning_rate": 5.219463329875622e-08,
      "loss": 1.5819,
      "step": 6335
    },
    {
      "epoch": 0.9555840434356383,
      "grad_norm": 0.3264049251970853,
      "learning_rate": 5.21798036778528e-08,
      "loss": 1.5493,
      "step": 6336
    },
    {
      "epoch": 0.9557348616243119,
      "grad_norm": 0.2673903371298123,
      "learning_rate": 5.216502408657271e-08,
      "loss": 1.5503,
      "step": 6337
    },
    {
      "epoch": 0.9558856798129854,
      "grad_norm": 0.25067687301545893,
      "learning_rate": 5.215029452823435e-08,
      "loss": 1.542,
      "step": 6338
    },
    {
      "epoch": 0.956036498001659,
      "grad_norm": 0.24137452966892348,
      "learning_rate": 5.213561500614498e-08,
      "loss": 1.5222,
      "step": 6339
    },
    {
      "epoch": 0.9561873161903326,
      "grad_norm": 0.2985998380684002,
      "learning_rate": 5.21209855236006e-08,
      "loss": 1.5531,
      "step": 6340
    },
    {
      "epoch": 0.956338134379006,
      "grad_norm": 0.2575262759632998,
      "learning_rate": 5.21064060838859e-08,
      "loss": 1.5532,
      "step": 6341
    },
    {
      "epoch": 0.9564889525676796,
      "grad_norm": 0.26226393673791865,
      "learning_rate": 5.2091876690274414e-08,
      "loss": 1.5486,
      "step": 6342
    },
    {
      "epoch": 0.9566397707563532,
      "grad_norm": 0.25781586813383506,
      "learning_rate": 5.2077397346028425e-08,
      "loss": 1.6117,
      "step": 6343
    },
    {
      "epoch": 0.9567905889450268,
      "grad_norm": 0.24805595398722974,
      "learning_rate": 5.2062968054398935e-08,
      "loss": 1.5863,
      "step": 6344
    },
    {
      "epoch": 0.9569414071337004,
      "grad_norm": 0.24439234848724492,
      "learning_rate": 5.2048588818625784e-08,
      "loss": 1.6135,
      "step": 6345
    },
    {
      "epoch": 0.9570922253223739,
      "grad_norm": 0.4701836790860895,
      "learning_rate": 5.2034259641937486e-08,
      "loss": 1.5165,
      "step": 6346
    },
    {
      "epoch": 0.9572430435110474,
      "grad_norm": 0.24201946522129109,
      "learning_rate": 5.201998052755139e-08,
      "loss": 1.5647,
      "step": 6347
    },
    {
      "epoch": 0.957393861699721,
      "grad_norm": 0.24220374354816995,
      "learning_rate": 5.200575147867356e-08,
      "loss": 1.4992,
      "step": 6348
    },
    {
      "epoch": 0.9575446798883945,
      "grad_norm": 0.2541763196845834,
      "learning_rate": 5.199157249849884e-08,
      "loss": 1.5549,
      "step": 6349
    },
    {
      "epoch": 0.9576954980770681,
      "grad_norm": 0.23701110036162357,
      "learning_rate": 5.19774435902108e-08,
      "loss": 1.5567,
      "step": 6350
    },
    {
      "epoch": 0.9578463162657417,
      "grad_norm": 0.2466884716259338,
      "learning_rate": 5.1963364756981814e-08,
      "loss": 1.6295,
      "step": 6351
    },
    {
      "epoch": 0.9579971344544153,
      "grad_norm": 0.2565904248345054,
      "learning_rate": 5.1949336001973e-08,
      "loss": 1.58,
      "step": 6352
    },
    {
      "epoch": 0.9581479526430887,
      "grad_norm": 0.26585377051506953,
      "learning_rate": 5.193535732833418e-08,
      "loss": 1.5371,
      "step": 6353
    },
    {
      "epoch": 0.9582987708317623,
      "grad_norm": 0.26269672466200317,
      "learning_rate": 5.192142873920401e-08,
      "loss": 1.6572,
      "step": 6354
    },
    {
      "epoch": 0.9584495890204359,
      "grad_norm": 0.2683391523084997,
      "learning_rate": 5.1907550237709877e-08,
      "loss": 1.594,
      "step": 6355
    },
    {
      "epoch": 0.9586004072091094,
      "grad_norm": 0.25262581475921414,
      "learning_rate": 5.189372182696786e-08,
      "loss": 1.5137,
      "step": 6356
    },
    {
      "epoch": 0.958751225397783,
      "grad_norm": 0.2730807174740343,
      "learning_rate": 5.187994351008288e-08,
      "loss": 1.5568,
      "step": 6357
    },
    {
      "epoch": 0.9589020435864565,
      "grad_norm": 0.25856596758541345,
      "learning_rate": 5.1866215290148544e-08,
      "loss": 1.5104,
      "step": 6358
    },
    {
      "epoch": 0.95905286177513,
      "grad_norm": 0.6752805259594179,
      "learning_rate": 5.1852537170247246e-08,
      "loss": 1.5019,
      "step": 6359
    },
    {
      "epoch": 0.9592036799638036,
      "grad_norm": 0.27077413598273226,
      "learning_rate": 5.183890915345015e-08,
      "loss": 1.5844,
      "step": 6360
    },
    {
      "epoch": 0.9593544981524772,
      "grad_norm": 0.2620727508929476,
      "learning_rate": 5.1825331242817094e-08,
      "loss": 1.5553,
      "step": 6361
    },
    {
      "epoch": 0.9595053163411508,
      "grad_norm": 0.24557656249048038,
      "learning_rate": 5.181180344139676e-08,
      "loss": 1.5322,
      "step": 6362
    },
    {
      "epoch": 0.9596561345298243,
      "grad_norm": 0.24812302527421484,
      "learning_rate": 5.179832575222651e-08,
      "loss": 1.6082,
      "step": 6363
    },
    {
      "epoch": 0.9598069527184978,
      "grad_norm": 0.26450100421447964,
      "learning_rate": 5.1784898178332455e-08,
      "loss": 1.5796,
      "step": 6364
    },
    {
      "epoch": 0.9599577709071714,
      "grad_norm": 0.24817179902484185,
      "learning_rate": 5.177152072272952e-08,
      "loss": 1.5181,
      "step": 6365
    },
    {
      "epoch": 0.960108589095845,
      "grad_norm": 0.31622965878024756,
      "learning_rate": 5.175819338842133e-08,
      "loss": 1.5046,
      "step": 6366
    },
    {
      "epoch": 0.9602594072845185,
      "grad_norm": 0.2484048976941183,
      "learning_rate": 5.174491617840023e-08,
      "loss": 1.547,
      "step": 6367
    },
    {
      "epoch": 0.9604102254731921,
      "grad_norm": 0.24385330701567645,
      "learning_rate": 5.1731689095647344e-08,
      "loss": 1.5372,
      "step": 6368
    },
    {
      "epoch": 0.9605610436618656,
      "grad_norm": 0.2516939386965962,
      "learning_rate": 5.171851214313254e-08,
      "loss": 1.5466,
      "step": 6369
    },
    {
      "epoch": 0.9607118618505391,
      "grad_norm": 0.2556617679977314,
      "learning_rate": 5.170538532381446e-08,
      "loss": 1.619,
      "step": 6370
    },
    {
      "epoch": 0.9608626800392127,
      "grad_norm": 0.25530410544496274,
      "learning_rate": 5.1692308640640416e-08,
      "loss": 1.5737,
      "step": 6371
    },
    {
      "epoch": 0.9610134982278863,
      "grad_norm": 0.25699963969499623,
      "learning_rate": 5.1679282096546535e-08,
