trainer_state-mistral-bnb8-QLORA-super_glue-copa-sequence_classification.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.0,
  "eval_steps": 1,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.8,
      "grad_norm": 156.40821838378906,
      "learning_rate": 2.5e-05,
      "loss": 1.7794,
      "step": 1
    },
    {
      "epoch": 0.8,
      "eval_accuracy": 0.54,
      "eval_loss": 1.8211030960083008,
      "eval_runtime": 4.6065,
      "eval_samples_per_second": 21.709,
      "eval_steps_per_second": 1.085,
      "step": 1
    },
    {
      "epoch": 1.6,
      "grad_norm": 189.74441528320312,
      "learning_rate": 5e-05,
      "loss": 1.6888,
      "step": 2
    },
    {
      "epoch": 1.6,
      "eval_accuracy": 0.45,
      "eval_loss": 6.309699058532715,
      "eval_runtime": 4.5954,
      "eval_samples_per_second": 21.761,
      "eval_steps_per_second": 1.088,
      "step": 2
    },
    {
      "epoch": 2.4,
      "grad_norm": 369.6210021972656,
      "learning_rate": 4.375e-05,
      "loss": 5.8817,
      "step": 3
    },
    {
      "epoch": 2.4,
      "eval_accuracy": 0.5,
      "eval_loss": 2.501729726791382,
      "eval_runtime": 4.59,
      "eval_samples_per_second": 21.786,
      "eval_steps_per_second": 1.089,
      "step": 3
    },
    {
      "epoch": 3.2,
      "grad_norm": 340.3645935058594,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 2.7331,
      "step": 4
    },
    {
      "epoch": 3.2,
      "eval_accuracy": 0.8,
      "eval_loss": 1.4943816661834717,
      "eval_runtime": 4.6115,
      "eval_samples_per_second": 21.685,
      "eval_steps_per_second": 1.084,
      "step": 4
    },
    {
      "epoch": 4.0,
      "grad_norm": 227.17994689941406,
      "learning_rate": 3.125e-05,
      "loss": 1.9356,
      "step": 5
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.87,
      "eval_loss": 0.731078028678894,
      "eval_runtime": 4.6084,
      "eval_samples_per_second": 21.699,
      "eval_steps_per_second": 1.085,
      "step": 5
    },
    {
      "epoch": 4.8,
      "grad_norm": 109.35914611816406,
      "learning_rate": 2.5e-05,
      "loss": 0.9016,
      "step": 6
    },
    {
      "epoch": 4.8,
      "eval_accuracy": 0.96,
      "eval_loss": 0.1448785364627838,
      "eval_runtime": 4.6033,
      "eval_samples_per_second": 21.724,
      "eval_steps_per_second": 1.086,
      "step": 6
    },
    {
      "epoch": 5.6,
      "grad_norm": 29.45122718811035,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.2567,
      "step": 7
    },
    {
      "epoch": 5.6,
      "eval_accuracy": 0.93,
      "eval_loss": 0.23716884851455688,
      "eval_runtime": 4.6029,
      "eval_samples_per_second": 21.725,
      "eval_steps_per_second": 1.086,
      "step": 7
    },
    {
      "epoch": 6.4,
      "grad_norm": 64.86052703857422,
      "learning_rate": 1.25e-05,
      "loss": 0.4689,
      "step": 8
    },
    {
      "epoch": 6.4,
      "eval_accuracy": 0.95,
      "eval_loss": 0.16914618015289307,
      "eval_runtime": 4.6114,
      "eval_samples_per_second": 21.686,
      "eval_steps_per_second": 1.084,
      "step": 8
    },
    {
      "epoch": 7.2,
      "grad_norm": 20.433061599731445,
      "learning_rate": 6.25e-06,
      "loss": 0.1207,
      "step": 9
    },
    {
      "epoch": 7.2,
      "eval_accuracy": 0.95,
      "eval_loss": 0.1778692603111267,
      "eval_runtime": 4.6066,
      "eval_samples_per_second": 21.708,
      "eval_steps_per_second": 1.085,
      "step": 9
    },
    {
      "epoch": 8.0,
      "grad_norm": 16.223140716552734,
      "learning_rate": 0.0,
      "loss": 0.1186,
      "step": 10
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.95,
      "eval_loss": 0.17223526537418365,
      "eval_runtime": 4.6175,
      "eval_samples_per_second": 21.657,
      "eval_steps_per_second": 1.083,
      "step": 10
    },
    {
      "epoch": 8.0,
      "step": 10,
      "total_flos": 7663110512640000.0,
      "train_loss": 1.588520422577858,
      "train_runtime": 555.6053,
      "train_samples_per_second": 7.199,
      "train_steps_per_second": 0.018
    }
  ],
  "logging_steps": 1,
  "max_steps": 10,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 7663110512640000.0,
  "train_batch_size": 20,
  "trial_name": null,
  "trial_params": null
}