trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5893518924713135,
  "best_model_checkpoint": "bert_base_km_50_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": 1.1089715957641602,
      "learning_rate": 4.9e-05,
      "loss": 0.6263,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6887254901960784,
      "eval_combined_score": 0.7452987825395369,
      "eval_f1": 0.8018720748829953,
      "eval_loss": 0.5977591872215271,
      "eval_runtime": 0.2712,
      "eval_samples_per_second": 1504.571,
      "eval_steps_per_second": 7.375,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4218831062316895,
      "learning_rate": 4.8e-05,
      "loss": 0.572,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7034313725490197,
      "eval_combined_score": 0.7587817385018524,
      "eval_f1": 0.8141321044546851,
      "eval_loss": 0.5893518924713135,
      "eval_runtime": 0.2717,
      "eval_samples_per_second": 1501.661,
      "eval_steps_per_second": 7.361,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.491407871246338,
      "learning_rate": 4.7e-05,
      "loss": 0.5108,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7034313725490197,
      "eval_combined_score": 0.7552244582043344,
      "eval_f1": 0.8070175438596491,
      "eval_loss": 0.6016624569892883,
      "eval_runtime": 0.2743,
      "eval_samples_per_second": 1487.603,
      "eval_steps_per_second": 7.292,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.704878091812134,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.4158,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6838235294117647,
      "eval_combined_score": 0.737373677185623,
      "eval_f1": 0.7909238249594813,
      "eval_loss": 0.6796642541885376,
      "eval_runtime": 0.2784,
      "eval_samples_per_second": 1465.639,
      "eval_steps_per_second": 7.185,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.1070427894592285,
      "learning_rate": 4.5e-05,
      "loss": 0.2727,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6446078431372549,
      "eval_combined_score": 0.6872946105816629,
      "eval_f1": 0.7299813780260708,
      "eval_loss": 0.8248001337051392,
      "eval_runtime": 0.2734,
      "eval_samples_per_second": 1492.086,
      "eval_steps_per_second": 7.314,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.6518099308013916,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.1534,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6176470588235294,
      "eval_combined_score": 0.6659663865546219,
      "eval_f1": 0.7142857142857143,
      "eval_loss": 1.088710904121399,
      "eval_runtime": 0.2757,
      "eval_samples_per_second": 1479.813,
      "eval_steps_per_second": 7.254,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 4.282205581665039,
      "learning_rate": 4.3e-05,
      "loss": 0.083,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6593137254901961,
      "eval_combined_score": 0.7062110368125936,
      "eval_f1": 0.7531083481349912,
      "eval_loss": 1.1340426206588745,
      "eval_runtime": 0.289,
      "eval_samples_per_second": 1411.96,
      "eval_steps_per_second": 6.921,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 3377819728711680.0,
      "train_loss": 0.37630167575109574,
      "train_runtime": 60.942,
      "train_samples_per_second": 3009.42,
      "train_steps_per_second": 12.307
    }
  ],
  "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
}