      "loss": 1.6024,
      "step": 6372
    },
    {
      "epoch": 0.9611643164165599,
      "grad_norm": 0.25245499108689606,
      "learning_rate": 5.1666305694457646e-08,
      "loss": 1.6041,
      "step": 6373
    },
    {
      "epoch": 0.9613151346052334,
      "grad_norm": 0.2495170948349133,
      "learning_rate": 5.16533794372873e-08,
      "loss": 1.5568,
      "step": 6374
    },
    {
      "epoch": 0.9614659527939069,
      "grad_norm": 0.2518591639384356,
      "learning_rate": 5.164050332793786e-08,
      "loss": 1.6363,
      "step": 6375
    },
    {
      "epoch": 0.9616167709825805,
      "grad_norm": 0.28856757112588033,
      "learning_rate": 5.162767736930037e-08,
      "loss": 1.5222,
      "step": 6376
    },
    {
      "epoch": 0.961767589171254,
      "grad_norm": 0.2541937128631411,
      "learning_rate": 5.161490156425461e-08,
      "loss": 1.5362,
      "step": 6377
    },
    {
      "epoch": 0.9619184073599276,
      "grad_norm": 0.338367967854621,
      "learning_rate": 5.1602175915669166e-08,
      "loss": 1.6243,
      "step": 6378
    },
    {
      "epoch": 0.9620692255486012,
      "grad_norm": 0.3627880610386085,
      "learning_rate": 5.158950042640128e-08,
      "loss": 1.5381,
      "step": 6379
    },
    {
      "epoch": 0.9622200437372748,
      "grad_norm": 0.282055593543598,
      "learning_rate": 5.1576875099297006e-08,
      "loss": 1.5844,
      "step": 6380
    },
    {
      "epoch": 0.9623708619259482,
      "grad_norm": 0.25053425306125043,
      "learning_rate": 5.156429993719104e-08,
      "loss": 1.5137,
      "step": 6381
    },
    {
      "epoch": 0.9625216801146218,
      "grad_norm": 0.44214031666076936,
      "learning_rate": 5.155177494290691e-08,
      "loss": 1.5772,
      "step": 6382
    },
    {
      "epoch": 0.9626724983032954,
      "grad_norm": 0.2482561354350461,
      "learning_rate": 5.1539300119256864e-08,
      "loss": 1.5061,
      "step": 6383
    },
    {
      "epoch": 0.962823316491969,
      "grad_norm": 0.26087108410790427,
      "learning_rate": 5.152687546904182e-08,
      "loss": 1.554,
      "step": 6384
    },
    {
      "epoch": 0.9629741346806425,
      "grad_norm": 0.283677071517818,
      "learning_rate": 5.151450099505147e-08,
      "loss": 1.5427,
      "step": 6385
    },
    {
      "epoch": 0.963124952869316,
      "grad_norm": 0.36027328623910293,
      "learning_rate": 5.1502176700064304e-08,
      "loss": 1.5881,
      "step": 6386
    },
    {
      "epoch": 0.9632757710579896,
      "grad_norm": 0.24136512996506998,
      "learning_rate": 5.1489902586847397e-08,
      "loss": 1.5579,
      "step": 6387
    },
    {
      "epoch": 0.9634265892466631,
      "grad_norm": 0.276326793171099,
      "learning_rate": 5.147767865815673e-08,
      "loss": 1.5579,
      "step": 6388
    },
    {
      "epoch": 0.9635774074353367,
      "grad_norm": 0.2661932526640572,
      "learning_rate": 5.14655049167369e-08,
      "loss": 1.5575,
      "step": 6389
    },
    {
      "epoch": 0.9637282256240103,
      "grad_norm": 0.2768456724796517,
      "learning_rate": 5.1453381365321224e-08,
      "loss": 1.5478,
      "step": 6390
    },
    {
      "epoch": 0.9638790438126839,
      "grad_norm": 0.2672029887206416,
      "learning_rate": 5.1441308006631854e-08,
      "loss": 1.5583,
      "step": 6391
    },
    {
      "epoch": 0.9640298620013573,
      "grad_norm": 0.28307822253509946,
      "learning_rate": 5.1429284843379576e-08,
      "loss": 1.5894,
      "step": 6392
    },
    {
      "epoch": 0.9641806801900309,
      "grad_norm": 0.2584585522942879,
      "learning_rate": 5.141731187826395e-08,
      "loss": 1.5169,
      "step": 6393
    },
    {
      "epoch": 0.9643314983787045,
      "grad_norm": 0.2556591975295269,
      "learning_rate": 5.140538911397326e-08,
      "loss": 1.5851,
      "step": 6394
    },
    {
      "epoch": 0.964482316567378,
      "grad_norm": 0.2720527790539065,
      "learning_rate": 5.139351655318451e-08,
      "loss": 1.561,
      "step": 6395
    },
    {
      "epoch": 0.9646331347560516,
      "grad_norm": 0.28976825110477883,
      "learning_rate": 5.138169419856344e-08,
      "loss": 1.519,
      "step": 6396
    },
    {
      "epoch": 0.9647839529447252,
      "grad_norm": 3.967530858341636,
      "learning_rate": 5.136992205276449e-08,
      "loss": 1.5251,
      "step": 6397
    },
    {
      "epoch": 0.9649347711333986,
      "grad_norm": 0.24041622479641886,
      "learning_rate": 5.135820011843088e-08,
      "loss": 1.5575,
      "step": 6398
    },
    {
      "epoch": 0.9650855893220722,
      "grad_norm": 0.2605820726704934,
      "learning_rate": 5.13465283981945e-08,
      "loss": 1.579,
      "step": 6399
    },
    {
      "epoch": 0.9652364075107458,
      "grad_norm": 0.2560085767952434,
      "learning_rate": 5.1334906894676e-08,
      "loss": 1.5716,
      "step": 6400
    },
    {
      "epoch": 0.9653872256994194,
      "grad_norm": 0.24748805942426388,
      "learning_rate": 5.1323335610484776e-08,
      "loss": 1.5731,
      "step": 6401
    },
    {
      "epoch": 0.965538043888093,
      "grad_norm": 0.2654173067765607,
      "learning_rate": 5.131181454821887e-08,
      "loss": 1.6424,
      "step": 6402
    },
    {
      "epoch": 0.9656888620767664,
      "grad_norm": 0.3101653209941184,
      "learning_rate": 5.1300343710465113e-08,
      "loss": 1.5601,
      "step": 6403
    },
    {
      "epoch": 0.96583968026544,
      "grad_norm": 0.24537041958505784,
      "learning_rate": 5.128892309979907e-08,
      "loss": 1.598,
      "step": 6404
    },
    {
      "epoch": 0.9659904984541136,
      "grad_norm": 0.2949769881137243,
      "learning_rate": 5.1277552718784966e-08,
      "loss": 1.5294,
      "step": 6405
    },
    {
      "epoch": 0.9661413166427871,
      "grad_norm": 0.3355136951414702,
      "learning_rate": 5.12662325699758e-08,
      "loss": 1.613,
      "step": 6406
    },
    {
      "epoch": 0.9662921348314607,
      "grad_norm": 0.24091864982323638,
      "learning_rate": 5.1254962655913256e-08,
      "loss": 1.5554,
      "step": 6407
    },
    {
      "epoch": 0.9664429530201343,
      "grad_norm": 0.2687209411320317,
      "learning_rate": 5.124374297912777e-08,
      "loss": 1.4844,
      "step": 6408
    },
    {
      "epoch": 0.9665937712088077,
      "grad_norm": 0.26312670298435775,
      "learning_rate": 5.123257354213851e-08,
