trainer_state.json

{
  "best_metric": 1.7336630821228027,
  "best_model_checkpoint": "/users/zyong2/data/zyong2/bigscience/data/processed/024/bloom-350m_az_continual-pretrain-reinit_100000samples_-1vocab_original/checkpoint-25000",
  "epoch": 2.194518203164432,
  "global_step": 25000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.22,
      "learning_rate": 9e-05,
      "loss": 2.9009,
      "step": 2500
    },
    {
      "epoch": 0.44,
      "learning_rate": 8e-05,
      "loss": 2.1899,
      "step": 5000
    },
    {
      "epoch": 0.44,
      "eval_loss": 2.127051830291748,
      "eval_runtime": 85.5613,
      "eval_samples_per_second": 9.479,
      "eval_steps_per_second": 4.745,
      "step": 5000
    },
    {
      "epoch": 0.66,
      "learning_rate": 7e-05,
      "loss": 1.9997,
      "step": 7500
    },
    {
      "epoch": 0.88,
      "learning_rate": 6e-05,
      "loss": 1.8925,
      "step": 10000
    },
    {
      "epoch": 0.88,
      "eval_loss": 1.907801866531372,
      "eval_runtime": 85.4903,
      "eval_samples_per_second": 9.486,
      "eval_steps_per_second": 4.749,
      "step": 10000
    },
    {
      "epoch": 1.1,
      "learning_rate": 5e-05,
      "loss": 1.7805,
      "step": 12500
    },
    {
      "epoch": 1.32,
      "learning_rate": 4e-05,
      "loss": 1.6941,
      "step": 15000
    },
    {
      "epoch": 1.32,
      "eval_loss": 1.8138355016708374,
      "eval_runtime": 85.5471,
      "eval_samples_per_second": 9.48,
      "eval_steps_per_second": 4.746,
      "step": 15000
    },
    {
      "epoch": 1.54,
      "learning_rate": 3e-05,
      "loss": 1.6636,
      "step": 17500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2e-05,
      "loss": 1.6343,
      "step": 20000
    },
    {
      "epoch": 1.76,
      "eval_loss": 1.7566851377487183,
      "eval_runtime": 85.1744,
      "eval_samples_per_second": 9.522,
      "eval_steps_per_second": 4.767,
      "step": 20000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1e-05,
      "loss": 1.6134,
      "step": 22500
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0,
      "loss": 1.4868,
      "step": 25000
    },
    {
      "epoch": 2.19,
      "eval_loss": 1.7336630821228027,
      "eval_runtime": 85.4669,
      "eval_samples_per_second": 9.489,
      "eval_steps_per_second": 4.75,
      "step": 25000
    },
    {
      "epoch": 2.19,
      "step": 25000,
      "total_flos": 3.714902240494879e+17,
      "train_loss": 1.885576103515625,
      "train_runtime": 68721.1156,
      "train_samples_per_second": 2.91,
      "train_steps_per_second": 0.364
    }
  ],
  "max_steps": 25000,
  "num_train_epochs": 3,
  "total_flos": 3.714902240494879e+17,
  "trial_name": null,
  "trial_params": null
}