trainer_state.json

{
  "best_metric": 0.9109132885932922,
  "best_model_checkpoint": "bert-combined/checkpoint-1500",
  "epoch": 1.1029411764705883,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.36764705882352944,
      "grad_norm": 3.704710006713867,
      "learning_rate": 9.080882352941177e-06,
      "loss": 0.9954,
      "step": 500
    },
    {
      "epoch": 0.36764705882352944,
      "eval_accuracy": 0.5526893523600439,
      "eval_loss": 0.9458832144737244,
      "eval_runtime": 39.4256,
      "eval_samples_per_second": 92.427,
      "eval_steps_per_second": 2.892,
      "step": 500
    },
    {
      "epoch": 0.7352941176470589,
      "grad_norm": 5.2961835861206055,
      "learning_rate": 8.161764705882354e-06,
      "loss": 0.9644,
      "step": 1000
    },
    {
      "epoch": 0.7352941176470589,
      "eval_accuracy": 0.5760153677277717,
      "eval_loss": 0.9222443699836731,
      "eval_runtime": 39.4362,
      "eval_samples_per_second": 92.402,
      "eval_steps_per_second": 2.891,
      "step": 1000
    },
    {
      "epoch": 1.1029411764705883,
      "grad_norm": 7.299610614776611,
      "learning_rate": 7.24264705882353e-06,
      "loss": 0.951,
      "step": 1500
    },
    {
      "epoch": 1.1029411764705883,
      "eval_accuracy": 0.5927552140504939,
      "eval_loss": 0.9109132885932922,
      "eval_runtime": 40.7062,
      "eval_samples_per_second": 89.519,
      "eval_steps_per_second": 2.801,
      "step": 1500
    }
  ],
  "logging_steps": 500,
  "max_steps": 5440,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "total_flos": 1.1886933874968432e+16,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}