      "loss": 1.5538,
      "step": 6409
    },
    {
      "epoch": 0.9667445893974813,
      "grad_norm": 0.28145274288958727,
      "learning_rate": 5.122145434745329e-08,
      "loss": 1.437,
      "step": 6410
    },
    {
      "epoch": 0.9668954075861549,
      "grad_norm": 0.2838846037459194,
      "learning_rate": 5.1210385397568745e-08,
      "loss": 1.586,
      "step": 6411
    },
    {
      "epoch": 0.9670462257748285,
      "grad_norm": 0.2528764946796678,
      "learning_rate": 5.119936669497015e-08,
      "loss": 1.6836,
      "step": 6412
    },
    {
      "epoch": 0.967197043963502,
      "grad_norm": 0.2562476919374889,
      "learning_rate": 5.118839824213151e-08,
      "loss": 1.5663,
      "step": 6413
    },
    {
      "epoch": 0.9673478621521755,
      "grad_norm": 0.24206565595635035,
      "learning_rate": 5.1177480041515576e-08,
      "loss": 1.5931,
      "step": 6414
    },
    {
      "epoch": 0.9674986803408491,
      "grad_norm": 0.2629260321961326,
      "learning_rate": 5.1166612095573824e-08,
      "loss": 1.5754,
      "step": 6415
    },
    {
      "epoch": 0.9676494985295226,
      "grad_norm": 0.25671472522188277,
      "learning_rate": 5.1155794406746385e-08,
      "loss": 1.5733,
      "step": 6416
    },
    {
      "epoch": 0.9678003167181962,
      "grad_norm": 0.4533755136092952,
      "learning_rate": 5.11450269774622e-08,
      "loss": 1.5637,
      "step": 6417
    },
    {
      "epoch": 0.9679511349068698,
      "grad_norm": 0.25153301510486026,
      "learning_rate": 5.11343098101388e-08,
      "loss": 1.5059,
      "step": 6418
    },
    {
      "epoch": 0.9681019530955434,
      "grad_norm": 0.2870340121695763,
      "learning_rate": 5.112364290718255e-08,
      "loss": 1.6028,
      "step": 6419
    },
    {
      "epoch": 0.9682527712842168,
      "grad_norm": 0.2654495116873873,
      "learning_rate": 5.111302627098848e-08,
      "loss": 1.5071,
      "step": 6420
    },
    {
      "epoch": 0.9684035894728904,
      "grad_norm": 0.24384356461296303,
      "learning_rate": 5.110245990394031e-08,
      "loss": 1.5198,
      "step": 6421
    },
    {
      "epoch": 0.968554407661564,
      "grad_norm": 0.24775464926176569,
      "learning_rate": 5.1091943808410505e-08,
      "loss": 1.5057,
      "step": 6422
    },
    {
      "epoch": 0.9687052258502376,
      "grad_norm": 0.2634349039600724,
      "learning_rate": 5.108147798676023e-08,
      "loss": 1.5917,
      "step": 6423
    },
    {
      "epoch": 0.9688560440389111,
      "grad_norm": 0.2475589464139171,
      "learning_rate": 5.107106244133939e-08,
      "loss": 1.5126,
      "step": 6424
    },
    {
      "epoch": 0.9690068622275847,
      "grad_norm": 0.2500812810143837,
      "learning_rate": 5.1060697174486566e-08,
      "loss": 1.5617,
      "step": 6425
    },
    {
      "epoch": 0.9691576804162582,
      "grad_norm": 0.26121275475871664,
      "learning_rate": 5.105038218852906e-08,
      "loss": 1.557,
      "step": 6426
    },
    {
      "epoch": 0.9693084986049317,
      "grad_norm": 0.27842912631507954,
      "learning_rate": 5.104011748578287e-08,
      "loss": 1.5295,
      "step": 6427
    },
    {
      "epoch": 0.9694593167936053,
      "grad_norm": 0.5338303687886577,
      "learning_rate": 5.102990306855276e-08,
      "loss": 1.5238,
      "step": 6428
    },
    {
      "epoch": 0.9696101349822789,
      "grad_norm": 0.25920039211508383,
      "learning_rate": 5.101973893913211e-08,
      "loss": 1.5476,
      "step": 6429
    },
    {
      "epoch": 0.9697609531709525,
      "grad_norm": 0.29257572169992907,
      "learning_rate": 5.100962509980314e-08,
      "loss": 1.5912,
      "step": 6430
    },
    {
      "epoch": 0.9699117713596259,
      "grad_norm": 0.4974711068219046,
      "learning_rate": 5.0999561552836636e-08,
      "loss": 1.578,
      "step": 6431
    },
    {
      "epoch": 0.9700625895482995,
      "grad_norm": 0.2692008434386123,
      "learning_rate": 5.098954830049217e-08,
      "loss": 1.5373,
      "step": 6432
    },
    {
      "epoch": 0.9702134077369731,
      "grad_norm": 0.24493602251064997,
      "learning_rate": 5.0979585345018e-08,
      "loss": 1.5222,
      "step": 6433
    },
    {
      "epoch": 0.9703642259256466,
      "grad_norm": 0.24580616475390038,
      "learning_rate": 5.0969672688651134e-08,
      "loss": 1.5055,
      "step": 6434
    },
    {
      "epoch": 0.9705150441143202,
      "grad_norm": 0.40525484455776534,
      "learning_rate": 5.095981033361724e-08,
      "loss": 1.5819,
      "step": 6435
    },
    {
      "epoch": 0.9706658623029938,
      "grad_norm": 0.5323926365254915,
      "learning_rate": 5.094999828213068e-08,
      "loss": 1.5993,
      "step": 6436
    },
    {
      "epoch": 0.9708166804916672,
      "grad_norm": 0.2758624089121172,
      "learning_rate": 5.0940236536394586e-08,
      "loss": 1.5756,
      "step": 6437
    },
    {
      "epoch": 0.9709674986803408,
      "grad_norm": 0.30537970805051545,
      "learning_rate": 5.093052509860073e-08,
      "loss": 1.5215,
      "step": 6438
    },
    {
      "epoch": 0.9711183168690144,
      "grad_norm": 0.25306210515763267,
      "learning_rate": 5.0920863970929584e-08,
      "loss": 1.5335,
      "step": 6439
    },
    {
      "epoch": 0.971269135057688,
      "grad_norm": 0.2793419447136214,
      "learning_rate": 5.091125315555039e-08,
      "loss": 1.5796,
      "step": 6440
    },
    {
      "epoch": 0.9714199532463615,
      "grad_norm": 0.2635853250708283,
      "learning_rate": 5.090169265462107e-08,
      "loss": 1.5372,
      "step": 6441
    },
    {
      "epoch": 0.9715707714350351,
      "grad_norm": 0.24773665496139735,
      "learning_rate": 5.0892182470288205e-08,
      "loss": 1.5705,
      "step": 6442
    },
    {
      "epoch": 0.9717215896237086,
      "grad_norm": 0.2602379534612917,
      "learning_rate": 5.08827226046871e-08,
      "loss": 1.6155,
      "step": 6443
    },
    {
      "epoch": 0.9718724078123822,
      "grad_norm": 0.24904634275831725,
      "learning_rate": 5.0873313059941794e-08,
      "loss": 1.5483,
      "step": 6444
    },
    {
      "epoch": 0.9720232260010557,
      "grad_norm": 0.2558030518578314,
      "learning_rate": 5.086395383816501e-08,
      "loss": 1.5874,
      "step": 6445
    },
    {
      "epoch": 0.9721740441897293,
      "grad_norm": 0.25354510496858634,
      "learning_rate": 5.0854644941458136e-08,
      "loss": 1.5827,
