trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.3973020613193512,
  "best_model_checkpoint": "bert_base_km_20_v2_qqp/checkpoint-2844",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 9954,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.2013742923736572,
      "learning_rate": 4.9e-05,
      "loss": 0.4797,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.792851842691071,
      "eval_combined_score": 0.7404029696707339,
      "eval_f1": 0.6879540966503969,
      "eval_loss": 0.4331514239311218,
      "eval_runtime": 25.8106,
      "eval_samples_per_second": 1566.409,
      "eval_steps_per_second": 6.122,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.980579376220703,
      "learning_rate": 4.8e-05,
      "loss": 0.371,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8158792975513233,
      "eval_combined_score": 0.7832912897669544,
      "eval_f1": 0.7507032819825854,
      "eval_loss": 0.3973020613193512,
      "eval_runtime": 25.3068,
      "eval_samples_per_second": 1597.594,
      "eval_steps_per_second": 6.243,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.9177956581115723,
      "learning_rate": 4.7e-05,
      "loss": 0.2919,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8166707890180559,
      "eval_combined_score": 0.7943677458221469,
      "eval_f1": 0.7720647026262378,
      "eval_loss": 0.4061584174633026,
      "eval_runtime": 25.3481,
      "eval_samples_per_second": 1594.991,
      "eval_steps_per_second": 6.233,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.2191033363342285,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.2234,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8318327974276527,
      "eval_combined_score": 0.8014820903016676,
      "eval_f1": 0.7711313831756825,
      "eval_loss": 0.4300062358379364,
      "eval_runtime": 25.3498,
      "eval_samples_per_second": 1594.884,
      "eval_steps_per_second": 6.233,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.6099495887756348,
      "learning_rate": 4.5e-05,
      "loss": 0.1687,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8354439772446204,
      "eval_combined_score": 0.805214801555648,
      "eval_f1": 0.7749856258666756,
      "eval_loss": 0.4925248324871063,
      "eval_runtime": 25.4043,
      "eval_samples_per_second": 1591.461,
      "eval_steps_per_second": 6.219,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 4.482016563415527,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.1294,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8354439772446204,
      "eval_combined_score": 0.802029713033676,
      "eval_f1": 0.7686154488227316,
      "eval_loss": 0.5531768798828125,
      "eval_runtime": 25.333,
      "eval_samples_per_second": 1595.945,
      "eval_steps_per_second": 6.237,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 5.143055438995361,
      "learning_rate": 4.3e-05,
      "loss": 0.1015,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8350234974029186,
      "eval_combined_score": 0.8048732726139154,
      "eval_f1": 0.7747230478249122,
      "eval_loss": 0.6021216511726379,
      "eval_runtime": 25.3526,
      "eval_samples_per_second": 1594.705,
      "eval_steps_per_second": 6.232,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 3.3506166766980096e+17,
      "train_loss": 0.2522317121037044,
      "train_runtime": 4251.0904,
      "train_samples_per_second": 4279.443,
      "train_steps_per_second": 16.725
    }
  ],
  "logging_steps": 1,
  "max_steps": 71100,
  "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": 3.3506166766980096e+17,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}