checkpoint-100/trainer_state.json

{
  "best_metric": 0.49068325757980347,
  "best_model_checkpoint": "output/checkpoint-100",
  "epoch": 0.08021658477890303,
  "eval_steps": 50,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0,
      "loss": 1.6453,
      "step": 1
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.777777777777779e-06,
      "loss": 1.3216,
      "step": 50
    },
    {
      "epoch": 0.04,
      "eval_accuracy": 0.8057142857142857,
      "eval_loss": 0.6264824271202087,
      "eval_runtime": 28.8159,
      "eval_samples_per_second": 6.073,
      "eval_steps_per_second": 1.527,
      "step": 50
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.088888888888889e-05,
      "loss": 0.918,
      "step": 100
    },
    {
      "epoch": 0.08,
      "eval_accuracy": 0.7885714285714286,
      "eval_loss": 0.49068325757980347,
      "eval_runtime": 26.1642,
      "eval_samples_per_second": 6.689,
      "eval_steps_per_second": 1.682,
      "step": 100
    }
  ],
  "logging_steps": 50,
  "max_steps": 1246,
  "num_train_epochs": 1,
  "save_steps": 50,
  "total_flos": 3.690320903467008e+16,
  "trial_name": null,
  "trial_params": null
}