      "step": 6446
    },
    {
      "epoch": 0.9723248623784029,
      "grad_norm": 0.250452222296652,
      "learning_rate": 5.084538637191133e-08,
      "loss": 1.5173,
      "step": 6447
    },
    {
      "epoch": 0.9724756805670763,
      "grad_norm": 0.2603913148074543,
      "learning_rate": 5.083617813160337e-08,
      "loss": 1.5508,
      "step": 6448
    },
    {
      "epoch": 0.9726264987557499,
      "grad_norm": 0.40930176727111084,
      "learning_rate": 5.082702022260179e-08,
      "loss": 1.5538,
      "step": 6449
    },
    {
      "epoch": 0.9727773169444235,
      "grad_norm": 0.2497056167405697,
      "learning_rate": 5.081791264696281e-08,
      "loss": 1.4941,
      "step": 6450
    },
    {
      "epoch": 0.9729281351330971,
      "grad_norm": 0.2550678396598978,
      "learning_rate": 5.080885540673135e-08,
      "loss": 1.5985,
      "step": 6451
    },
    {
      "epoch": 0.9730789533217706,
      "grad_norm": 0.2710859709151231,
      "learning_rate": 5.079984850394102e-08,
      "loss": 1.5682,
      "step": 6452
    },
    {
      "epoch": 0.9732297715104442,
      "grad_norm": 0.24717971177406345,
      "learning_rate": 5.079089194061412e-08,
      "loss": 1.5253,
      "step": 6453
    },
    {
      "epoch": 0.9733805896991177,
      "grad_norm": 0.24035360506745604,
      "learning_rate": 5.078198571876169e-08,
      "loss": 1.5121,
      "step": 6454
    },
    {
      "epoch": 0.9735314078877912,
      "grad_norm": 0.24981177472656527,
      "learning_rate": 5.077312984038339e-08,
      "loss": 1.5128,
      "step": 6455
    },
    {
      "epoch": 0.9736822260764648,
      "grad_norm": 0.2522474695358754,
      "learning_rate": 5.076432430746767e-08,
      "loss": 1.581,
      "step": 6456
    },
    {
      "epoch": 0.9738330442651384,
      "grad_norm": 0.2617174417485487,
      "learning_rate": 5.075556912199161e-08,
      "loss": 1.5678,
      "step": 6457
    },
    {
      "epoch": 0.973983862453812,
      "grad_norm": 0.24953025638107984,
      "learning_rate": 5.0746864285920995e-08,
      "loss": 1.587,
      "step": 6458
    },
    {
      "epoch": 0.9741346806424854,
      "grad_norm": 0.261470405145867,
      "learning_rate": 5.0738209801210296e-08,
      "loss": 1.5478,
      "step": 6459
    },
    {
      "epoch": 0.974285498831159,
      "grad_norm": 0.2520795520193516,
      "learning_rate": 5.0729605669802766e-08,
      "loss": 1.5086,
      "step": 6460
    },
    {
      "epoch": 0.9744363170198326,
      "grad_norm": 0.3244120204281703,
      "learning_rate": 5.0721051893630216e-08,
      "loss": 1.6022,
      "step": 6461
    },
    {
      "epoch": 0.9745871352085062,
      "grad_norm": 0.26256172500630426,
      "learning_rate": 5.071254847461322e-08,
      "loss": 1.5372,
      "step": 6462
    },
    {
      "epoch": 0.9747379533971797,
      "grad_norm": 0.23743723766173824,
      "learning_rate": 5.0704095414661084e-08,
      "loss": 1.5135,
      "step": 6463
    },
    {
      "epoch": 0.9748887715858533,
      "grad_norm": 0.29242581161455566,
      "learning_rate": 5.0695692715671756e-08,
      "loss": 1.553,
      "step": 6464
    },
    {
      "epoch": 0.9750395897745268,
      "grad_norm": 0.3117616067572284,
      "learning_rate": 5.0687340379531864e-08,
      "loss": 1.5107,
      "step": 6465
    },
    {
      "epoch": 0.9751904079632003,
      "grad_norm": 0.24530123805488938,
      "learning_rate": 5.067903840811679e-08,
      "loss": 1.5374,
      "step": 6466
    },
    {
      "epoch": 0.9753412261518739,
      "grad_norm": 0.24600277932465653,
      "learning_rate": 5.0670786803290514e-08,
      "loss": 1.5453,
      "step": 6467
    },
    {
      "epoch": 0.9754920443405475,
      "grad_norm": 0.24954792101445722,
      "learning_rate": 5.066258556690581e-08,
      "loss": 1.5429,
      "step": 6468
    },
    {
      "epoch": 0.975642862529221,
      "grad_norm": 0.3665210065866042,
      "learning_rate": 5.065443470080404e-08,
      "loss": 1.5777,
      "step": 6469
    },
    {
      "epoch": 0.9757936807178946,
      "grad_norm": 0.24878039696491128,
      "learning_rate": 5.0646334206815386e-08,
      "loss": 1.5602,
      "step": 6470
    },
    {
      "epoch": 0.9759444989065681,
      "grad_norm": 0.271514195335307,
      "learning_rate": 5.063828408675858e-08,
      "loss": 1.5214,
      "step": 6471
    },
    {
      "epoch": 0.9760953170952417,
      "grad_norm": 0.2663042992823365,
      "learning_rate": 5.063028434244113e-08,
      "loss": 1.5173,
      "step": 6472
    },
    {
      "epoch": 0.9762461352839152,
      "grad_norm": 0.23886729520101155,
      "learning_rate": 5.062233497565922e-08,
      "loss": 1.5354,
      "step": 6473
    },
    {
      "epoch": 0.9763969534725888,
      "grad_norm": 0.26097968730758986,
      "learning_rate": 5.0614435988197694e-08,
      "loss": 1.5131,
      "step": 6474
    },
    {
      "epoch": 0.9765477716612624,
      "grad_norm": 0.2957295881693974,
      "learning_rate": 5.060658738183012e-08,
      "loss": 1.563,
      "step": 6475
    },
    {
      "epoch": 0.9766985898499358,
      "grad_norm": 0.25273515987338896,
      "learning_rate": 5.059878915831877e-08,
      "loss": 1.5387,
      "step": 6476
    },
    {
      "epoch": 0.9768494080386094,
      "grad_norm": 0.24849187140414217,
      "learning_rate": 5.059104131941452e-08,
      "loss": 1.5624,
      "step": 6477
    },
    {
      "epoch": 0.977000226227283,
      "grad_norm": 0.26336256406420006,
      "learning_rate": 5.058334386685699e-08,
      "loss": 1.5553,
      "step": 6478
    },
    {
      "epoch": 0.9771510444159566,
      "grad_norm": 0.2505690502220122,
      "learning_rate": 5.0575696802374496e-08,
      "loss": 1.5526,
      "step": 6479
    },
    {
      "epoch": 0.9773018626046301,
      "grad_norm": 0.23844152313984407,
      "learning_rate": 5.056810012768405e-08,
      "loss": 1.5126,
      "step": 6480
    },
    {
      "epoch": 0.9774526807933037,
      "grad_norm": 0.3296402229519655,
      "learning_rate": 5.056055384449129e-08,
      "loss": 1.5577,
      "step": 6481
    },
    {
      "epoch": 0.9776034989819772,
      "grad_norm": 0.251433847339324,
      "learning_rate": 5.055305795449061e-08,
      "loss": 1.5756,
      "step": 6482
    },
    {
      "epoch": 0.9777543171706508,
      "grad_norm": 0.23745387328194084,
      "learning_rate": 5.054561245936502e-08,
      "loss": 1.5269,
