trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5934136509895325,
  "best_model_checkpoint": "tiny_bert_rand_5_v2_mrpc/checkpoint-30",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 105,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.8713007569313049,
      "learning_rate": 4.9e-05,
      "loss": 0.6344,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7107843137254902,
      "eval_combined_score": 0.7673324553702078,
      "eval_f1": 0.8238805970149253,
      "eval_loss": 0.6029195785522461,
      "eval_runtime": 0.1384,
      "eval_samples_per_second": 2947.361,
      "eval_steps_per_second": 14.448,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1738929748535156,
      "learning_rate": 4.8e-05,
      "loss": 0.5907,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6985294117647058,
      "eval_combined_score": 0.7521083077780875,
      "eval_f1": 0.8056872037914692,
      "eval_loss": 0.5934136509895325,
      "eval_runtime": 0.1454,
      "eval_samples_per_second": 2806.829,
      "eval_steps_per_second": 13.759,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8110806345939636,
      "learning_rate": 4.7e-05,
      "loss": 0.5576,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7058823529411765,
      "eval_combined_score": 0.7614777618364419,
      "eval_f1": 0.8170731707317073,
      "eval_loss": 0.6198119521141052,
      "eval_runtime": 0.1393,
      "eval_samples_per_second": 2928.317,
      "eval_steps_per_second": 14.354,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.3468823432922363,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.5277,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6838235294117647,
      "eval_combined_score": 0.7261128419051642,
      "eval_f1": 0.7684021543985637,
      "eval_loss": 0.60675048828125,
      "eval_runtime": 0.1456,
      "eval_samples_per_second": 2801.609,
      "eval_steps_per_second": 13.733,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.008892297744751,
      "learning_rate": 4.5e-05,
      "loss": 0.449,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.696078431372549,
      "eval_combined_score": 0.7418748321246307,
      "eval_f1": 0.7876712328767124,
      "eval_loss": 0.6314738988876343,
      "eval_runtime": 0.1554,
      "eval_samples_per_second": 2626.006,
      "eval_steps_per_second": 12.873,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.5793399810791016,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.3555,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6642156862745098,
      "eval_combined_score": 0.7054904679062013,
      "eval_f1": 0.7467652495378928,
      "eval_loss": 0.7845552563667297,
      "eval_runtime": 0.1323,
      "eval_samples_per_second": 3083.78,
      "eval_steps_per_second": 15.117,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.1420841217041016,
      "learning_rate": 4.3e-05,
      "loss": 0.2687,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.678921568627451,
      "eval_combined_score": 0.722704991087344,
      "eval_f1": 0.7664884135472371,
      "eval_loss": 0.8464624285697937,
      "eval_runtime": 0.1327,
      "eval_samples_per_second": 3073.457,
      "eval_steps_per_second": 15.066,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 673316591603712.0,
      "train_loss": 0.4833688463483538,
      "train_runtime": 19.8418,
      "train_samples_per_second": 9243.133,
      "train_steps_per_second": 37.799
    }
  ],
  "logging_steps": 1,
  "max_steps": 750,
  "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": 673316591603712.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}