trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.42420217394828796,
  "best_model_checkpoint": "tiny_bert_rand_50_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": 1.624103307723999,
      "learning_rate": 4.9e-05,
      "loss": 0.4957,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7809300024734108,
      "eval_combined_score": 0.7270215614876743,
      "eval_f1": 0.6731131205019376,
      "eval_loss": 0.4535508453845978,
      "eval_runtime": 11.7106,
      "eval_samples_per_second": 3452.424,
      "eval_steps_per_second": 13.492,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.412515163421631,
      "learning_rate": 4.8e-05,
      "loss": 0.4098,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8002226069750186,
      "eval_combined_score": 0.7547840157117707,
      "eval_f1": 0.7093454244485228,
      "eval_loss": 0.42420217394828796,
      "eval_runtime": 11.4883,
      "eval_samples_per_second": 3519.22,
      "eval_steps_per_second": 13.753,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.119178533554077,
      "learning_rate": 4.7e-05,
      "loss": 0.3539,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8104377937175364,
      "eval_combined_score": 0.7727062749623714,
      "eval_f1": 0.7349747562072065,
      "eval_loss": 0.4246242642402649,
      "eval_runtime": 11.8603,
      "eval_samples_per_second": 3408.855,
      "eval_steps_per_second": 13.322,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.7941720485687256,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3091,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8165718525847143,
      "eval_combined_score": 0.7736160635771687,
      "eval_f1": 0.730660274569623,
      "eval_loss": 0.435099333524704,
      "eval_runtime": 11.4817,
      "eval_samples_per_second": 3521.267,
      "eval_steps_per_second": 13.761,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.839110851287842,
      "learning_rate": 4.5e-05,
      "loss": 0.272,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8204303734850359,
      "eval_combined_score": 0.7880094232089871,
      "eval_f1": 0.7555884729329383,
      "eval_loss": 0.43761980533599854,
      "eval_runtime": 11.4705,
      "eval_samples_per_second": 3524.709,
      "eval_steps_per_second": 13.775,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.09541654586792,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.2409,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8220133564185012,
      "eval_combined_score": 0.7904060212406911,
      "eval_f1": 0.7587986860628813,
      "eval_loss": 0.4505470395088196,
      "eval_runtime": 11.4624,
      "eval_samples_per_second": 3527.198,
      "eval_steps_per_second": 13.784,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 2.9798851013183594,
      "learning_rate": 4.3e-05,
      "loss": 0.2152,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8274795943606233,
      "eval_combined_score": 0.7946189856217504,
      "eval_f1": 0.7617583768828774,
      "eval_loss": 0.49352139234542847,
      "eval_runtime": 11.4875,
      "eval_samples_per_second": 3519.465,
      "eval_steps_per_second": 13.754,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 6.678940801217126e+16,
      "train_loss": 0.32807474441022455,
      "train_runtime": 1407.977,
      "train_samples_per_second": 12920.879,
      "train_steps_per_second": 50.498
    }
  ],
  "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": 6.678940801217126e+16,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}