trainer_state.json

{
  "best_metric": 0.9905983805656433,
  "best_model_checkpoint": "th_cl_13epochs_lora_pos_neg/checkpoint-96",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 96,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.3125,
      "grad_norm": 28.00575065612793,
      "learning_rate": 8.958333333333335e-05,
      "loss": 0.4959,
      "step": 10
    },
    {
      "epoch": 0.625,
      "grad_norm": 26.350143432617188,
      "learning_rate": 7.916666666666666e-05,
      "loss": 0.6218,
      "step": 20
    },
    {
      "epoch": 0.9375,
      "grad_norm": 31.33298110961914,
      "learning_rate": 6.875e-05,
      "loss": 0.7442,
      "step": 30
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5111111111111111,
      "eval_balanced_accuracy": 0.515,
      "eval_loss": 1.0410274267196655,
      "eval_runtime": 86.3882,
      "eval_samples_per_second": 0.521,
      "eval_steps_per_second": 0.069,
      "step": 32
    },
    {
      "epoch": 1.25,
      "grad_norm": 32.04572677612305,
      "learning_rate": 5.833333333333334e-05,
      "loss": 0.3933,
      "step": 40
    },
    {
      "epoch": 1.5625,
      "grad_norm": 27.043235778808594,
      "learning_rate": 4.791666666666667e-05,
      "loss": 0.3407,
      "step": 50
    },
    {
      "epoch": 1.875,
      "grad_norm": 20.865083694458008,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.4596,
      "step": 60
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5111111111111111,
      "eval_balanced_accuracy": 0.5118577075098814,
      "eval_loss": 1.0013209581375122,
      "eval_runtime": 87.3168,
      "eval_samples_per_second": 0.515,
      "eval_steps_per_second": 0.069,
      "step": 64
    },
    {
      "epoch": 2.1875,
      "grad_norm": 7.867964744567871,
      "learning_rate": 2.7083333333333332e-05,
      "loss": 0.317,
      "step": 70
    },
    {
      "epoch": 2.5,
      "grad_norm": 21.569629669189453,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.2957,
      "step": 80
    },
    {
      "epoch": 2.8125,
      "grad_norm": 14.822030067443848,
      "learning_rate": 6.25e-06,
      "loss": 0.2864,
      "step": 90
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.5111111111111111,
      "eval_balanced_accuracy": 0.5118577075098814,
      "eval_loss": 0.9905983805656433,
      "eval_runtime": 86.9479,
      "eval_samples_per_second": 0.518,
      "eval_steps_per_second": 0.069,
      "step": 96
    },
    {
      "epoch": 3.0,
      "step": 96,
      "total_flos": 1.6176865332953088e+16,
      "train_loss": 0.4266343352695306,
      "train_runtime": 4808.092,
      "train_samples_per_second": 0.157,
      "train_steps_per_second": 0.02
    }
  ],
  "logging_steps": 10,
  "max_steps": 96,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.6176865332953088e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}