trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.4182646870613098,
  "best_model_checkpoint": "tiny_bert_rand_10_v2_qqp/checkpoint-2844",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 9954,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 1.803529977798462,
      "learning_rate": 4.9e-05,
      "loss": 0.4943,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7848627256987386,
      "eval_combined_score": 0.7281569804364358,
      "eval_f1": 0.6714512351741331,
      "eval_loss": 0.4519537091255188,
      "eval_runtime": 11.8129,
      "eval_samples_per_second": 3422.525,
      "eval_steps_per_second": 13.375,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.601994752883911,
      "learning_rate": 4.8e-05,
      "loss": 0.4048,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8032401681919367,
      "eval_combined_score": 0.7630360626333015,
      "eval_f1": 0.7228319570746664,
      "eval_loss": 0.4182646870613098,
      "eval_runtime": 11.8478,
      "eval_samples_per_second": 3412.46,
      "eval_steps_per_second": 13.336,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.3220903873443604,
      "learning_rate": 4.7e-05,
      "loss": 0.3482,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8168686618847391,
      "eval_combined_score": 0.7782930105458288,
      "eval_f1": 0.7397173592069184,
      "eval_loss": 0.4183485507965088,
      "eval_runtime": 11.4704,
      "eval_samples_per_second": 3524.719,
      "eval_steps_per_second": 13.775,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.0365071296691895,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3031,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8224833044768736,
      "eval_combined_score": 0.7854838109908462,
      "eval_f1": 0.7484843175048187,
      "eval_loss": 0.4340476095676422,
      "eval_runtime": 11.5172,
      "eval_samples_per_second": 3510.413,
      "eval_steps_per_second": 13.719,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.2530550956726074,
      "learning_rate": 4.5e-05,
      "loss": 0.2648,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8270096463022508,
      "eval_combined_score": 0.7906478810488207,
      "eval_f1": 0.7542861157953906,
      "eval_loss": 0.45441848039627075,
      "eval_runtime": 11.5431,
      "eval_samples_per_second": 3502.526,
      "eval_steps_per_second": 13.688,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.48345947265625,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.234,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8286668315607222,
      "eval_combined_score": 0.7960081108710653,
      "eval_f1": 0.7633493901814082,
      "eval_loss": 0.4517841041088104,
      "eval_runtime": 11.5196,
      "eval_samples_per_second": 3509.675,
      "eval_steps_per_second": 13.716,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 5.132691860198975,
      "learning_rate": 4.3e-05,
      "loss": 0.2093,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8285431610190452,
      "eval_combined_score": 0.7997612065171875,
      "eval_f1": 0.7709792520153297,
      "eval_loss": 0.48305028676986694,
      "eval_runtime": 12.0865,
      "eval_samples_per_second": 3345.064,
      "eval_steps_per_second": 13.072,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 6.678940801217126e+16,
      "train_loss": 0.32263902844687126,
      "train_runtime": 1410.2891,
      "train_samples_per_second": 12899.695,
      "train_steps_per_second": 50.415
    }
  ],
  "logging_steps": 1,
  "max_steps": 71100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 5
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.678940801217126e+16,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}