trainer_state.json

{
  "best_metric": 82.66116908888208,
  "best_model_checkpoint": "/root/turkic_qa/tr_uzn_models/tr_uzn_xlm_roberta_large_model/checkpoint-3248",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 4060,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "step": 812,
      "train_exact_match": 61.63836163836164,
      "train_f1": 79.31791325882773,
      "train_runtime": 38.4548,
      "train_samples_per_second": 43.168,
      "train_steps_per_second": 1.56
    },
    {
      "epoch": 1.0,
      "grad_norm": 30.129222869873047,
      "learning_rate": 1e-05,
      "loss": 2.7411,
      "step": 812
    },
    {
      "epoch": 1.0,
      "eval_exact_match": 59.46875,
      "eval_f1": 77.35390101597599,
      "eval_runtime": 118.1334,
      "eval_samples_per_second": 43.265,
      "eval_steps_per_second": 1.549,
      "step": 812
    },
    {
      "epoch": 2.0,
      "step": 1624,
      "train_exact_match": 75.62437562437563,
      "train_f1": 87.86513543878338,
      "train_runtime": 37.8364,
      "train_samples_per_second": 43.001,
      "train_steps_per_second": 1.559
    },
    {
      "epoch": 2.0,
      "grad_norm": 113.6387710571289,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.8103,
      "step": 1624
    },
    {
      "epoch": 2.0,
      "eval_exact_match": 66.59375,
      "eval_f1": 81.70014057851566,
      "eval_runtime": 118.1729,
      "eval_samples_per_second": 43.25,
      "eval_steps_per_second": 1.549,
      "step": 1624
    },
    {
      "epoch": 3.0,
      "step": 2436,
      "train_exact_match": 81.31868131868131,
      "train_f1": 91.60071635098949,
      "train_runtime": 36.5906,
      "train_samples_per_second": 43.153,
      "train_steps_per_second": 1.558
    },
    {
      "epoch": 3.0,
      "grad_norm": 58.30486297607422,
      "learning_rate": 5e-06,
      "loss": 0.5305,
      "step": 2436
    },
    {
      "epoch": 3.0,
      "eval_exact_match": 66.9375,
      "eval_f1": 82.51782698770623,
      "eval_runtime": 118.3438,
      "eval_samples_per_second": 43.188,
      "eval_steps_per_second": 1.546,
      "step": 2436
    },
    {
      "epoch": 4.0,
      "step": 3248,
      "train_exact_match": 83.61638361638362,
      "train_f1": 93.04689171723578,
      "train_runtime": 36.8939,
      "train_samples_per_second": 43.178,
      "train_steps_per_second": 1.545
    },
    {
      "epoch": 4.0,
      "grad_norm": 11.345389366149902,
      "learning_rate": 2.5e-06,
      "loss": 0.3795,
      "step": 3248
    },
    {
      "epoch": 4.0,
      "eval_exact_match": 67.8125,
      "eval_f1": 82.66116908888208,
      "eval_runtime": 118.298,
      "eval_samples_per_second": 43.204,
      "eval_steps_per_second": 1.547,
      "step": 3248
    },
    {
      "epoch": 5.0,
      "step": 4060,
      "train_exact_match": 86.91308691308691,
      "train_f1": 93.86544501027228,
      "train_runtime": 36.964,
      "train_samples_per_second": 43.177,
      "train_steps_per_second": 1.542
    },
    {
      "epoch": 5.0,
      "grad_norm": 37.001564025878906,
      "learning_rate": 0.0,
      "loss": 0.2878,
      "step": 4060
    },
    {
      "epoch": 5.0,
      "eval_exact_match": 67.5625,
      "eval_f1": 82.48067936529503,
      "eval_runtime": 118.5494,
      "eval_samples_per_second": 43.113,
      "eval_steps_per_second": 1.544,
      "step": 4060
    },
    {
      "epoch": 5.0,
      "step": 4060,
      "total_flos": 7.910493982198272e+16,
      "train_loss": 0.9498688984387027,
      "train_runtime": 7336.9641,
      "train_samples_per_second": 15.479,
      "train_steps_per_second": 0.553
    }
  ],
  "logging_steps": 500,
  "max_steps": 4060,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 7.910493982198272e+16,
  "train_batch_size": 28,
  "trial_name": null,
  "trial_params": null
}