trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5820168852806091,
  "best_model_checkpoint": "bert_base_km_20_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": 1.330278754234314,
      "learning_rate": 4.9e-05,
      "loss": 0.6264,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7107843137254902,
      "eval_combined_score": 0.760840874811463,
      "eval_f1": 0.8108974358974359,
      "eval_loss": 0.6000066995620728,
      "eval_runtime": 0.2813,
      "eval_samples_per_second": 1450.372,
      "eval_steps_per_second": 7.11,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7250771522521973,
      "learning_rate": 4.8e-05,
      "loss": 0.5604,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7132352941176471,
      "eval_combined_score": 0.7659199726402188,
      "eval_f1": 0.8186046511627907,
      "eval_loss": 0.5820168852806091,
      "eval_runtime": 0.277,
      "eval_samples_per_second": 1472.717,
      "eval_steps_per_second": 7.219,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.663827419281006,
      "learning_rate": 4.7e-05,
      "loss": 0.4642,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6862745098039216,
      "eval_combined_score": 0.7324106113033448,
      "eval_f1": 0.7785467128027682,
      "eval_loss": 0.6450715661048889,
      "eval_runtime": 0.281,
      "eval_samples_per_second": 1451.831,
      "eval_steps_per_second": 7.117,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.8037514686584473,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3411,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6838235294117647,
      "eval_combined_score": 0.72895204491604,
      "eval_f1": 0.7740805604203153,
      "eval_loss": 0.7175894379615784,
      "eval_runtime": 0.2762,
      "eval_samples_per_second": 1476.972,
      "eval_steps_per_second": 7.24,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.8671875,
      "learning_rate": 4.5e-05,
      "loss": 0.1998,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6911764705882353,
      "eval_combined_score": 0.7388085742771685,
      "eval_f1": 0.7864406779661017,
      "eval_loss": 0.9084722399711609,
      "eval_runtime": 0.2766,
      "eval_samples_per_second": 1475.0,
      "eval_steps_per_second": 7.23,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.317751169204712,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.1089,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6004901960784313,
      "eval_combined_score": 0.6450070028011204,
      "eval_f1": 0.6895238095238095,
      "eval_loss": 1.1335161924362183,
      "eval_runtime": 0.2999,
      "eval_samples_per_second": 1360.656,
      "eval_steps_per_second": 6.67,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 5.213982105255127,
      "learning_rate": 4.3e-05,
      "loss": 0.0526,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6470588235294118,
      "eval_combined_score": 0.6958698372966208,
      "eval_f1": 0.7446808510638298,
      "eval_loss": 1.4105342626571655,
      "eval_runtime": 0.3111,
      "eval_samples_per_second": 1311.584,
      "eval_steps_per_second": 6.429,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 3377819728711680.0,
      "train_loss": 0.33620565221423193,
      "train_runtime": 59.6114,
      "train_samples_per_second": 3076.594,
      "train_steps_per_second": 12.581
    }
  ],
  "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
}