{
  "best_metric": 76.09384754933019,
  "best_model_checkpoint": "/root/turkic_qa/ru_uzn_models/ru_uzn_xlm_roberta_base_squad_model/checkpoint-5022",
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 5580,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "step": 558,
      "train_exact_match": 54.345654345654346,
      "train_f1": 72.79270519378026,
      "train_runtime": 11.7013,
      "train_samples_per_second": 90.247,
      "train_steps_per_second": 3.248
    },
    {
      "epoch": 1.0,
      "grad_norm": 80.72039031982422,
      "learning_rate": 5e-06,
      "loss": 1.7207,
      "step": 558
    },
    {
      "epoch": 1.0,
      "eval_exact_match": 52.21875,
      "eval_f1": 71.55991959566774,
      "eval_runtime": 37.0088,
      "eval_samples_per_second": 90.762,
      "eval_steps_per_second": 3.242,
      "step": 558
    },
    {
      "epoch": 2.0,
      "step": 1116,
      "train_exact_match": 59.54045954045954,
      "train_f1": 78.95661212870544,
      "train_runtime": 11.6455,
      "train_samples_per_second": 89.39,
      "train_steps_per_second": 3.263
    },
    {
      "epoch": 2.0,
      "grad_norm": 73.76553344726562,
      "learning_rate": 1e-05,
      "loss": 1.352,
      "step": 1116
    },
    {
      "epoch": 2.0,
      "eval_exact_match": 55.46875,
      "eval_f1": 74.01581668435861,
      "eval_runtime": 37.1575,
      "eval_samples_per_second": 90.399,
      "eval_steps_per_second": 3.229,
      "step": 1116
    },
    {
      "epoch": 3.0,
      "step": 1674,
      "train_exact_match": 66.13386613386614,
      "train_f1": 83.54611322142382,
      "train_runtime": 11.6678,
      "train_samples_per_second": 90.163,
      "train_steps_per_second": 3.257
    },
    {
      "epoch": 3.0,
      "grad_norm": 73.22183227539062,
      "learning_rate": 8.750000000000001e-06,
      "loss": 1.1278,
      "step": 1674
    },
    {
      "epoch": 3.0,
      "eval_exact_match": 57.40625,
      "eval_f1": 75.65492929099143,
      "eval_runtime": 37.0792,
      "eval_samples_per_second": 90.59,
      "eval_steps_per_second": 3.236,
      "step": 1674
    },
    {
      "epoch": 4.0,
      "step": 2232,
      "train_exact_match": 67.23276723276723,
      "train_f1": 85.47766787697041,
      "train_runtime": 11.6568,
      "train_samples_per_second": 89.733,
      "train_steps_per_second": 3.26
    },
    {
      "epoch": 4.0,
      "grad_norm": 12.79330825805664,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.9505,
      "step": 2232
    },
    {
      "epoch": 4.0,
      "eval_exact_match": 57.9375,
      "eval_f1": 75.99186608404226,
      "eval_runtime": 37.0588,
      "eval_samples_per_second": 90.64,
      "eval_steps_per_second": 3.238,
      "step": 2232
    },
    {
      "epoch": 5.0,
      "step": 2790,
      "train_exact_match": 73.62637362637362,
      "train_f1": 88.41775048247749,
      "train_runtime": 11.6282,
      "train_samples_per_second": 89.61,
      "train_steps_per_second": 3.268
    },
    {
      "epoch": 5.0,
      "grad_norm": 57.348087310791016,
      "learning_rate": 6.25e-06,
      "loss": 0.8175,
      "step": 2790
    },
    {
      "epoch": 5.0,
      "eval_exact_match": 58.34375,
      "eval_f1": 76.03000703182897,
      "eval_runtime": 37.1357,
      "eval_samples_per_second": 90.452,
      "eval_steps_per_second": 3.231,
      "step": 2790
    },
    {
      "epoch": 6.0,
      "step": 3348,
      "train_exact_match": 78.62137862137862,
