trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 3210,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.78,
      "learning_rate": 4.221183800623053e-05,
      "loss": 2.765,
      "step": 500
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.442367601246106e-05,
      "loss": 2.5475,
      "step": 1000
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.663551401869159e-05,
      "loss": 2.403,
      "step": 1500
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.884735202492212e-05,
      "loss": 2.3577,
      "step": 2000
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.1059190031152649e-05,
      "loss": 2.3018,
      "step": 2500
    },
    {
      "epoch": 4.67,
      "learning_rate": 3.2710280373831774e-06,
      "loss": 2.2848,
      "step": 3000
    },
    {
      "epoch": 5.0,
      "step": 3210,
      "total_flos": 6704753541120000.0,
      "train_loss": 2.4309686013099934,
      "train_runtime": 1406.649,
      "train_samples_per_second": 9.121,
      "train_steps_per_second": 2.282
    }
  ],
  "logging_steps": 500,
  "max_steps": 3210,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 6704753541120000.0,
  "trial_name": null,
  "trial_params": null
}