{
  "best_metric": 0.7774001359939575,
  "best_model_checkpoint": "experts/expert-16/checkpoint-200",
  "epoch": 0.06337135614702155,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0002,
      "loss": 0.8331,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8289,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.9038,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8491,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.8155,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.7896,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.7835,
      "step": 70
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8827,
      "step": 80
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.861,
      "step": 90
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.7879,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.803,
      "step": 110
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.8212,
      "step": 120
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.8075,
      "step": 130
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.9263,
      "step": 140
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7969,
      "step": 150
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7883,
      "step": 160
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7582,
      "step": 170
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8095,
      "step": 180
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8614,
      "step": 190
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8674,
      "step": 200
    },
    {
      "epoch": 0.06,
      "eval_loss": 0.7774001359939575,
      "eval_runtime": 149.8878,
      "eval_samples_per_second": 6.672,
      "eval_steps_per_second": 3.336,
      "step": 200
    },
    {
      "epoch": 0.06,
      "mmlu_eval_accuracy": 0.4759538024775667,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.619096915108728,
      "step": 200
    }
  ],
  "max_steps": 10000,
  "num_train_epochs": 4,
  "total_flos": 6.125001405058253e+16,
  "trial_name": null,
  "trial_params": null
}