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": 4.369685173034668,
      "learning_rate": 4.242424242424243e-05,
      "loss": 0.4661,
      "step": 500
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.9891207218170166,
      "learning_rate": 3.484848484848485e-05,
      "loss": 0.3327,
      "step": 1000
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.819967269897461,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.2258,
      "step": 1500
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7621924877166748,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 0.1772,
      "step": 2000
    },
    {
      "epoch": 2.27,
      "grad_norm": 21.205034255981445,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 0.1106,
      "step": 2500
    },
    {
      "epoch": 2.73,
      "grad_norm": 8.332528114318848,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.0867,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "step": 3300,
      "total_flos": 2669084030031120.0,
      "train_loss": 0.21816863666881214,
      "train_runtime": 391.9097,
      "train_samples_per_second": 134.71,
      "train_steps_per_second": 8.42
    }
  ],
  "logging_steps": 500,
  "max_steps": 3300,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2669084030031120.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}