trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.3983299732208252,
  "best_model_checkpoint": "bert_base_km_10_v1_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": 1.8717503547668457,
      "learning_rate": 4.9e-05,
      "loss": 0.4756,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7900321543408361,
      "eval_combined_score": 0.7309520672751686,
      "eval_f1": 0.671871980209501,
      "eval_loss": 0.4438171982765198,
      "eval_runtime": 25.8505,
      "eval_samples_per_second": 1563.992,
      "eval_steps_per_second": 6.112,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.944045305252075,
      "learning_rate": 4.8e-05,
      "loss": 0.3636,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8171407370764284,
      "eval_combined_score": 0.7832100402579552,
      "eval_f1": 0.7492793434394818,
      "eval_loss": 0.3983299732208252,
      "eval_runtime": 25.2779,
      "eval_samples_per_second": 1599.421,
      "eval_steps_per_second": 6.251,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.461521863937378,
      "learning_rate": 4.7e-05,
      "loss": 0.2766,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8278258718773188,
      "eval_combined_score": 0.7976598880746693,
      "eval_f1": 0.7674939042720198,
      "eval_loss": 0.419876366853714,
      "eval_runtime": 25.2754,
      "eval_samples_per_second": 1599.582,
      "eval_steps_per_second": 6.251,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.752514600753784,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.2059,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8327232253277269,
      "eval_combined_score": 0.7966158709591289,
      "eval_f1": 0.7605085165905309,
      "eval_loss": 0.46601402759552,
      "eval_runtime": 25.4001,
      "eval_samples_per_second": 1591.723,
      "eval_steps_per_second": 6.22,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.1064395904541016,
      "learning_rate": 4.5e-05,
      "loss": 0.1549,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8351471679445956,
      "eval_combined_score": 0.8002196381329486,
      "eval_f1": 0.7652921083213016,
      "eval_loss": 0.5142884254455566,
      "eval_runtime": 25.3616,
      "eval_samples_per_second": 1594.14,
      "eval_steps_per_second": 6.23,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 5.583347320556641,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.1209,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8362849369280237,
      "eval_combined_score": 0.8034324064211018,
      "eval_f1": 0.7705798759141798,
      "eval_loss": 0.5888400673866272,
      "eval_runtime": 25.2856,
      "eval_samples_per_second": 1598.932,
      "eval_steps_per_second": 6.249,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.643422842025757,
      "learning_rate": 4.3e-05,
      "loss": 0.0957,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8387830818698986,
      "eval_combined_score": 0.8079193833210486,
      "eval_f1": 0.7770556847721987,
      "eval_loss": 0.6113874316215515,
      "eval_runtime": 25.3801,
      "eval_samples_per_second": 1592.979,
      "eval_steps_per_second": 6.225,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 3.3506166766980096e+17,
      "train_loss": 0.2418764561612323,
      "train_runtime": 4259.546,
      "train_samples_per_second": 4270.948,
      "train_steps_per_second": 16.692
    }
  ],
  "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
}