trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 3333,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.45,
      "learning_rate": 2.5499549954995502e-05,
      "loss": 3.7031,
      "step": 500
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.0999099909991e-05,
      "loss": 2.1276,
      "step": 1000
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.64986498649865e-05,
      "loss": 1.8363,
      "step": 1500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.1998199819981998e-05,
      "loss": 1.6986,
      "step": 2000
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.497749774977498e-06,
      "loss": 1.6041,
      "step": 2500
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.997299729972997e-06,
      "loss": 1.5504,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "step": 3333,
      "train_runtime": 5721.0485,
      "train_samples_per_second": 0.583
    }
  ],
  "max_steps": 3333,
  "num_train_epochs": 3,
  "total_flos": 633709039885608960,
  "trial_name": null,
  "trial_params": null
}