trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 2250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5e-05,
      "loss": 11.1,
      "step": 1
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.8933333333333336e-05,
      "loss": 0.6822,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_cer": 0.0418,
      "eval_loss": 0.26874810457229614,
      "eval_runtime": 341.3378,
      "eval_samples_per_second": 5.859,
      "eval_steps_per_second": 0.732,
      "step": 750
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.782222222222222e-05,
      "loss": 0.2781,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6733333333333335e-05,
      "loss": 0.1595,
      "step": 1500
    },
    {
      "epoch": 2.0,
      "eval_cer": 0.0094,
      "eval_loss": 0.14129523932933807,
      "eval_runtime": 339.7395,
      "eval_samples_per_second": 5.887,
      "eval_steps_per_second": 0.736,
      "step": 1500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.6444444444444445e-06,
      "loss": 0.0481,
      "step": 2000
    },
    {
      "epoch": 3.0,
      "eval_cer": 0.0029,
      "eval_loss": 0.07172464579343796,
      "eval_runtime": 338.5668,
      "eval_samples_per_second": 5.907,
      "eval_steps_per_second": 0.738,
      "step": 2250
    },
    {
      "epoch": 3.0,
      "step": 2250,
      "total_flos": 1.3469133437927424e+19,
      "train_loss": 0.2679765387641059,
      "train_runtime": 3403.503,
      "train_samples_per_second": 5.289,
      "train_steps_per_second": 0.661
    }
  ],
  "logging_steps": 500,
  "max_steps": 2250,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.3469133437927424e+19,
  "trial_name": null,
  "trial_params": null
}