      "step": 6483
    },
    {
      "epoch": 0.9779051353593243,
      "grad_norm": 0.2431422102768408,
      "learning_rate": 5.053821736078627e-08,
      "loss": 1.5274,
      "step": 6484
    },
    {
      "epoch": 0.9780559535479979,
      "grad_norm": 0.34551121164950643,
      "learning_rate": 5.053087266041475e-08,
      "loss": 1.5602,
      "step": 6485
    },
    {
      "epoch": 0.9782067717366715,
      "grad_norm": 0.25410630677517887,
      "learning_rate": 5.052357835989963e-08,
      "loss": 1.4949,
      "step": 6486
    },
    {
      "epoch": 0.978357589925345,
      "grad_norm": 0.26531649825766734,
      "learning_rate": 5.05163344608786e-08,
      "loss": 1.4913,
      "step": 6487
    },
    {
      "epoch": 0.9785084081140185,
      "grad_norm": 0.26781315412225526,
      "learning_rate": 5.05091409649782e-08,
      "loss": 1.5396,
      "step": 6488
    },
    {
      "epoch": 0.9786592263026921,
      "grad_norm": 0.2520580299099175,
      "learning_rate": 5.050199787381354e-08,
      "loss": 1.5628,
      "step": 6489
    },
    {
      "epoch": 0.9788100444913657,
      "grad_norm": 0.3804986834247732,
      "learning_rate": 5.049490518898845e-08,
      "loss": 1.5316,
      "step": 6490
    },
    {
      "epoch": 0.9789608626800392,
      "grad_norm": 1.244187494723972,
      "learning_rate": 5.048786291209548e-08,
      "loss": 1.5458,
      "step": 6491
    },
    {
      "epoch": 0.9791116808687128,
      "grad_norm": 0.23424370161157604,
      "learning_rate": 5.048087104471578e-08,
      "loss": 1.5631,
      "step": 6492
    },
    {
      "epoch": 0.9792624990573863,
      "grad_norm": 0.25032519116588015,
      "learning_rate": 5.047392958841927e-08,
      "loss": 1.5827,
      "step": 6493
    },
    {
      "epoch": 0.9794133172460598,
      "grad_norm": 0.24731029803020294,
      "learning_rate": 5.0467038544764466e-08,
      "loss": 1.5252,
      "step": 6494
    },
    {
      "epoch": 0.9795641354347334,
      "grad_norm": 0.2624072760981343,
      "learning_rate": 5.046019791529864e-08,
      "loss": 1.6048,
      "step": 6495
    },
    {
      "epoch": 0.979714953623407,
      "grad_norm": 0.2924639644547172,
      "learning_rate": 5.045340770155771e-08,
      "loss": 1.5289,
      "step": 6496
    },
    {
      "epoch": 0.9798657718120806,
      "grad_norm": 0.24884598029170174,
      "learning_rate": 5.0446667905066276e-08,
      "loss": 1.5927,
      "step": 6497
    },
    {
      "epoch": 0.9800165900007541,
      "grad_norm": 0.27924775476960245,
      "learning_rate": 5.0439978527337586e-08,
      "loss": 1.5717,
      "step": 6498
    },
    {
      "epoch": 0.9801674081894276,
      "grad_norm": 0.25301986563185713,
      "learning_rate": 5.043333956987366e-08,
      "loss": 1.5623,
      "step": 6499
    },
    {
      "epoch": 0.9803182263781012,
      "grad_norm": 0.27688842373854006,
      "learning_rate": 5.0426751034165076e-08,
      "loss": 1.5664,
      "step": 6500
    },
    {
      "epoch": 0.9804690445667748,
      "grad_norm": 0.24445084259797586,
      "learning_rate": 5.0420212921691204e-08,
      "loss": 1.59,
      "step": 6501
    },
    {
      "epoch": 0.9806198627554483,
      "grad_norm": 0.24344584631254262,
      "learning_rate": 5.041372523392e-08,
      "loss": 1.5412,
      "step": 6502
    },
    {
      "epoch": 0.9807706809441219,
      "grad_norm": 0.2526570631672275,
      "learning_rate": 5.040728797230817e-08,
      "loss": 1.5631,
      "step": 6503
    },
    {
      "epoch": 0.9809214991327954,
      "grad_norm": 0.3186108326288003,
      "learning_rate": 5.040090113830105e-08,
      "loss": 1.5801,
      "step": 6504
    },
    {
      "epoch": 0.9810723173214689,
      "grad_norm": 0.2986905201284329,
      "learning_rate": 5.039456473333268e-08,
      "loss": 1.5891,
      "step": 6505
    },
    {
      "epoch": 0.9812231355101425,
      "grad_norm": 0.24819163687733814,
      "learning_rate": 5.0388278758825745e-08,
      "loss": 1.5818,
      "step": 6506
    },
    {
      "epoch": 0.9813739536988161,
      "grad_norm": 0.2576431292514359,
      "learning_rate": 5.038204321619165e-08,
      "loss": 1.5484,
      "step": 6507
    },
    {
      "epoch": 0.9815247718874897,
      "grad_norm": 1.1781340006336096,
      "learning_rate": 5.037585810683046e-08,
      "loss": 1.5901,
      "step": 6508
    },
    {
      "epoch": 0.9816755900761632,
      "grad_norm": 0.25925081343451833,
      "learning_rate": 5.03697234321309e-08,
      "loss": 1.5672,
      "step": 6509
    },
    {
      "epoch": 0.9818264082648367,
      "grad_norm": 0.24633666848558364,
      "learning_rate": 5.0363639193470386e-08,
      "loss": 1.5217,
      "step": 6510
    },
    {
      "epoch": 0.9819772264535103,
      "grad_norm": 0.2390602820930981,
      "learning_rate": 5.0357605392215e-08,
      "loss": 1.5268,
      "step": 6511
    },
    {
      "epoch": 0.9821280446421838,
      "grad_norm": 0.26011007605635605,
      "learning_rate": 5.0351622029719506e-08,
      "loss": 1.5172,
      "step": 6512
    },
    {
      "epoch": 0.9822788628308574,
      "grad_norm": 0.24076616630485573,
      "learning_rate": 5.034568910732737e-08,
      "loss": 1.6221,
      "step": 6513
    },
    {
      "epoch": 0.982429681019531,
      "grad_norm": 0.27744182607071605,
      "learning_rate": 5.033980662637067e-08,
      "loss": 1.6523,
      "step": 6514
    },
    {
      "epoch": 0.9825804992082046,
      "grad_norm": 0.26065840247835165,
      "learning_rate": 5.03339745881702e-08,
      "loss": 1.6047,
      "step": 6515
    },
    {
      "epoch": 0.982731317396878,
      "grad_norm": 0.24838784838301078,
      "learning_rate": 5.0328192994035456e-08,
      "loss": 1.5765,
      "step": 6516
    },
    {
      "epoch": 0.9828821355855516,
      "grad_norm": 0.24992476928306281,
      "learning_rate": 5.0322461845264545e-08,
      "loss": 1.5043,
      "step": 6517
    },
    {
      "epoch": 0.9830329537742252,
      "grad_norm": 0.24636638294071833,
      "learning_rate": 5.031678114314429e-08,
      "loss": 1.5504,
      "step": 6518
    },
    {
      "epoch": 0.9831837719628987,
      "grad_norm": 0.25322163197820124,
      "learning_rate": 5.031115088895015e-08,
      "loss": 1.5645,
      "step": 6519
    },
    {
      "epoch": 0.9833345901515723,
      "grad_norm": 0.28106231255069447,
      "learning_rate": 5.0305571083946305e-08,
      "loss": 1.5788,
      "step": 6520
