run-32/checkpoint-1032/trainer_state.json

{
  "best_metric": 0.8126666666666666,
  "best_model_checkpoint": "Bert_tinybert-distilled/run-32/checkpoint-860",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 1032,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.8815119862556458,
      "learning_rate": 0.00019666908200653298,
      "loss": 0.5311,
      "step": 172
    },
    {
      "epoch": 1.0,
      "eval_AUC": 0.8264693333333334,
      "eval_BAC": 0.7413333333333333,
      "eval_MCC": 0.48401777889745745,
      "eval_PR_AUC": 0.7834479949491041,
      "eval_accuracy": 0.7413333333333333,
      "eval_loss": 0.47986286878585815,
      "eval_runtime": 6.254,
      "eval_samples_per_second": 239.845,
      "eval_steps_per_second": 7.515,
      "step": 172
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.847335696220398,
      "learning_rate": 0.00016389090167211084,
      "loss": 0.4906,
      "step": 344
    },
    {
      "epoch": 2.0,
      "eval_AUC": 0.8593288888888888,
      "eval_BAC": 0.7753333333333333,
      "eval_MCC": 0.5809536114294386,
      "eval_PR_AUC": 0.8208211551656307,
      "eval_accuracy": 0.7753333333333333,
      "eval_loss": 0.46062198281288147,
      "eval_runtime": 6.2859,
      "eval_samples_per_second": 238.629,
      "eval_steps_per_second": 7.477,
      "step": 344
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.6832430362701416,
      "learning_rate": 0.00013111272133768865,
      "loss": 0.4723,
      "step": 516
    },
    {
      "epoch": 3.0,
      "eval_AUC": 0.8711697777777777,
      "eval_BAC": 0.8086666666666666,
      "eval_MCC": 0.6201182679233241,
      "eval_PR_AUC": 0.8332422338463044,
      "eval_accuracy": 0.8086666666666666,
      "eval_loss": 0.42795059084892273,
      "eval_runtime": 6.25,
      "eval_samples_per_second": 240.001,
      "eval_steps_per_second": 7.52,
      "step": 516
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.4761440753936768,
      "learning_rate": 9.833454100326649e-05,
      "loss": 0.4616,
      "step": 688
    },
    {
      "epoch": 4.0,
      "eval_AUC": 0.8759911111111109,
      "eval_BAC": 0.808,
      "eval_MCC": 0.6199002844761639,
      "eval_PR_AUC": 0.8424113281283504,
      "eval_accuracy": 0.808,
      "eval_loss": 0.4210560619831085,
      "eval_runtime": 6.2121,
      "eval_samples_per_second": 241.463,
      "eval_steps_per_second": 7.566,
      "step": 688
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.6606786251068115,
      "learning_rate": 6.555636066884433e-05,
      "loss": 0.4489,
      "step": 860
    },
    {
      "epoch": 5.0,
      "eval_AUC": 0.8827644444444446,
      "eval_BAC": 0.8126666666666666,
      "eval_MCC": 0.6317961066346743,
      "eval_PR_AUC": 0.8607459277405258,
      "eval_accuracy": 0.8126666666666666,
      "eval_loss": 0.4215475618839264,
      "eval_runtime": 6.2337,
      "eval_samples_per_second": 240.629,
      "eval_steps_per_second": 7.54,
      "step": 860
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.446411371231079,
      "learning_rate": 3.277818033442216e-05,
      "loss": 0.445,
      "step": 1032
    },
    {
      "epoch": 6.0,
      "eval_AUC": 0.881552,
      "eval_BAC": 0.8106666666666666,
      "eval_MCC": 0.628637794247896,
      "eval_PR_AUC": 0.8568648106694358,
      "eval_accuracy": 0.8106666666666666,
      "eval_loss": 0.42053866386413574,
      "eval_runtime": 6.2148,
      "eval_samples_per_second": 241.361,
      "eval_steps_per_second": 7.563,
      "step": 1032
    }
  ],
  "logging_steps": 500,
  "max_steps": 1204,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2702266380000.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.8238347003864188,
    "learning_rate": 0.00022944726234095517,
    "num_train_epochs": 7,
    "temperature": 4
  }
}