trainer_state.json

{
  "best_global_step": 820,
  "best_metric": 0.6186944246292114,
  "best_model_checkpoint": "bert_base_km_5_v2_qnli/checkpoint-820",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 2870,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.429147243499756,
      "learning_rate": 4.9e-05,
      "loss": 0.664,
      "step": 410
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.638843126487278,
      "eval_loss": 0.6346889138221741,
      "eval_runtime": 3.4402,
      "eval_samples_per_second": 1587.979,
      "eval_steps_per_second": 6.395,
      "step": 410
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.4559290409088135,
      "learning_rate": 4.8e-05,
      "loss": 0.6207,
      "step": 820
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6538531942156325,
      "eval_loss": 0.6186944246292114,
      "eval_runtime": 3.3981,
      "eval_samples_per_second": 1607.64,
      "eval_steps_per_second": 6.474,
      "step": 820
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.064799785614014,
      "learning_rate": 4.7e-05,
      "loss": 0.5424,
      "step": 1230
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6511074501189822,
      "eval_loss": 0.6316582560539246,
      "eval_runtime": 3.4193,
      "eval_samples_per_second": 1597.706,
      "eval_steps_per_second": 6.434,
      "step": 1230
    },
    {
      "epoch": 4.0,
      "grad_norm": 5.049038887023926,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.431,
      "step": 1640
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6836902800658978,
      "eval_loss": 0.6447038054466248,
      "eval_runtime": 3.4239,
      "eval_samples_per_second": 1595.541,
      "eval_steps_per_second": 6.425,
      "step": 1640
    },
    {
      "epoch": 5.0,
      "grad_norm": 5.788658142089844,
      "learning_rate": 4.5e-05,
      "loss": 0.3039,
      "step": 2050
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.680578436756361,
      "eval_loss": 0.7469810247421265,
      "eval_runtime": 3.4209,
      "eval_samples_per_second": 1596.956,
      "eval_steps_per_second": 6.431,
      "step": 2050
    },
    {
      "epoch": 6.0,
      "grad_norm": 9.675091743469238,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.2059,
      "step": 2460
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6851546769174446,
      "eval_loss": 0.8305718302726746,
      "eval_runtime": 3.4127,
      "eval_samples_per_second": 1600.766,
      "eval_steps_per_second": 6.446,
      "step": 2460
    },
    {
      "epoch": 7.0,
      "grad_norm": 7.534983158111572,
      "learning_rate": 4.3e-05,
      "loss": 0.146,
      "step": 2870
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6719751052535237,
      "eval_loss": 1.0975512266159058,
      "eval_runtime": 3.4055,
      "eval_samples_per_second": 1604.176,
      "eval_steps_per_second": 6.46,
      "step": 2870
    },
    {
      "epoch": 7.0,
      "step": 2870,
      "total_flos": 9.645664445050368e+16,
      "train_loss": 0.4162792006436125,
      "train_runtime": 1214.3467,
      "train_samples_per_second": 4312.73,
      "train_steps_per_second": 16.882
    }
  ],
  "logging_steps": 1,
  "max_steps": 20500,
  "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": 9.645664445050368e+16,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}