trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5948534607887268,
  "best_model_checkpoint": "bert_base_km_50_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": 1.3010720014572144,
      "learning_rate": 4.9e-05,
      "loss": 0.6303,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6838235294117647,
      "eval_combined_score": 0.7412877397448839,
      "eval_f1": 0.7987519500780031,
      "eval_loss": 0.6091092824935913,
      "eval_runtime": 0.2809,
      "eval_samples_per_second": 1452.299,
      "eval_steps_per_second": 7.119,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8739886283874512,
      "learning_rate": 4.8e-05,
      "loss": 0.5743,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7132352941176471,
      "eval_combined_score": 0.7670311233222232,
      "eval_f1": 0.8208269525267994,
      "eval_loss": 0.5948534607887268,
      "eval_runtime": 0.2838,
      "eval_samples_per_second": 1437.56,
      "eval_steps_per_second": 7.047,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6283915042877197,
      "learning_rate": 4.7e-05,
      "loss": 0.5186,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7009803921568627,
      "eval_combined_score": 0.7563543936092956,
      "eval_f1": 0.8117283950617284,
      "eval_loss": 0.6003393530845642,
      "eval_runtime": 0.2903,
      "eval_samples_per_second": 1405.556,
      "eval_steps_per_second": 6.89,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.6895073652267456,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.4417,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6764705882352942,
      "eval_combined_score": 0.7244421906693712,
      "eval_f1": 0.7724137931034483,
      "eval_loss": 0.6426466107368469,
      "eval_runtime": 0.2719,
      "eval_samples_per_second": 1500.477,
      "eval_steps_per_second": 7.355,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.009364128112793,
      "learning_rate": 4.5e-05,
      "loss": 0.3244,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6691176470588235,
      "eval_combined_score": 0.7111584578986987,
      "eval_f1": 0.753199268738574,
      "eval_loss": 0.7227545976638794,
      "eval_runtime": 0.2795,
      "eval_samples_per_second": 1459.752,
      "eval_steps_per_second": 7.156,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.109133005142212,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.2037,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.678921568627451,
      "eval_combined_score": 0.722704991087344,
      "eval_f1": 0.7664884135472371,
      "eval_loss": 0.841673731803894,
      "eval_runtime": 0.2863,
      "eval_samples_per_second": 1425.321,
      "eval_steps_per_second": 6.987,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 2.244492769241333,
      "learning_rate": 4.3e-05,
      "loss": 0.1058,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6715686274509803,
      "eval_combined_score": 0.716141456582633,
      "eval_f1": 0.7607142857142857,
      "eval_loss": 1.0139018297195435,
      "eval_runtime": 0.2851,
      "eval_samples_per_second": 1431.127,
      "eval_steps_per_second": 7.015,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 3377819728711680.0,
      "train_loss": 0.3998178311756679,
      "train_runtime": 58.584,
      "train_samples_per_second": 3130.549,
      "train_steps_per_second": 12.802
    }
  ],
  "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
}