trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.842105263157895,
  "eval_steps": 500,
  "global_step": 20,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 8.293423652648926,
      "learning_rate": 0.0,
      "loss": 1.0907,
      "step": 1
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 8.356363296508789,
      "learning_rate": 1e-05,
      "loss": 1.1208,
      "step": 2
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 8.640625953674316,
      "learning_rate": 1e-05,
      "loss": 1.1098,
      "step": 3
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 5.440302848815918,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 4
    },
    {
      "epoch": 1.2105263157894737,
      "grad_norm": 6.2790846824646,
      "learning_rate": 1e-05,
      "loss": 1.1493,
      "step": 5
    },
    {
      "epoch": 1.4210526315789473,
      "grad_norm": 6.609096050262451,
      "learning_rate": 1e-05,
      "loss": 0.8753,
      "step": 6
    },
    {
      "epoch": 1.631578947368421,
      "grad_norm": 12.102945327758789,
      "learning_rate": 1e-05,
      "loss": 0.9279,
      "step": 7
    },
    {
      "epoch": 1.8421052631578947,
      "grad_norm": 7.788964748382568,
      "learning_rate": 1e-05,
      "loss": 0.8725,
      "step": 8
    },
    {
      "epoch": 2.2105263157894735,
      "grad_norm": 11.209437370300293,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 9
    },
    {
      "epoch": 2.4210526315789473,
      "grad_norm": 6.773443222045898,
      "learning_rate": 1e-05,
      "loss": 0.8019,
      "step": 10
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 6.824408054351807,
      "learning_rate": 1e-05,
      "loss": 0.7652,
      "step": 11
    },
    {
      "epoch": 2.8421052631578947,
      "grad_norm": 6.099692344665527,
      "learning_rate": 1e-05,
      "loss": 0.7196,
      "step": 12
    },
    {
      "epoch": 3.2105263157894735,
      "grad_norm": 9.494585990905762,
      "learning_rate": 1e-05,
      "loss": 0.8614,
      "step": 13
    },
    {
      "epoch": 3.4210526315789473,
      "grad_norm": 5.957844257354736,
      "learning_rate": 1e-05,
      "loss": 0.6315,
      "step": 14
    },
    {
      "epoch": 3.6315789473684212,
      "grad_norm": 4.620523929595947,
      "learning_rate": 1e-05,
      "loss": 0.6296,
      "step": 15
    },
    {
      "epoch": 3.8421052631578947,
      "grad_norm": 4.814481735229492,
      "learning_rate": 1e-05,
      "loss": 0.61,
      "step": 16
    },
    {
      "epoch": 4.2105263157894735,
      "grad_norm": 4.219420433044434,
      "learning_rate": 1e-05,
      "loss": 0.7063,
      "step": 17
    },
    {
      "epoch": 4.421052631578947,
      "grad_norm": 4.6533026695251465,
      "learning_rate": 1e-05,
      "loss": 0.5283,
      "step": 18
    },
    {
      "epoch": 4.631578947368421,
      "grad_norm": 11.828014373779297,
      "learning_rate": 1e-05,
      "loss": 0.4826,
      "step": 19
    },
    {
      "epoch": 4.842105263157895,
      "grad_norm": 5.108757972717285,
      "learning_rate": 1e-05,
      "loss": 0.5179,
      "step": 20
    },
    {
      "epoch": 4.842105263157895,
      "step": 20,
      "total_flos": 93712012541952.0,
      "train_loss": 0.8187338769435882,
      "train_runtime": 200.6985,
      "train_samples_per_second": 7.499,
      "train_steps_per_second": 0.1
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 20,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 93712012541952.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}