trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 100,
  "global_step": 64,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.62,
      "grad_norm": 1.1892509460449219,
      "learning_rate": 0.00016875,
      "loss": 0.5447,
      "step": 10
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.315897822380066,
      "learning_rate": 0.0001375,
      "loss": 0.408,
      "step": 20
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7632724046707153,
      "learning_rate": 0.00010625000000000001,
      "loss": 0.2444,
      "step": 30
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.5031312704086304,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.1791,
      "step": 40
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6666581630706787,
      "learning_rate": 4.375e-05,
      "loss": 0.1459,
      "step": 50
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.8803956508636475,
      "learning_rate": 1.25e-05,
      "loss": 0.1187,
      "step": 60
    },
    {
      "epoch": 4.0,
      "step": 64,
      "total_flos": 7.749476775936e+16,
      "train_loss": 0.2629110822454095,
      "train_runtime": 64.8307,
      "train_samples_per_second": 15.425,
      "train_steps_per_second": 0.987
    }
  ],
  "logging_steps": 10,
  "max_steps": 64,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 100,
  "total_flos": 7.749476775936e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}