    },
    {
      "epoch": 0.9834854083402458,
      "grad_norm": 0.2553265807543555,
      "learning_rate": 5.030004172938559e-08,
      "loss": 1.5801,
      "step": 6521
    },
    {
      "epoch": 0.9836362265289194,
      "grad_norm": 0.2637114564187194,
      "learning_rate": 5.029456282650948e-08,
      "loss": 1.5569,
      "step": 6522
    },
    {
      "epoch": 0.9837870447175929,
      "grad_norm": 0.25801532365856095,
      "learning_rate": 5.028913437654816e-08,
      "loss": 1.5832,
      "step": 6523
    },
    {
      "epoch": 0.9839378629062665,
      "grad_norm": 0.24938123649171232,
      "learning_rate": 5.028375638072049e-08,
      "loss": 1.5996,
      "step": 6524
    },
    {
      "epoch": 0.9840886810949401,
      "grad_norm": 0.3509246362741187,
      "learning_rate": 5.027842884023395e-08,
      "loss": 1.6541,
      "step": 6525
    },
    {
      "epoch": 0.9842394992836137,
      "grad_norm": 0.3108705383194768,
      "learning_rate": 5.027315175628478e-08,
      "loss": 1.5554,
      "step": 6526
    },
    {
      "epoch": 0.9843903174722871,
      "grad_norm": 0.24457863554702414,
      "learning_rate": 5.026792513005779e-08,
      "loss": 1.5216,
      "step": 6527
    },
    {
      "epoch": 0.9845411356609607,
      "grad_norm": 0.26069030549313427,
      "learning_rate": 5.02627489627265e-08,
      "loss": 1.5288,
      "step": 6528
    },
    {
      "epoch": 0.9846919538496343,
      "grad_norm": 0.24747632060797806,
      "learning_rate": 5.025762325545319e-08,
      "loss": 1.5329,
      "step": 6529
    },
    {
      "epoch": 0.9848427720383078,
      "grad_norm": 0.2757738150564607,
      "learning_rate": 5.025254800938867e-08,
      "loss": 1.522,
      "step": 6530
    },
    {
      "epoch": 0.9849935902269814,
      "grad_norm": 0.26346233480909437,
      "learning_rate": 5.024752322567248e-08,
      "loss": 1.4972,
      "step": 6531
    },
    {
      "epoch": 0.985144408415655,
      "grad_norm": 0.25503332889486574,
      "learning_rate": 5.024254890543284e-08,
      "loss": 1.6004,
      "step": 6532
    },
    {
      "epoch": 0.9852952266043284,
      "grad_norm": 0.24756468469656975,
      "learning_rate": 5.023762504978663e-08,
      "loss": 1.4894,
      "step": 6533
    },
    {
      "epoch": 0.985446044793002,
      "grad_norm": 0.24112569122639588,
      "learning_rate": 5.0232751659839414e-08,
      "loss": 1.5443,
      "step": 6534
    },
    {
      "epoch": 0.9855968629816756,
      "grad_norm": 0.2719279864027767,
      "learning_rate": 5.0227928736685394e-08,
      "loss": 1.6128,
      "step": 6535
    },
    {
      "epoch": 0.9857476811703492,
      "grad_norm": 1.1656734563879858,
      "learning_rate": 5.0223156281407456e-08,
      "loss": 1.568,
      "step": 6536
    },
    {
      "epoch": 0.9858984993590227,
      "grad_norm": 0.28870201154578834,
      "learning_rate": 5.0218434295077155e-08,
      "loss": 1.5104,
      "step": 6537
    },
    {
      "epoch": 0.9860493175476962,
      "grad_norm": 0.25067612590232297,
      "learning_rate": 5.0213762778754706e-08,
      "loss": 1.5537,
      "step": 6538
    },
    {
      "epoch": 0.9862001357363698,
      "grad_norm": 0.3021000063641636,
      "learning_rate": 5.0209141733489046e-08,
      "loss": 1.5324,
      "step": 6539
    },
    {
      "epoch": 0.9863509539250433,
      "grad_norm": 0.26216923631130623,
      "learning_rate": 5.020457116031768e-08,
      "loss": 1.5644,
      "step": 6540
    },
    {
      "epoch": 0.9865017721137169,
      "grad_norm": 0.2656526626019676,
      "learning_rate": 5.0200051060266824e-08,
      "loss": 1.5785,
      "step": 6541
    },
    {
      "epoch": 0.9866525903023905,
      "grad_norm": 0.24813683444140744,
      "learning_rate": 5.0195581434351445e-08,
      "loss": 1.595,
      "step": 6542
    },
    {
      "epoch": 0.9868034084910641,
      "grad_norm": 0.24865637037750893,
      "learning_rate": 5.019116228357505e-08,
      "loss": 1.626,
      "step": 6543
    },
    {
      "epoch": 0.9869542266797375,
      "grad_norm": 0.24656800678491447,
      "learning_rate": 5.018679360892989e-08,
      "loss": 1.6314,
      "step": 6544
    },
    {
      "epoch": 0.9871050448684111,
      "grad_norm": 0.2483444651125244,
      "learning_rate": 5.018247541139684e-08,
      "loss": 1.5515,
      "step": 6545
    },
    {
      "epoch": 0.9872558630570847,
      "grad_norm": 0.2527243702013985,
      "learning_rate": 5.017820769194546e-08,
      "loss": 1.6493,
      "step": 6546
    },
    {
      "epoch": 0.9874066812457583,
      "grad_norm": 0.2517723592329947,
      "learning_rate": 5.0173990451533994e-08,
      "loss": 1.5266,
      "step": 6547
    },
    {
      "epoch": 0.9875574994344318,
      "grad_norm": 0.25317069918610413,
      "learning_rate": 5.016982369110933e-08,
      "loss": 1.56,
      "step": 6548
    },
    {
      "epoch": 0.9877083176231053,
      "grad_norm": 0.24608514103588505,
      "learning_rate": 5.016570741160703e-08,
      "loss": 1.5486,
      "step": 6549
    },
    {
      "epoch": 0.9878591358117789,
      "grad_norm": 0.2647537292214903,
      "learning_rate": 5.0161641613951295e-08,
      "loss": 1.5151,
      "step": 6550
    },
    {
      "epoch": 0.9880099540004524,
      "grad_norm": 0.24074774570637517,
      "learning_rate": 5.015762629905507e-08,
      "loss": 1.5666,
      "step": 6551
    },
    {
      "epoch": 0.988160772189126,
      "grad_norm": 0.26007525915617,
      "learning_rate": 5.0153661467819845e-08,
      "loss": 1.5034,
      "step": 6552
    },
    {
      "epoch": 0.9883115903777996,
      "grad_norm": 0.2516640942926316,
      "learning_rate": 5.014974712113589e-08,
      "loss": 1.5718,
      "step": 6553
    },
    {
      "epoch": 0.9884624085664732,
      "grad_norm": 0.24720404120112693,
      "learning_rate": 5.014588325988206e-08,
      "loss": 1.5559,
      "step": 6554
    },
    {
      "epoch": 0.9886132267551466,
      "grad_norm": 0.24841952692707817,
      "learning_rate": 5.0142069884925905e-08,
      "loss": 1.5363,
      "step": 6555
    },
    {
      "epoch": 0.9887640449438202,
      "grad_norm": 0.2486734212115933,
      "learning_rate": 5.013830699712365e-08,
      "loss": 1.4945,
      "step": 6556
    },
    {
      "epoch": 0.9889148631324938,
      "grad_norm": 0.24751562722403198,
      "learning_rate": 5.0134594597320185e-08,
      "loss": 1.6056,
      "step": 6557
    },
