checkpoint-1000/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.3222591362126246,
  "eval_steps": 1000,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 9.966777408637875e-07,
      "loss": 2.3466,
      "step": 300
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.993355481727575e-06,
      "loss": 2.3107,
      "step": 600
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.9900332225913626e-06,
      "loss": 2.2553,
      "step": 900
    },
    {
      "epoch": 3.32,
      "eval_loss": 2.210308074951172,
      "eval_runtime": 21.2106,
      "eval_samples_per_second": 80.007,
      "eval_steps_per_second": 5.045,
      "step": 1000
    }
  ],
  "logging_steps": 300,
  "max_steps": 6020,
  "num_train_epochs": 20,
  "save_steps": 1000,
  "total_flos": 5.172190049437286e+16,
  "trial_name": null,
  "trial_params": null
}