trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.39002183079719543,
  "best_model_checkpoint": "distilbert_rand_100_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": 2.175678253173828,
      "learning_rate": 4.9e-05,
      "loss": 0.4782,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7897600791491467,
      "eval_combined_score": 0.7315566977868553,
      "eval_f1": 0.6733533164245639,
      "eval_loss": 0.43471285700798035,
      "eval_runtime": 16.686,
      "eval_samples_per_second": 2422.983,
      "eval_steps_per_second": 9.469,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.3648293018341064,
      "learning_rate": 4.8e-05,
      "loss": 0.37,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8204303734850359,
      "eval_combined_score": 0.7890212764861078,
      "eval_f1": 0.7576121794871795,
      "eval_loss": 0.39002183079719543,
      "eval_runtime": 16.3282,
      "eval_samples_per_second": 2476.081,
      "eval_steps_per_second": 9.676,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.203273296356201,
      "learning_rate": 4.7e-05,
      "loss": 0.2974,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8244620331437051,
      "eval_combined_score": 0.7965354083425806,
      "eval_f1": 0.7686087835414561,
      "eval_loss": 0.4048449397087097,
      "eval_runtime": 16.4092,
      "eval_samples_per_second": 2463.861,
      "eval_steps_per_second": 9.629,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 4.714377403259277,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.2389,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8310660400692556,
      "eval_combined_score": 0.7940892078013418,
      "eval_f1": 0.7571123755334281,
      "eval_loss": 0.4422609806060791,
      "eval_runtime": 16.367,
      "eval_samples_per_second": 2470.213,
      "eval_steps_per_second": 9.654,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.3094966411590576,
      "learning_rate": 4.5e-05,
      "loss": 0.1926,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8335147167944595,
      "eval_combined_score": 0.7998875455695047,
      "eval_f1": 0.7662603743445497,
      "eval_loss": 0.4478583335876465,
      "eval_runtime": 16.2704,
      "eval_samples_per_second": 2484.887,
      "eval_steps_per_second": 9.711,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.7431838512420654,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.1563,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8346772198862231,
      "eval_combined_score": 0.8063013317186448,
      "eval_f1": 0.7779254435510665,
      "eval_loss": 0.5162534117698669,
      "eval_runtime": 16.3371,
      "eval_samples_per_second": 2474.731,
      "eval_steps_per_second": 9.671,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 5.1978912353515625,
      "learning_rate": 4.3e-05,
      "loss": 0.1291,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8348503586445709,
      "eval_combined_score": 0.8053836986363103,
      "eval_f1": 0.7759170386280498,
      "eval_loss": 0.5895099639892578,
      "eval_runtime": 16.2465,
      "eval_samples_per_second": 2488.535,
      "eval_steps_per_second": 9.725,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 1.686920659598684e+17,
      "train_loss": 0.2660790754413662,
      "train_runtime": 2426.9703,
      "train_samples_per_second": 7495.889,
      "train_steps_per_second": 29.296
    }
  ],
  "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": 1.686920659598684e+17,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}