trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 3339,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.45,
      "grad_norm": 5.29541015625,
      "learning_rate": 4.251272836178497e-05,
      "loss": 3.2051,
      "step": 500
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.331775188446045,
      "learning_rate": 3.502545672356993e-05,
      "loss": 2.8264,
      "step": 1000
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.496217250823975,
      "learning_rate": 2.75381850853549e-05,
      "loss": 2.6611,
      "step": 1500
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.267590522766113,
      "learning_rate": 2.0050913447139864e-05,
      "loss": 2.5819,
      "step": 2000
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.9951887130737305,
      "learning_rate": 1.2563641808924828e-05,
      "loss": 2.5289,
      "step": 2500
    },
    {
      "epoch": 2.7,
      "grad_norm": 5.234454154968262,
      "learning_rate": 5.0763701707097935e-06,
      "loss": 2.5066,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "step": 3339,
      "total_flos": 480295528237080.0,
      "train_loss": 2.6927158644757037,
      "train_runtime": 1168.1664,
      "train_samples_per_second": 91.459,
      "train_steps_per_second": 2.858
    }
  ],
  "logging_steps": 500,
  "max_steps": 3339,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 480295528237080.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}