trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 3300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.45,
      "grad_norm": 4.830384254455566,
      "learning_rate": 4.242424242424243e-05,
      "loss": 0.5576,
      "step": 500
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.6578476428985596,
      "learning_rate": 3.484848484848485e-05,
      "loss": 0.4035,
      "step": 1000
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.386310577392578,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.2828,
      "step": 1500
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7881163358688354,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 0.2468,
      "step": 2000
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0737249851226807,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 0.1608,
      "step": 2500
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.20467181503772736,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.1321,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "step": 3300,
      "total_flos": 2871828253461180.0,
      "train_loss": 0.2811542418508819,
      "train_runtime": 449.7464,
      "train_samples_per_second": 117.386,
      "train_steps_per_second": 7.337
    }
  ],
  "logging_steps": 500,
  "max_steps": 3300,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2871828253461180.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}