trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5926130414009094,
  "best_model_checkpoint": "bert_base_km_5_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": 1.390352725982666,
      "learning_rate": 4.9e-05,
      "loss": 0.6268,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7058823529411765,
      "eval_combined_score": 0.7617557661362417,
      "eval_f1": 0.817629179331307,
      "eval_loss": 0.5994035005569458,
      "eval_runtime": 0.2782,
      "eval_samples_per_second": 1466.706,
      "eval_steps_per_second": 7.19,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7472285032272339,
      "learning_rate": 4.8e-05,
      "loss": 0.555,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7181372549019608,
      "eval_combined_score": 0.7676536830948595,
      "eval_f1": 0.8171701112877583,
      "eval_loss": 0.5926130414009094,
      "eval_runtime": 0.2849,
      "eval_samples_per_second": 1432.312,
      "eval_steps_per_second": 7.021,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.221715211868286,
      "learning_rate": 4.7e-05,
      "loss": 0.45,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6936274509803921,
      "eval_combined_score": 0.7445223997946151,
      "eval_f1": 0.795417348608838,
      "eval_loss": 0.6521729230880737,
      "eval_runtime": 0.2697,
      "eval_samples_per_second": 1512.718,
      "eval_steps_per_second": 7.415,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 4.633323669433594,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.2999,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.678921568627451,
      "eval_combined_score": 0.7255477408354647,
      "eval_f1": 0.7721739130434783,
      "eval_loss": 0.8733688592910767,
      "eval_runtime": 0.2733,
      "eval_samples_per_second": 1492.947,
      "eval_steps_per_second": 7.318,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.693900108337402,
      "learning_rate": 4.5e-05,
      "loss": 0.1672,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6004901960784313,
      "eval_combined_score": 0.6432123427405645,
      "eval_f1": 0.6859344894026975,
      "eval_loss": 1.059749960899353,
      "eval_runtime": 0.2877,
      "eval_samples_per_second": 1418.13,
      "eval_steps_per_second": 6.952,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.3627774715423584,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.0863,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6323529411764706,
      "eval_combined_score": 0.6778000868243976,
      "eval_f1": 0.7232472324723247,
      "eval_loss": 1.2696877717971802,
      "eval_runtime": 0.274,
      "eval_samples_per_second": 1489.267,
      "eval_steps_per_second": 7.3,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.4011316299438477,
      "learning_rate": 4.3e-05,
      "loss": 0.0536,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6593137254901961,
      "eval_combined_score": 0.7100355202321893,
      "eval_f1": 0.7607573149741824,
      "eval_loss": 1.4808331727981567,
      "eval_runtime": 0.3198,
      "eval_samples_per_second": 1275.628,
      "eval_steps_per_second": 6.253,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 3377819728711680.0,
      "train_loss": 0.31983384234564644,
      "train_runtime": 59.582,
      "train_samples_per_second": 3078.112,
      "train_steps_per_second": 12.588
    }
  ],
  "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
}