trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5978386998176575,
  "best_model_checkpoint": "bert_base_km_5_v2_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": 1.4363431930541992,
      "learning_rate": 4.9e-05,
      "loss": 0.6316,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6887254901960784,
      "eval_combined_score": 0.7424365813741227,
      "eval_f1": 0.7961476725521669,
      "eval_loss": 0.6041475534439087,
      "eval_runtime": 0.3074,
      "eval_samples_per_second": 1327.06,
      "eval_steps_per_second": 6.505,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8027812242507935,
      "learning_rate": 4.8e-05,
      "loss": 0.5501,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7156862745098039,
      "eval_combined_score": 0.7605277010132911,
      "eval_f1": 0.8053691275167785,
      "eval_loss": 0.5978386998176575,
      "eval_runtime": 0.2786,
      "eval_samples_per_second": 1464.265,
      "eval_steps_per_second": 7.178,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.0269715785980225,
      "learning_rate": 4.7e-05,
      "loss": 0.4475,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6642156862745098,
      "eval_combined_score": 0.7129774083546462,
      "eval_f1": 0.7617391304347826,
      "eval_loss": 0.6494813561439514,
      "eval_runtime": 0.2962,
      "eval_samples_per_second": 1377.329,
      "eval_steps_per_second": 6.752,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.375882625579834,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3135,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6715686274509803,
      "eval_combined_score": 0.7198673587081892,
      "eval_f1": 0.7681660899653979,
      "eval_loss": 0.8098564743995667,
      "eval_runtime": 0.2668,
      "eval_samples_per_second": 1529.24,
      "eval_steps_per_second": 7.496,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.542888164520264,
      "learning_rate": 4.5e-05,
      "loss": 0.1742,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.5882352941176471,
      "eval_combined_score": 0.6300551470588236,
      "eval_f1": 0.671875,
      "eval_loss": 1.151023268699646,
      "eval_runtime": 0.275,
      "eval_samples_per_second": 1483.816,
      "eval_steps_per_second": 7.274,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 4.128825664520264,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.0858,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6274509803921569,
      "eval_combined_score": 0.6735040879820563,
      "eval_f1": 0.7195571955719557,
      "eval_loss": 1.1825387477874756,
      "eval_runtime": 0.3035,
      "eval_samples_per_second": 1344.189,
      "eval_steps_per_second": 6.589,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 2.8958871364593506,
      "learning_rate": 4.3e-05,
      "loss": 0.0492,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6642156862745098,
      "eval_combined_score": 0.7181311376464063,
      "eval_f1": 0.7720465890183028,
      "eval_loss": 1.3776967525482178,
      "eval_runtime": 0.3059,
      "eval_samples_per_second": 1333.975,
      "eval_steps_per_second": 6.539,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 3377819728711680.0,
      "train_loss": 0.32170329434531075,
      "train_runtime": 58.7394,
      "train_samples_per_second": 3122.266,
      "train_steps_per_second": 12.768
    }
  ],
  "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
}