trainer_state.json

{
  "best_metric": 0.9972401103955841,
  "best_model_checkpoint": "data/train-test/xlnet-large-cased//model/checkpoint-3352",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 3352,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.3,
      "grad_norm": 0.6528932452201843,
      "learning_rate": 2.552505966587112e-05,
      "loss": 0.0674,
      "step": 500
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.035294823348522186,
      "learning_rate": 2.1050119331742245e-05,
      "loss": 0.0058,
      "step": 1000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.07718423008918762,
      "learning_rate": 1.6575178997613365e-05,
      "loss": 0.0041,
      "step": 1500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": null,
      "eval_f1": 0.9853211009174311,
      "eval_loss": 0.00522511824965477,
      "eval_precision": 0.9826166514181153,
      "eval_recall": 0.9880404783808647,
      "eval_runtime": 8.7222,
      "eval_samples_per_second": 111.211,
      "eval_steps_per_second": 6.994,
      "step": 1676
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.03843804448843002,
      "learning_rate": 1.2100238663484488e-05,
      "loss": 0.003,
      "step": 2000
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.008623383939266205,
      "learning_rate": 7.625298329355609e-06,
      "loss": 0.0015,
      "step": 2500
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.005918263457715511,
      "learning_rate": 3.1503579952267305e-06,
      "loss": 0.0012,
      "step": 3000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": null,
      "eval_f1": 0.9972401103955841,
      "eval_loss": 0.002058240817859769,
      "eval_precision": 0.9972401103955841,
      "eval_recall": 0.9972401103955841,
      "eval_runtime": 8.7323,
      "eval_samples_per_second": 111.082,
      "eval_steps_per_second": 6.986,
      "step": 3352
    }
  ],
  "logging_steps": 500,
  "max_steps": 3352,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 6566441558618250.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}