trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.0012697727424234248,
  "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": 3e-05,
      "loss": 1.3736,
      "step": 50
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.0156368017196655,
      "eval_runtime": 6181.8269,
      "eval_samples_per_second": 5.662,
      "eval_steps_per_second": 0.708,
      "step": 50
    },
    {
      "epoch": 0.0,
      "learning_rate": 3e-05,
      "loss": 0.9336,
      "step": 100
    },
    {
      "epoch": 0.0,
      "eval_loss": 0.9014769196510315,
      "eval_runtime": 6086.4963,
      "eval_samples_per_second": 5.751,
      "eval_steps_per_second": 0.719,
      "step": 100
    }
  ],
  "logging_steps": 50,
  "max_steps": 1000000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 13,
  "save_steps": 50,
  "total_flos": 8522209045045248.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}