trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.7862484652304946,
  "eval_steps": 500,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.22,
      "learning_rate": 4.375e-05,
      "loss": 6.9178,
      "step": 500
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 5.889,
      "step": 1000
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.125e-05,
      "loss": 5.4496,
      "step": 1500
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.5e-05,
      "loss": 5.187,
      "step": 2000
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 5.008,
      "step": 2500
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.25e-05,
      "loss": 4.8763,
      "step": 3000
    },
    {
      "epoch": 1.56,
      "learning_rate": 6.25e-06,
      "loss": 4.7881,
      "step": 3500
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0,
      "loss": 4.7419,
      "step": 4000
    },
    {
      "epoch": 1.79,
      "step": 4000,
      "total_flos": 2.67561995599872e+17,
      "train_loss": 0.59273583984375,
      "train_runtime": 6564.3415,
      "train_samples_per_second": 77.997,
      "train_steps_per_second": 0.609
    }
  ],
  "logging_steps": 500,
  "max_steps": 4000,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 2.67561995599872e+17,
  "trial_name": null,
  "trial_params": null
}