    {
      "epoch": 0.9890656813211673,
      "grad_norm": 0.261117367244557,
      "learning_rate": 5.0130932686349036e-08,
      "loss": 1.4504,
      "step": 6558
    },
    {
      "epoch": 0.9892164995098409,
      "grad_norm": 0.26202295217714505,
      "learning_rate": 5.012732126503238e-08,
      "loss": 1.5029,
      "step": 6559
    },
    {
      "epoch": 0.9893673176985145,
      "grad_norm": 0.24211227802711202,
      "learning_rate": 5.012376033418115e-08,
      "loss": 1.6037,
      "step": 6560
    },
    {
      "epoch": 0.989518135887188,
      "grad_norm": 0.26180647304427285,
      "learning_rate": 5.012024989459483e-08,
      "loss": 1.5926,
      "step": 6561
    },
    {
      "epoch": 0.9896689540758615,
      "grad_norm": 0.4177446113873971,
      "learning_rate": 5.011678994706165e-08,
      "loss": 1.542,
      "step": 6562
    },
    {
      "epoch": 0.9898197722645351,
      "grad_norm": 0.2539076636184375,
      "learning_rate": 5.011338049235842e-08,
      "loss": 1.6005,
      "step": 6563
    },
    {
      "epoch": 0.9899705904532087,
      "grad_norm": 0.2615796253744187,
      "learning_rate": 5.011002153125071e-08,
      "loss": 1.5413,
      "step": 6564
    },
    {
      "epoch": 0.9901214086418823,
      "grad_norm": 0.2960432928116584,
      "learning_rate": 5.010671306449267e-08,
      "loss": 1.631,
      "step": 6565
    },
    {
      "epoch": 0.9902722268305557,
      "grad_norm": 0.24457914186893268,
      "learning_rate": 5.010345509282717e-08,
      "loss": 1.5597,
      "step": 6566
    },
    {
      "epoch": 0.9904230450192293,
      "grad_norm": 0.24825750469450225,
      "learning_rate": 5.010024761698572e-08,
      "loss": 1.5476,
      "step": 6567
    },
    {
      "epoch": 0.9905738632079029,
      "grad_norm": 0.2572946824640707,
      "learning_rate": 5.0097090637688494e-08,
      "loss": 1.5519,
      "step": 6568
    },
    {
      "epoch": 0.9907246813965764,
      "grad_norm": 0.2859707766855372,
      "learning_rate": 5.0093984155644316e-08,
      "loss": 1.6001,
      "step": 6569
    },
    {
      "epoch": 0.99087549958525,
      "grad_norm": 0.24960053608418814,
      "learning_rate": 5.009092817155066e-08,
      "loss": 1.6219,
      "step": 6570
    },
    {
      "epoch": 0.9910263177739236,
      "grad_norm": 0.24169959370913324,
      "learning_rate": 5.008792268609373e-08,
      "loss": 1.5545,
      "step": 6571
    },
    {
      "epoch": 0.991177135962597,
      "grad_norm": 0.26847186988473953,
      "learning_rate": 5.00849676999483e-08,
      "loss": 1.6139,
      "step": 6572
    },
    {
      "epoch": 0.9913279541512706,
      "grad_norm": 0.24507802090346645,
      "learning_rate": 5.0082063213777876e-08,
      "loss": 1.5341,
      "step": 6573
    },
    {
      "epoch": 0.9914787723399442,
      "grad_norm": 0.24703925950319292,
      "learning_rate": 5.007920922823461e-08,
      "loss": 1.5049,
      "step": 6574
    },
    {
      "epoch": 0.9916295905286178,
      "grad_norm": 0.25278765284185717,
      "learning_rate": 5.007640574395928e-08,
      "loss": 1.5668,
      "step": 6575
    },
    {
      "epoch": 0.9917804087172913,
      "grad_norm": 0.25022451568994536,
      "learning_rate": 5.007365276158134e-08,
      "loss": 1.5477,
      "step": 6576
    },
    {
      "epoch": 0.9919312269059649,
      "grad_norm": 0.26152573903290816,
      "learning_rate": 5.0070950281718946e-08,
      "loss": 1.5317,
      "step": 6577
    },
    {
      "epoch": 0.9920820450946384,
      "grad_norm": 0.3338312726106041,
      "learning_rate": 5.006829830497889e-08,
      "loss": 1.5627,
      "step": 6578
    },
    {
      "epoch": 0.992232863283312,
      "grad_norm": 0.242145683944422,
      "learning_rate": 5.006569683195659e-08,
      "loss": 1.5797,
      "step": 6579
    },
    {
      "epoch": 0.9923836814719855,
      "grad_norm": 0.2458225912180188,
      "learning_rate": 5.006314586323615e-08,
      "loss": 1.4727,
      "step": 6580
    },
    {
      "epoch": 0.9925344996606591,
      "grad_norm": 0.2600251176366966,
      "learning_rate": 5.006064539939035e-08,
      "loss": 1.57,
      "step": 6581
    },
    {
      "epoch": 0.9926853178493327,
      "grad_norm": 0.24992863036321655,
      "learning_rate": 5.005819544098063e-08,
      "loss": 1.6949,
      "step": 6582
    },
    {
      "epoch": 0.9928361360380061,
      "grad_norm": 0.2593715127852107,
      "learning_rate": 5.005579598855705e-08,
      "loss": 1.591,
      "step": 6583
    },
    {
      "epoch": 0.9929869542266797,
      "grad_norm": 0.2599446696035754,
      "learning_rate": 5.005344704265839e-08,
      "loss": 1.585,
      "step": 6584
    },
    {
      "epoch": 0.9931377724153533,
      "grad_norm": 0.3857416727018453,
      "learning_rate": 5.005114860381201e-08,
      "loss": 1.5521,
      "step": 6585
    },
    {
      "epoch": 0.9932885906040269,
      "grad_norm": 0.25227183012276794,
      "learning_rate": 5.004890067253402e-08,
      "loss": 1.52,
      "step": 6586
    },
    {
      "epoch": 0.9934394087927004,
      "grad_norm": 0.270258631240915,
      "learning_rate": 5.0046703249329124e-08,
      "loss": 1.5643,
      "step": 6587
    },
    {
      "epoch": 0.993590226981374,
      "grad_norm": 0.2996266870859598,
      "learning_rate": 5.004455633469073e-08,
      "loss": 1.5496,
      "step": 6588
    },
    {
      "epoch": 0.9937410451700475,
      "grad_norm": 0.25308604768433385,
      "learning_rate": 5.0042459929100854e-08,
      "loss": 1.5256,
      "step": 6589
    },
    {
      "epoch": 0.993891863358721,
      "grad_norm": 0.25207421806595026,
      "learning_rate": 5.00404140330302e-08,
      "loss": 1.5814,
      "step": 6590
    },
    {
      "epoch": 0.9940426815473946,
      "grad_norm": 0.2695325793514296,
      "learning_rate": 5.003841864693814e-08,
      "loss": 1.6237,
      "step": 6591
    },
    {
      "epoch": 0.9941934997360682,
      "grad_norm": 0.620517491915544,
      "learning_rate": 5.003647377127272e-08,
      "loss": 1.5484,
      "step": 6592
    },
    {
      "epoch": 0.9943443179247418,
      "grad_norm": 0.2450898709745292,
      "learning_rate": 5.003457940647057e-08,
      "loss": 1.5471,
      "step": 6593
    },
    {
      "epoch": 0.9944951361134152,
      "grad_norm": 0.2587619496177874,
      "learning_rate": 5.0032735552957104e-08,
      "loss": 1.5542,
      "step": 6594
    },
    {
      "epoch": 0.9946459543020888,
