trainer_state.json

{
  "best_metric": 3.5487985610961914,
  "best_model_checkpoint": "/users/zyong2/data/zyong2/bigscience/data/processed/024/bloom-350m_az_bitfit_1000samples_-1vocab_original-frozen/checkpoint-25000",
  "epoch": 181.15855855855855,
  "global_step": 25000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 18.12,
      "learning_rate": 9e-05,
      "loss": 4.2544,
      "step": 2500
    },
    {
      "epoch": 36.23,
      "learning_rate": 8e-05,
      "loss": 3.9123,
      "step": 5000
    },
    {
      "epoch": 36.23,
      "eval_loss": 3.843057632446289,
      "eval_runtime": 513.4541,
      "eval_samples_per_second": 9.555,
      "eval_steps_per_second": 4.777,
      "step": 5000
    },
    {
      "epoch": 54.35,
      "learning_rate": 7e-05,
      "loss": 3.7584,
      "step": 7500
    },
    {
      "epoch": 72.46,
      "learning_rate": 6e-05,
      "loss": 3.6674,
      "step": 10000
    },
    {
      "epoch": 72.46,
      "eval_loss": 3.6707117557525635,
      "eval_runtime": 512.9035,
      "eval_samples_per_second": 9.565,
      "eval_steps_per_second": 4.783,
      "step": 10000
    },
    {
      "epoch": 90.58,
      "learning_rate": 5e-05,
      "loss": 3.6049,
      "step": 12500
    },
    {
      "epoch": 108.69,
      "learning_rate": 4e-05,
      "loss": 3.5633,
      "step": 15000
    },
    {
      "epoch": 108.69,
      "eval_loss": 3.59525203704834,
      "eval_runtime": 512.888,
      "eval_samples_per_second": 9.565,
      "eval_steps_per_second": 4.783,
      "step": 15000
    },
    {
      "epoch": 126.81,
      "learning_rate": 3e-05,
      "loss": 3.5333,
      "step": 17500
    },
    {
      "epoch": 144.92,
      "learning_rate": 2e-05,
      "loss": 3.5125,
      "step": 20000
    },
    {
      "epoch": 144.92,
      "eval_loss": 3.559532403945923,
      "eval_runtime": 512.8641,
      "eval_samples_per_second": 9.566,
      "eval_steps_per_second": 4.783,
      "step": 20000
    },
    {
      "epoch": 163.04,
      "learning_rate": 1e-05,
      "loss": 3.501,
      "step": 22500
    },
    {
      "epoch": 181.16,
      "learning_rate": 0.0,
      "loss": 3.493,
      "step": 25000
    },
    {
      "epoch": 181.16,
      "eval_loss": 3.5487985610961914,
      "eval_runtime": 512.9062,
      "eval_samples_per_second": 9.565,
      "eval_steps_per_second": 4.783,
      "step": 25000
    },
    {
      "epoch": 181.16,
      "step": 25000,
      "total_flos": 3.7349994596715725e+17,
      "train_loss": 3.6800425,
      "train_runtime": 52880.0792,
      "train_samples_per_second": 3.782,
      "train_steps_per_second": 0.473
    }
  ],
  "max_steps": 25000,
  "num_train_epochs": 182,
  "total_flos": 3.7349994596715725e+17,
  "trial_name": null,
  "trial_params": null
}