trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.380952380952381,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.79,
      "learning_rate": 1.470899470899471e-05,
      "loss": 1.2451,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.8770800232887268,
      "eval_runtime": 1453.2483,
      "eval_samples_per_second": 3.453,
      "eval_steps_per_second": 0.432,
      "step": 630
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.417989417989418e-06,
      "loss": 0.9493,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.7568654417991638,
      "eval_runtime": 1504.7075,
      "eval_samples_per_second": 3.335,
      "eval_steps_per_second": 0.417,
      "step": 1260
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.126984126984127e-06,
      "loss": 0.8542,
      "step": 1500
    }
  ],
  "max_steps": 1890,
  "num_train_epochs": 3,
  "total_flos": 1553384715102000.0,
  "trial_name": null,
  "trial_params": null
}