checkpoint-4/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8,
  "eval_steps": 500,
  "global_step": 4,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2,
      "grad_norm": 1.2988848686218262,
      "learning_rate": 9.333333333333334e-05,
      "loss": 1.4722,
      "step": 1
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0152840614318848,
      "learning_rate": 8.666666666666667e-05,
      "loss": 1.4304,
      "step": 2
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8597587943077087,
      "learning_rate": 8e-05,
      "loss": 1.3624,
      "step": 3
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8003940582275391,
      "learning_rate": 7.333333333333333e-05,
      "loss": 1.3047,
      "step": 4
    }
  ],
  "logging_steps": 1,
  "max_steps": 15,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.1369028085743616e+16,
  "train_batch_size": 100,
  "trial_name": null,
  "trial_params": null
}