trainer_state.json

{
  "best_metric": 83.0892405361961,
  "best_model_checkpoint": "/root/turkic_qa/tr_kaz_models/tr_kaz_xlm_roberta_large_model/checkpoint-2996",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 3745,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "step": 749,
      "train_exact_match": 61.13886113886114,
      "train_f1": 79.15769355969495,
      "train_runtime": 34.957,
      "train_samples_per_second": 43.482,
      "train_steps_per_second": 1.573
    },
    {
      "epoch": 1.0,
      "grad_norm": 48.082645416259766,
      "learning_rate": 1e-05,
      "loss": 2.8895,
      "step": 749
    },
    {
      "epoch": 1.0,
      "eval_exact_match": 58.53125,
      "eval_f1": 77.23870107427594,
      "eval_runtime": 105.5703,
      "eval_samples_per_second": 43.743,
      "eval_steps_per_second": 1.563,
      "step": 749
    },
    {
      "epoch": 2.0,
      "step": 1498,
      "train_exact_match": 74.12587412587412,
      "train_f1": 86.83221994171714,
      "train_runtime": 33.8383,
      "train_samples_per_second": 43.797,
      "train_steps_per_second": 1.566
    },
    {
      "epoch": 2.0,
      "grad_norm": 21.845104217529297,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.8623,
      "step": 1498
    },
    {
      "epoch": 2.0,
      "eval_exact_match": 65.71875,
      "eval_f1": 81.74157715854905,
      "eval_runtime": 105.3081,
      "eval_samples_per_second": 43.852,
      "eval_steps_per_second": 1.567,
      "step": 1498
    },
    {
      "epoch": 3.0,
      "step": 2247,
      "train_exact_match": 81.21878121878122,
      "train_f1": 91.78878436784974,
      "train_runtime": 34.2658,
      "train_samples_per_second": 43.688,
      "train_steps_per_second": 1.576
    },
    {
      "epoch": 3.0,
      "grad_norm": 24.28180503845215,
      "learning_rate": 5e-06,
      "loss": 0.5683,
      "step": 2247
    },
    {
      "epoch": 3.0,
      "eval_exact_match": 66.8125,
      "eval_f1": 82.38007432685225,
      "eval_runtime": 105.2928,
      "eval_samples_per_second": 43.859,
      "eval_steps_per_second": 1.567,
      "step": 2247
    },
    {
      "epoch": 4.0,
      "step": 2996,
      "train_exact_match": 83.81618381618381,
      "train_f1": 93.12987614032765,
      "train_runtime": 34.4648,
      "train_samples_per_second": 43.697,
      "train_steps_per_second": 1.567
    },
    {
      "epoch": 4.0,
      "grad_norm": 40.392921447753906,
      "learning_rate": 2.5e-06,
      "loss": 0.4119,
      "step": 2996
    },
    {
      "epoch": 4.0,
      "eval_exact_match": 67.78125,
      "eval_f1": 83.0892405361961,
      "eval_runtime": 105.4844,
      "eval_samples_per_second": 43.779,
      "eval_steps_per_second": 1.564,
      "step": 2996
    },
    {
      "epoch": 5.0,
      "step": 3745,
      "train_exact_match": 86.41358641358642,
      "train_f1": 94.67861425658204,
      "train_runtime": 34.6799,
      "train_samples_per_second": 43.137,
      "train_steps_per_second": 1.557
    },
    {
      "epoch": 5.0,
      "grad_norm": 17.308034896850586,
      "learning_rate": 0.0,
      "loss": 0.3072,
      "step": 3745
    },
    {
      "epoch": 5.0,
      "eval_exact_match": 68.21875,
      "eval_f1": 82.89346018544795,
      "eval_runtime": 106.6861,
      "eval_samples_per_second": 43.286,
      "eval_steps_per_second": 1.547,
      "step": 3745
    },
    {
      "epoch": 5.0,
      "step": 3745,
      "total_flos": 7.301726548858368e+16,
      "train_loss": 1.0078204329405354,
      "train_runtime": 6759.161,
      "train_samples_per_second": 15.509,
      "train_steps_per_second": 0.554
    }
  ],
  "logging_steps": 500,
  "max_steps": 3745,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 7.301726548858368e+16,
  "train_batch_size": 28,
  "trial_name": null,
  "trial_params": null
}