trainer_state-gemma-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": 167.4876708984375,
      "learning_rate": 2.5e-05,
      "loss": 1.0597,
      "step": 1
    },
    {
      "epoch": 0.8,
      "eval_accuracy": 0.45,
      "eval_loss": 1.2138588428497314,
      "eval_runtime": 5.8912,
      "eval_samples_per_second": 16.974,
      "eval_steps_per_second": 0.849,
      "step": 1
    },
    {
      "epoch": 1.6,
      "grad_norm": 158.76622009277344,
      "learning_rate": 5e-05,
      "loss": 1.057,
      "step": 2
    },
    {
      "epoch": 1.6,
      "eval_accuracy": 0.55,
      "eval_loss": 1.4417834281921387,
      "eval_runtime": 5.9115,
      "eval_samples_per_second": 16.916,
      "eval_steps_per_second": 0.846,
      "step": 2
    },
    {
      "epoch": 2.4,
      "grad_norm": 256.3013610839844,
      "learning_rate": 4.375e-05,
      "loss": 1.5118,
      "step": 3
    },
    {
      "epoch": 2.4,
      "eval_accuracy": 0.55,
      "eval_loss": 1.3326141834259033,
      "eval_runtime": 5.8623,
      "eval_samples_per_second": 17.058,
      "eval_steps_per_second": 0.853,
      "step": 3
    },
    {
      "epoch": 3.2,
      "grad_norm": 250.87896728515625,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 1.4061,
      "step": 4
    },
    {
      "epoch": 3.2,
      "eval_accuracy": 0.8,
      "eval_loss": 0.46391355991363525,
      "eval_runtime": 5.8855,
      "eval_samples_per_second": 16.991,
      "eval_steps_per_second": 0.85,
      "step": 4
    },
    {
      "epoch": 4.0,
      "grad_norm": 95.89608001708984,
      "learning_rate": 3.125e-05,
      "loss": 0.4755,
      "step": 5
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.78,
      "eval_loss": 0.5776048898696899,
      "eval_runtime": 5.8627,
      "eval_samples_per_second": 17.057,
      "eval_steps_per_second": 0.853,
      "step": 5
    },
    {
      "epoch": 4.8,
      "grad_norm": 125.85784912109375,
      "learning_rate": 2.5e-05,
      "loss": 0.5092,
      "step": 6
    },
    {
      "epoch": 4.8,
      "eval_accuracy": 0.82,
      "eval_loss": 0.4626426696777344,
      "eval_runtime": 5.8521,
      "eval_samples_per_second": 17.088,
      "eval_steps_per_second": 0.854,
      "step": 6
    },
    {
      "epoch": 5.6,
      "grad_norm": 86.73672485351562,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.3827,
      "step": 7
    },
    {
      "epoch": 5.6,
      "eval_accuracy": 0.9,
      "eval_loss": 0.26935383677482605,
      "eval_runtime": 5.8986,
      "eval_samples_per_second": 16.953,
      "eval_steps_per_second": 0.848,
      "step": 7
    },
    {
      "epoch": 6.4,
      "grad_norm": 28.821346282958984,
      "learning_rate": 1.25e-05,
      "loss": 0.195,
      "step": 8
    },
    {
      "epoch": 6.4,
      "eval_accuracy": 0.93,
      "eval_loss": 0.2368904948234558,
      "eval_runtime": 5.8651,
      "eval_samples_per_second": 17.05,
      "eval_steps_per_second": 0.852,
      "step": 8
    },
    {
      "epoch": 7.2,
      "grad_norm": 5.473712921142578,
      "learning_rate": 6.25e-06,
      "loss": 0.1953,
      "step": 9
    },
    {
      "epoch": 7.2,
      "eval_accuracy": 0.94,
      "eval_loss": 0.24502474069595337,
      "eval_runtime": 5.9017,
      "eval_samples_per_second": 16.944,
      "eval_steps_per_second": 0.847,
      "step": 9
    },
    {
      "epoch": 8.0,
      "grad_norm": 20.565353393554688,
      "learning_rate": 0.0,
      "loss": 0.177,
      "step": 10
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.94,
      "eval_loss": 0.2417210191488266,
      "eval_runtime": 5.8796,
      "eval_samples_per_second": 17.008,
      "eval_steps_per_second": 0.85,
      "step": 10
    },
    {
      "epoch": 8.0,
      "step": 10,
      "total_flos": 7831941212282880.0,
      "train_loss": 0.6969200998544693,
      "train_runtime": 664.8566,
      "train_samples_per_second": 6.016,
      "train_steps_per_second": 0.015
    }
  ],
  "logging_steps": 1,
  "max_steps": 10,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 7831941212282880.0,
  "train_batch_size": 20,
  "trial_name": null,
  "trial_params": null
}