      "train_f1": 90.6016744099524,
      "train_runtime": 11.6628,
      "train_samples_per_second": 89.515,
      "train_steps_per_second": 3.258
    },
    {
      "epoch": 6.0,
      "grad_norm": 90.37775421142578,
      "learning_rate": 5e-06,
      "loss": 0.7139,
      "step": 3348
    },
    {
      "epoch": 6.0,
      "eval_exact_match": 58.8125,
      "eval_f1": 75.74834491209981,
      "eval_runtime": 37.1026,
      "eval_samples_per_second": 90.533,
      "eval_steps_per_second": 3.234,
      "step": 3348
    },
    {
      "epoch": 7.0,
      "step": 3906,
      "train_exact_match": 78.82117882117882,
      "train_f1": 92.66952734441121,
      "train_runtime": 11.5936,
      "train_samples_per_second": 89.705,
      "train_steps_per_second": 3.278
    },
    {
      "epoch": 7.0,
      "grad_norm": 47.05348587036133,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.6345,
      "step": 3906
    },
    {
      "epoch": 7.0,
      "eval_exact_match": 58.90625,
      "eval_f1": 76.05687183756119,
      "eval_runtime": 37.0467,
      "eval_samples_per_second": 90.669,
      "eval_steps_per_second": 3.239,
      "step": 3906
    },
    {
      "epoch": 8.0,
      "step": 4464,
      "train_exact_match": 80.21978021978022,
      "train_f1": 92.08409867123048,
      "train_runtime": 11.6489,
      "train_samples_per_second": 89.708,
      "train_steps_per_second": 3.262
    },
    {
      "epoch": 8.0,
      "grad_norm": 45.41628646850586,
      "learning_rate": 2.5e-06,
      "loss": 0.5799,
      "step": 4464
    },
    {
      "epoch": 8.0,
      "eval_exact_match": 58.9375,
      "eval_f1": 75.91207274041706,
      "eval_runtime": 37.0443,
      "eval_samples_per_second": 90.675,
      "eval_steps_per_second": 3.239,
      "step": 4464
    },
    {
      "epoch": 9.0,
      "step": 5022,
      "train_exact_match": 83.11688311688312,
      "train_f1": 93.78525979268544,
      "train_runtime": 11.634,
      "train_samples_per_second": 89.737,
      "train_steps_per_second": 3.266
    },
    {
      "epoch": 9.0,
      "grad_norm": 66.75804901123047,
      "learning_rate": 1.25e-06,
      "loss": 0.5267,
      "step": 5022
    },
    {
      "epoch": 9.0,
      "eval_exact_match": 58.96875,
      "eval_f1": 76.09384754933019,
      "eval_runtime": 37.1002,
      "eval_samples_per_second": 90.538,
      "eval_steps_per_second": 3.234,
      "step": 5022
    },
    {
      "epoch": 10.0,
      "step": 5580,
      "train_exact_match": 81.71828171828172,
      "train_f1": 93.01859252339878,
      "train_runtime": 11.6538,
      "train_samples_per_second": 89.842,
      "train_steps_per_second": 3.261
    },
    {
      "epoch": 10.0,
      "grad_norm": 77.05963897705078,
      "learning_rate": 0.0,
      "loss": 0.4995,
      "step": 5580
    },
    {
      "epoch": 10.0,
      "eval_exact_match": 58.71875,
      "eval_f1": 76.0502313865696,
      "eval_runtime": 37.0653,
      "eval_samples_per_second": 90.624,
      "eval_steps_per_second": 3.238,
      "step": 5580
    },
    {
      "epoch": 10.0,
      "step": 5580,
      "total_flos": 3.056976081243648e+16,
      "train_loss": 0.8922852929775006,
      "train_runtime": 3581.679,
      "train_samples_per_second": 43.552,
      "train_steps_per_second": 1.558
    }
  ],
  "logging_steps": 500,
  "max_steps": 5580,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 3.056976081243648e+16,
  "train_batch_size": 28,
  "trial_name": null,
  "trial_params": null
}