trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.4247690737247467,
  "best_model_checkpoint": "tiny_bert_rand_20_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": 1.8136016130447388,
      "learning_rate": 4.9e-05,
      "loss": 0.4957,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7811526094484293,
      "eval_combined_score": 0.7260174593143054,
      "eval_f1": 0.6708823091801815,
      "eval_loss": 0.4586934447288513,
      "eval_runtime": 12.5393,
      "eval_samples_per_second": 3224.252,
      "eval_steps_per_second": 12.6,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.50662899017334,
      "learning_rate": 4.8e-05,
      "loss": 0.4075,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7999010635666585,
      "eval_combined_score": 0.7540059510357466,
      "eval_f1": 0.7081108385048348,
      "eval_loss": 0.4247690737247467,
      "eval_runtime": 11.4753,
      "eval_samples_per_second": 3523.222,
      "eval_steps_per_second": 13.769,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.9053895473480225,
      "learning_rate": 4.7e-05,
      "loss": 0.3493,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8118723720009894,
      "eval_combined_score": 0.7714405006135941,
      "eval_f1": 0.7310086292261989,
      "eval_loss": 0.4262579083442688,
      "eval_runtime": 11.5633,
      "eval_samples_per_second": 3496.395,
      "eval_steps_per_second": 13.664,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.459416151046753,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3035,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8147909967845659,
      "eval_combined_score": 0.7715858640336062,
      "eval_f1": 0.7283807312826466,
      "eval_loss": 0.4371027946472168,
      "eval_runtime": 11.5128,
      "eval_samples_per_second": 3511.732,
      "eval_steps_per_second": 13.724,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.1791892051696777,
      "learning_rate": 4.5e-05,
      "loss": 0.2653,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8201088300766758,
      "eval_combined_score": 0.7845669931121155,
      "eval_f1": 0.7490251561475552,
      "eval_loss": 0.4532000720500946,
      "eval_runtime": 11.4708,
      "eval_samples_per_second": 3524.604,
      "eval_steps_per_second": 13.774,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.570615768432617,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.2333,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.822508038585209,
      "eval_combined_score": 0.7889636307445473,
      "eval_f1": 0.7554192229038855,
      "eval_loss": 0.45731690526008606,
      "eval_runtime": 11.4996,
      "eval_samples_per_second": 3515.775,
      "eval_steps_per_second": 13.74,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 5.915837287902832,
      "learning_rate": 4.3e-05,
      "loss": 0.2073,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8237447440019787,
      "eval_combined_score": 0.7946761349113243,
      "eval_f1": 0.7656075258206697,
      "eval_loss": 0.4748603403568268,
      "eval_runtime": 11.4587,
      "eval_samples_per_second": 3528.314,
      "eval_steps_per_second": 13.789,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 6.678940801217126e+16,
      "train_loss": 0.32312122089737794,
      "train_runtime": 1408.2712,
      "train_samples_per_second": 12918.179,
      "train_steps_per_second": 50.487
    }
  ],
  "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
}