trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "eval_steps": 500,
  "global_step": 1920,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.08,
      "learning_rate": 4.4791666666666673e-05,
      "loss": 6.9421,
      "step": 200
    },
    {
      "epoch": 4.17,
      "learning_rate": 3.958333333333333e-05,
      "loss": 4.8424,
      "step": 400
    },
    {
      "epoch": 6.25,
      "learning_rate": 3.4375e-05,
      "loss": 4.0383,
      "step": 600
    },
    {
      "epoch": 8.33,
      "learning_rate": 2.916666666666667e-05,
      "loss": 3.4406,
      "step": 800
    },
    {
      "epoch": 10.42,
      "learning_rate": 2.3958333333333334e-05,
      "loss": 2.9599,
      "step": 1000
    },
    {
      "epoch": 12.5,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 2.5537,
      "step": 1200
    },
    {
      "epoch": 14.58,
      "learning_rate": 1.3541666666666666e-05,
      "loss": 2.2096,
      "step": 1400
    },
    {
      "epoch": 16.67,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.944,
      "step": 1600
    },
    {
      "epoch": 18.75,
      "learning_rate": 3.125e-06,
      "loss": 1.7395,
      "step": 1800
    },
    {
      "epoch": 20.0,
      "step": 1920,
      "total_flos": 3.670508706988032e+16,
      "train_loss": 3.2965724309285482,
      "train_runtime": 9164.1975,
      "train_samples_per_second": 1.674,
      "train_steps_per_second": 0.21
    }
  ],
  "logging_steps": 200,
  "max_steps": 1920,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 200,
  "total_flos": 3.670508706988032e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}