      "grad_norm": 0.25998338882700783,
      "learning_rate": 5.003094221114625e-08,
      "loss": 1.6032,
      "step": 6595
    },
    {
      "epoch": 0.9947967724907624,
      "grad_norm": 0.2450711120072768,
      "learning_rate": 5.00291993814407e-08,
      "loss": 1.5722,
      "step": 6596
    },
    {
      "epoch": 0.994947590679436,
      "grad_norm": 0.29549061519392755,
      "learning_rate": 5.0027507064231754e-08,
      "loss": 1.5365,
      "step": 6597
    },
    {
      "epoch": 0.9950984088681095,
      "grad_norm": 0.24591074110055633,
      "learning_rate": 5.0025865259899395e-08,
      "loss": 1.5489,
      "step": 6598
    },
    {
      "epoch": 0.9952492270567831,
      "grad_norm": 0.2536244535593241,
      "learning_rate": 5.002427396881227e-08,
      "loss": 1.5374,
      "step": 6599
    },
    {
      "epoch": 0.9954000452454566,
      "grad_norm": 0.25277275561801404,
      "learning_rate": 5.0022733191327645e-08,
      "loss": 1.5392,
      "step": 6600
    },
    {
      "epoch": 0.9955508634341301,
      "grad_norm": 0.24256667203681012,
      "learning_rate": 5.0021242927791477e-08,
      "loss": 1.5,
      "step": 6601
    },
    {
      "epoch": 0.9957016816228037,
      "grad_norm": 0.262718000941228,
      "learning_rate": 5.001980317853838e-08,
      "loss": 1.4774,
      "step": 6602
    },
    {
      "epoch": 0.9958524998114773,
      "grad_norm": 0.26014642746601724,
      "learning_rate": 5.0018413943891636e-08,
      "loss": 1.5499,
      "step": 6603
    },
    {
      "epoch": 0.9960033180001509,
      "grad_norm": 0.27238693268092284,
      "learning_rate": 5.001707522416311e-08,
      "loss": 1.4962,
      "step": 6604
    },
    {
      "epoch": 0.9961541361888244,
      "grad_norm": 0.36523839539495767,
      "learning_rate": 5.001578701965347e-08,
      "loss": 1.5421,
      "step": 6605
    },
    {
      "epoch": 0.9963049543774979,
      "grad_norm": 0.26968890414008384,
      "learning_rate": 5.0014549330651884e-08,
      "loss": 1.5056,
      "step": 6606
    },
    {
      "epoch": 0.9964557725661715,
      "grad_norm": 0.25810388348729857,
      "learning_rate": 5.0013362157436276e-08,
      "loss": 1.4773,
      "step": 6607
    },
    {
      "epoch": 0.996606590754845,
      "grad_norm": 0.27029654290389415,
      "learning_rate": 5.0012225500273184e-08,
      "loss": 1.6149,
      "step": 6608
    },
    {
      "epoch": 0.9967574089435186,
      "grad_norm": 0.289186425325794,
      "learning_rate": 5.001113935941785e-08,
      "loss": 1.5884,
      "step": 6609
    },
    {
      "epoch": 0.9969082271321922,
      "grad_norm": 0.25404931837643585,
      "learning_rate": 5.001010373511414e-08,
      "loss": 1.4801,
      "step": 6610
    },
    {
      "epoch": 0.9970590453208656,
      "grad_norm": 0.2837220956254411,
      "learning_rate": 5.0009118627594556e-08,
      "loss": 1.5325,
      "step": 6611
    },
    {
      "epoch": 0.9972098635095392,
      "grad_norm": 0.2474702028118016,
      "learning_rate": 5.000818403708031e-08,
      "loss": 1.5786,
      "step": 6612
    },
    {
      "epoch": 0.9973606816982128,
      "grad_norm": 0.2526014041177881,
      "learning_rate": 5.000729996378121e-08,
      "loss": 1.5544,
      "step": 6613
    },
    {
      "epoch": 0.9975114998868864,
      "grad_norm": 0.26273147726823376,
      "learning_rate": 5.000646640789582e-08,
      "loss": 1.4948,
      "step": 6614
    },
    {
      "epoch": 0.9976623180755599,
      "grad_norm": 0.2950688429471966,
      "learning_rate": 5.000568336961123e-08,
      "loss": 1.5897,
      "step": 6615
    },
    {
      "epoch": 0.9978131362642335,
      "grad_norm": 0.25656955862427994,
      "learning_rate": 5.0004950849103275e-08,
      "loss": 1.5216,
      "step": 6616
    },
    {
      "epoch": 0.997963954452907,
      "grad_norm": 0.2862369388063159,
      "learning_rate": 5.000426884653646e-08,
      "loss": 1.5406,
      "step": 6617
    },
    {
      "epoch": 0.9981147726415805,
      "grad_norm": 0.34020761343391465,
      "learning_rate": 5.000363736206387e-08,
      "loss": 1.4978,
      "step": 6618
    },
    {
      "epoch": 0.9982655908302541,
      "grad_norm": 0.2857148184005516,
      "learning_rate": 5.00030563958273e-08,
      "loss": 1.5223,
      "step": 6619
    },
    {
      "epoch": 0.9984164090189277,
      "grad_norm": 0.30446226275025295,
      "learning_rate": 5.0002525947957226e-08,
      "loss": 1.5554,
      "step": 6620
    },
    {
      "epoch": 0.9985672272076013,
      "grad_norm": 0.2615054343779194,
      "learning_rate": 5.000204601857272e-08,
      "loss": 1.5789,
      "step": 6621
    },
    {
      "epoch": 0.9987180453962748,
      "grad_norm": 0.27178486454582657,
      "learning_rate": 5.0001616607781566e-08,
      "loss": 1.5491,
      "step": 6622
    },
    {
      "epoch": 0.9988688635849483,
      "grad_norm": 0.25279549756065184,
      "learning_rate": 5.000123771568012e-08,
      "loss": 1.5493,
      "step": 6623
    },
    {
      "epoch": 0.9990196817736219,
      "grad_norm": 0.45219351448552675,
      "learning_rate": 5.0000909342353545e-08,
      "loss": 1.5669,
      "step": 6624
    },
    {
      "epoch": 0.9991704999622955,
      "grad_norm": 0.273081492101918,
      "learning_rate": 5.0000631487875463e-08,
      "loss": 1.5694,
      "step": 6625
    },
    {
      "epoch": 0.999321318150969,
      "grad_norm": 0.2470353500890094,
      "learning_rate": 5.000040415230837e-08,
      "loss": 1.5355,
      "step": 6626
    },
    {
      "epoch": 0.9994721363396426,
      "grad_norm": 0.2749090211943406,
      "learning_rate": 5.000022733570323e-08,
      "loss": 1.5207,
      "step": 6627
    },
    {
      "epoch": 0.9996229545283161,
      "grad_norm": 0.2488845965843425,
      "learning_rate": 5.000010103809979e-08,
      "loss": 1.564,
      "step": 6628
    },
    {
      "epoch": 0.9997737727169896,
      "grad_norm": 0.24595496745924694,
      "learning_rate": 5.000002525952637e-08,
      "loss": 1.5456,
      "step": 6629
    },
    {
      "epoch": 0.9999245909056632,
      "grad_norm": 0.2667207051411267,
      "learning_rate": 5e-08,
      "loss": 1.5243,
      "step": 6630
    }
  ],
  "logging_steps": 1,
  "max_steps": 6630,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1326,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8649147427061760.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}