trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 3468,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.43,
      "learning_rate": 4.2791234140715114e-05,
      "loss": 0.725,
      "step": 500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.558246828143022e-05,
      "loss": 0.3879,
      "step": 1000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8373702422145332e-05,
      "loss": 0.2897,
      "step": 1500
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.116493656286044e-05,
      "loss": 0.2463,
      "step": 2000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.395617070357555e-05,
      "loss": 0.2074,
      "step": 2500
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.747404844290659e-06,
      "loss": 0.1688,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "step": 3468,
      "total_flos": 2.370754172808069e+17,
      "train_loss": 0.31381386621600615,
      "train_runtime": 2692.8552,
      "train_samples_per_second": 659.248,
      "train_steps_per_second": 1.288
    }
  ],
  "logging_steps": 500,
  "max_steps": 3468,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2.370754172808069e+17,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}