trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.938547486033519,
  "eval_steps": 100,
  "global_step": 400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.12,
      "learning_rate": 2.272727272727273e-05,
      "loss": 0.4667,
      "step": 50
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.0202020202020203e-05,
      "loss": 0.2407,
      "step": 100
    },
    {
      "epoch": 2.23,
      "eval_loss": 0.18808360397815704,
      "eval_runtime": 19.8928,
      "eval_samples_per_second": 1.91,
      "eval_steps_per_second": 0.251,
      "step": 100
    },
    {
      "epoch": 3.35,
      "learning_rate": 1.7676767676767676e-05,
      "loss": 0.1581,
      "step": 150
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 0.1116,
      "step": 200
    },
    {
      "epoch": 4.47,
      "eval_loss": 0.08594753593206406,
      "eval_runtime": 19.8891,
      "eval_samples_per_second": 1.911,
      "eval_steps_per_second": 0.251,
      "step": 200
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.2626262626262628e-05,
      "loss": 0.0846,
      "step": 250
    },
    {
      "epoch": 6.7,
      "learning_rate": 1.0101010101010101e-05,
      "loss": 0.0708,
      "step": 300
    },
    {
      "epoch": 6.7,
      "eval_loss": 0.06138959154486656,
      "eval_runtime": 19.8913,
      "eval_samples_per_second": 1.91,
      "eval_steps_per_second": 0.251,
      "step": 300
    },
    {
      "epoch": 7.82,
      "learning_rate": 7.5757575757575764e-06,
      "loss": 0.0641,
      "step": 350
    },
    {
      "epoch": 8.94,
      "learning_rate": 5.050505050505051e-06,
      "loss": 0.0547,
      "step": 400
    },
    {
      "epoch": 8.94,
      "eval_loss": 0.05079507827758789,
      "eval_runtime": 19.894,
      "eval_samples_per_second": 1.91,
      "eval_steps_per_second": 0.251,
      "step": 400
    }
  ],
  "logging_steps": 50,
  "max_steps": 500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 12,
  "save_steps": 50,
  "total_flos": 5.228947662962688e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}