trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.6164862513542175,
  "best_model_checkpoint": "bert_base_km_100_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": 1.257430076599121,
      "learning_rate": 4.9e-05,
      "loss": 0.6278,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6813725490196079,
      "eval_combined_score": 0.7400670794633644,
      "eval_f1": 0.7987616099071208,
      "eval_loss": 0.6242939233779907,
      "eval_runtime": 0.2756,
      "eval_samples_per_second": 1480.504,
      "eval_steps_per_second": 7.257,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5219076871871948,
      "learning_rate": 4.8e-05,
      "loss": 0.5651,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7058823529411765,
      "eval_combined_score": 0.7614777618364419,
      "eval_f1": 0.8170731707317073,
      "eval_loss": 0.6164862513542175,
      "eval_runtime": 0.2706,
      "eval_samples_per_second": 1507.706,
      "eval_steps_per_second": 7.391,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.705201506614685,
      "learning_rate": 4.7e-05,
      "loss": 0.489,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.696078431372549,
      "eval_combined_score": 0.7517659237608086,
      "eval_f1": 0.8074534161490683,
      "eval_loss": 0.6588466763496399,
      "eval_runtime": 0.2766,
      "eval_samples_per_second": 1475.276,
      "eval_steps_per_second": 7.232,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.3031058311462402,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3881,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6813725490196079,
      "eval_combined_score": 0.7312586650821945,
      "eval_f1": 0.7811447811447811,
      "eval_loss": 0.7228206396102905,
      "eval_runtime": 0.2874,
      "eval_samples_per_second": 1419.49,
      "eval_steps_per_second": 6.958,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.1388323307037354,
      "learning_rate": 4.5e-05,
      "loss": 0.2718,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6004901960784313,
      "eval_combined_score": 0.650703813635546,
      "eval_f1": 0.7009174311926606,
      "eval_loss": 0.8791874647140503,
      "eval_runtime": 0.2849,
      "eval_samples_per_second": 1432.17,
      "eval_steps_per_second": 7.02,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.5961313247680664,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.165,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.625,
      "eval_combined_score": 0.6780536028119508,
      "eval_f1": 0.7311072056239016,
      "eval_loss": 1.0607061386108398,
      "eval_runtime": 0.275,
      "eval_samples_per_second": 1483.722,
      "eval_steps_per_second": 7.273,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 2.8271677494049072,
      "learning_rate": 4.3e-05,
      "loss": 0.0868,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.625,
      "eval_combined_score": 0.6789921465968587,
      "eval_f1": 0.7329842931937173,
      "eval_loss": 1.1696970462799072,
      "eval_runtime": 0.2861,
      "eval_samples_per_second": 1425.908,
      "eval_steps_per_second": 6.99,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 3377819728711680.0,
      "train_loss": 0.37051124686286563,
      "train_runtime": 58.9087,
      "train_samples_per_second": 3113.291,
      "train_steps_per_second": 12.732
    }
  ],
  "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
}