checkpoint-500/trainer_state.json

{
  "best_metric": 0.48825809359550476,
  "best_model_checkpoint": "mikhail-panzo/zlm_b32_le4_s8000/checkpoint-500",
  "epoch": 0.2094240837696335,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.020942408376963352,
      "grad_norm": 3.1775436401367188,
      "learning_rate": 2.35e-06,
      "loss": 1.1228,
      "step": 50
    },
    {
      "epoch": 0.041884816753926704,
      "grad_norm": 4.18980073928833,
      "learning_rate": 4.85e-06,
      "loss": 0.8304,
      "step": 100
    },
    {
      "epoch": 0.06282722513089005,
      "grad_norm": 4.5355024337768555,
      "learning_rate": 7.35e-06,
      "loss": 0.7701,
      "step": 150
    },
    {
      "epoch": 0.08376963350785341,
      "grad_norm": 3.9590957164764404,
      "learning_rate": 9.85e-06,
      "loss": 0.7096,
      "step": 200
    },
    {
      "epoch": 0.10471204188481675,
      "grad_norm": 5.419675350189209,
      "learning_rate": 1.235e-05,
      "loss": 0.6518,
      "step": 250
    },
    {
      "epoch": 0.1256544502617801,
      "grad_norm": 3.1052777767181396,
      "learning_rate": 1.485e-05,
      "loss": 0.6326,
      "step": 300
    },
    {
      "epoch": 0.14659685863874344,
      "grad_norm": 3.4649453163146973,
      "learning_rate": 1.7349999999999998e-05,
      "loss": 0.6241,
      "step": 350
    },
    {
      "epoch": 0.16753926701570682,
      "grad_norm": 4.962624549865723,
      "learning_rate": 1.985e-05,
      "loss": 0.5723,
      "step": 400
    },
    {
      "epoch": 0.18848167539267016,
      "grad_norm": 2.7586095333099365,
      "learning_rate": 2.235e-05,
      "loss": 0.5764,
      "step": 450
    },
    {
      "epoch": 0.2094240837696335,
      "grad_norm": 3.5610299110412598,
      "learning_rate": 2.485e-05,
      "loss": 0.5552,
      "step": 500
    },
    {
      "epoch": 0.2094240837696335,
      "eval_loss": 0.48825809359550476,
      "eval_runtime": 277.5892,
      "eval_samples_per_second": 30.581,
      "eval_steps_per_second": 3.826,
      "step": 500
    }
  ],
  "logging_steps": 50,
  "max_steps": 8000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2218281463842048.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}