{
  "best_metric": 0.5742304921150208,
  "best_model_checkpoint": "./output_v2/7b_cluster024_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_024/checkpoint-200",
  "epoch": 0.33167495854063017,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.6328,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.616,
      "step": 20
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.6065,
      "step": 30
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.5889,
      "step": 40
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.6014,
      "step": 50
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.5784,
      "step": 60
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.5916,
      "step": 70
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.5858,
      "step": 80
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.5908,
      "step": 90
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.5695,
      "step": 100
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.5748,
      "step": 110
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.5461,
      "step": 120
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.5738,
      "step": 130
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.544,
      "step": 140
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.5646,
      "step": 150
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.5767,
      "step": 160
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.5783,
      "step": 170
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.5553,
      "step": 180
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.5455,
      "step": 190
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.5544,
      "step": 200
    },
    {
      "epoch": 0.33,
      "eval_loss": 0.5742304921150208,
      "eval_runtime": 241.4438,
      "eval_samples_per_second": 4.142,
      "eval_steps_per_second": 2.071,
      "step": 200
    },
    {
      "epoch": 0.33,
      "mmlu_eval_accuracy": 0.4702447044631963,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.75,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.5555555555555556,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.171094535030522,
      "step": 200
    }
  ],
  "max_steps": 5000,
  "num_train_epochs": 9,
  "total_flos": 5.118582424190976e+16,
  "trial_name": null,
  "trial_params": null
}