trainer_state.json

{
  "best_global_step": 20,
  "best_metric": 0.395206093788147,
  "best_model_checkpoint": "/content/models/gemma_qlora_lmh/checkpoint-20",
  "epoch": 1.25,
  "eval_steps": 20,
  "global_step": 20,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 1.949600937962532,
      "epoch": 1.25,
      "grad_norm": 3.921875,
      "learning_rate": 4.0625000000000005e-06,
      "loss": 0.5369,
      "mean_token_accuracy": 0.8124339818954468,
      "num_tokens": 58731.0,
      "step": 20
    },
    {
      "epoch": 1.25,
      "eval_entropy": 1.9308812655508518,
      "eval_loss": 0.395206093788147,
      "eval_mean_token_accuracy": 0.8587497491389513,
      "eval_num_tokens": 58731.0,
      "eval_runtime": 5.9613,
      "eval_samples_per_second": 83.874,
      "eval_steps_per_second": 5.368,
      "step": 20
    }
  ],
  "logging_steps": 20,
  "max_steps": 32,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 20,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 926435344545792.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}