trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5872551798820496,
  "best_model_checkpoint": "bert_base_km_100_v1_mrpc/checkpoint-30",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 105,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 1.2941228151321411,
      "learning_rate": 4.9e-05,
      "loss": 0.6199,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6887254901960784,
      "eval_combined_score": 0.7414454031207296,
      "eval_f1": 0.7941653160453809,
      "eval_loss": 0.6157417893409729,
      "eval_runtime": 0.2728,
      "eval_samples_per_second": 1495.805,
      "eval_steps_per_second": 7.332,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5449658632278442,
      "learning_rate": 4.8e-05,
      "loss": 0.5544,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7107843137254902,
      "eval_combined_score": 0.7626248612652609,
      "eval_f1": 0.8144654088050315,
      "eval_loss": 0.5872551798820496,
      "eval_runtime": 0.2743,
      "eval_samples_per_second": 1487.298,
      "eval_steps_per_second": 7.291,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8436791896820068,
      "learning_rate": 4.7e-05,
      "loss": 0.4708,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7083333333333334,
      "eval_combined_score": 0.7586610486891385,
      "eval_f1": 0.8089887640449438,
      "eval_loss": 0.5949360728263855,
      "eval_runtime": 0.306,
      "eval_samples_per_second": 1333.505,
      "eval_steps_per_second": 6.537,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.9913930892944336,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3405,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7181372549019608,
      "eval_combined_score": 0.7614455374679583,
      "eval_f1": 0.8047538200339559,
      "eval_loss": 0.6693371534347534,
      "eval_runtime": 0.2718,
      "eval_samples_per_second": 1501.318,
      "eval_steps_per_second": 7.359,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.109549045562744,
      "learning_rate": 4.5e-05,
      "loss": 0.1938,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6838235294117647,
      "eval_combined_score": 0.7204428381522101,
      "eval_f1": 0.7570621468926554,
      "eval_loss": 0.7770265340805054,
      "eval_runtime": 0.2877,
      "eval_samples_per_second": 1417.936,
      "eval_steps_per_second": 6.951,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.1045870780944824,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.0851,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7107843137254902,
      "eval_combined_score": 0.7540175864160097,
      "eval_f1": 0.7972508591065293,
      "eval_loss": 0.9557104110717773,
      "eval_runtime": 0.2733,
      "eval_samples_per_second": 1492.865,
      "eval_steps_per_second": 7.318,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 1.8864613771438599,
      "learning_rate": 4.3e-05,
      "loss": 0.0382,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.678921568627451,
      "eval_combined_score": 0.7267241233842935,
      "eval_f1": 0.774526678141136,
      "eval_loss": 1.2726539373397827,
      "eval_runtime": 0.2707,
      "eval_samples_per_second": 1507.419,
      "eval_steps_per_second": 7.389,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 3377819728711680.0,
      "train_loss": 0.32894962344850814,
      "train_runtime": 58.9337,
      "train_samples_per_second": 3111.971,
      "train_steps_per_second": 12.726
    }
  ],
  "logging_steps": 1,
  "max_steps": 750,
  "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": 3377819728711680.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}