trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.3920540511608124,
  "best_model_checkpoint": "bert_base_km_20_v1_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": 2.017176866531372,
      "learning_rate": 4.9e-05,
      "loss": 0.4811,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7807321296067277,
      "eval_combined_score": 0.7114269828123259,
      "eval_f1": 0.6421218360179242,
      "eval_loss": 0.4579184949398041,
      "eval_runtime": 26.0123,
      "eval_samples_per_second": 1554.263,
      "eval_steps_per_second": 6.074,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.95531964302063,
      "learning_rate": 4.8e-05,
      "loss": 0.3726,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8163739797180312,
      "eval_combined_score": 0.7867925232490345,
      "eval_f1": 0.7572110667800379,
      "eval_loss": 0.3920540511608124,
      "eval_runtime": 25.2701,
      "eval_samples_per_second": 1599.916,
      "eval_steps_per_second": 6.252,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.366745710372925,
      "learning_rate": 4.7e-05,
      "loss": 0.2901,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8243630967103636,
      "eval_combined_score": 0.7945958553596527,
      "eval_f1": 0.7648286140089419,
      "eval_loss": 0.3958491384983063,
      "eval_runtime": 25.3413,
      "eval_samples_per_second": 1595.42,
      "eval_steps_per_second": 6.235,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.1108007431030273,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.2176,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8303240168191937,
      "eval_combined_score": 0.7973735468711353,
      "eval_f1": 0.7644230769230769,
      "eval_loss": 0.44668564200401306,
      "eval_runtime": 25.356,
      "eval_samples_per_second": 1594.496,
      "eval_steps_per_second": 6.231,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.534144639968872,
      "learning_rate": 4.5e-05,
      "loss": 0.1641,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8321543408360128,
      "eval_combined_score": 0.798826496563559,
      "eval_f1": 0.7654986522911051,
      "eval_loss": 0.4786772131919861,
      "eval_runtime": 25.3806,
      "eval_samples_per_second": 1592.949,
      "eval_steps_per_second": 6.225,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 4.56919002532959,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.1263,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8334899826861242,
      "eval_combined_score": 0.8009475508586806,
      "eval_f1": 0.7684051190312371,
      "eval_loss": 0.5397569537162781,
      "eval_runtime": 25.3607,
      "eval_samples_per_second": 1594.201,
      "eval_steps_per_second": 6.23,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 6.266264915466309,
      "learning_rate": 4.3e-05,
      "loss": 0.0993,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8356418501113035,
      "eval_combined_score": 0.8081493620914663,
      "eval_f1": 0.780656874071629,
      "eval_loss": 0.5794485807418823,
      "eval_runtime": 25.3072,
      "eval_samples_per_second": 1597.566,
      "eval_steps_per_second": 6.243,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 3.3506166766980096e+17,
      "train_loss": 0.2501540595037956,
      "train_runtime": 4264.0449,
      "train_samples_per_second": 4266.442,
      "train_steps_per_second": 16.674
    }
  ],
  "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": 3.3506166766980096e+17,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}