trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 125,
  "global_step": 1250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2,
      "grad_norm": 4.678056716918945,
      "learning_rate": 2.25e-05,
      "loss": 0.6949,
      "step": 125
    },
    {
      "epoch": 0.2,
      "eval_accuracy": 0.669,
      "eval_loss": 0.6145870685577393,
      "eval_runtime": 881.9981,
      "eval_samples_per_second": 5.669,
      "eval_steps_per_second": 0.709,
      "step": 125
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.371523380279541,
      "learning_rate": 2e-05,
      "loss": 0.5479,
      "step": 250
    },
    {
      "epoch": 0.4,
      "eval_accuracy": 0.6992,
      "eval_loss": 0.5730966329574585,
      "eval_runtime": 870.806,
      "eval_samples_per_second": 5.742,
      "eval_steps_per_second": 0.718,
      "step": 250
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.348880767822266,
      "learning_rate": 1.75e-05,
      "loss": 0.5114,
      "step": 375
    },
    {
      "epoch": 0.6,
      "eval_accuracy": 0.7038,
      "eval_loss": 0.5597503781318665,
      "eval_runtime": 872.0256,
      "eval_samples_per_second": 5.734,
      "eval_steps_per_second": 0.717,
      "step": 375
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.614448547363281,
      "learning_rate": 1.5e-05,
      "loss": 0.5035,
      "step": 500
    },
    {
      "epoch": 0.8,
      "eval_accuracy": 0.7102,
      "eval_loss": 0.553583025932312,
      "eval_runtime": 871.7948,
      "eval_samples_per_second": 5.735,
      "eval_steps_per_second": 0.717,
      "step": 500
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.813840389251709,
      "learning_rate": 1.25e-05,
      "loss": 0.4973,
      "step": 625
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7166,
      "eval_loss": 0.5452517867088318,
      "eval_runtime": 872.2804,
      "eval_samples_per_second": 5.732,
      "eval_steps_per_second": 0.717,
      "step": 625
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.915520191192627,
      "learning_rate": 1e-05,
      "loss": 0.4803,
      "step": 750
    },
    {
      "epoch": 1.2,
      "eval_accuracy": 0.7174,
      "eval_loss": 0.5457369089126587,
      "eval_runtime": 872.4916,
      "eval_samples_per_second": 5.731,
      "eval_steps_per_second": 0.716,
      "step": 750
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.6513495445251465,
      "learning_rate": 7.5e-06,
      "loss": 0.4632,
      "step": 875
    },
    {
      "epoch": 1.4,
      "eval_accuracy": 0.7198,
      "eval_loss": 0.5453068614006042,
      "eval_runtime": 871.2609,
      "eval_samples_per_second": 5.739,
      "eval_steps_per_second": 0.717,
      "step": 875
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.503047466278076,
      "learning_rate": 5e-06,
      "loss": 0.4632,
      "step": 1000
    },
    {
      "epoch": 1.6,
      "eval_accuracy": 0.7258,
      "eval_loss": 0.544837236404419,
      "eval_runtime": 872.099,
      "eval_samples_per_second": 5.733,
      "eval_steps_per_second": 0.717,
      "step": 1000
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.498978137969971,
      "learning_rate": 2.5e-06,
      "loss": 0.4779,
      "step": 1125
    },
    {
      "epoch": 1.8,
      "eval_accuracy": 0.724,
      "eval_loss": 0.5422948598861694,
      "eval_runtime": 871.4244,
      "eval_samples_per_second": 5.738,
      "eval_steps_per_second": 0.717,
      "step": 1125
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.752194404602051,
      "learning_rate": 0.0,
      "loss": 0.4596,
      "step": 1250
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.723,
      "eval_loss": 0.5430094599723816,
      "eval_runtime": 872.4628,
      "eval_samples_per_second": 5.731,
      "eval_steps_per_second": 0.716,
      "step": 1250
    }
  ],
  "logging_steps": 125,
  "max_steps": 1250,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 125,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 12,
  "trial_name": null,
  "trial_params": null
}