checkpoint-200/trainer_state.json

{
  "best_metric": 0.32424649596214294,
  "best_model_checkpoint": "./output_v2/7b_cluster03_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_03/checkpoint-200",
  "epoch": 0.6866952789699571,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8627,
      "step": 10
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.4107,
      "step": 20
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.3691,
      "step": 30
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.3979,
      "step": 40
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.3495,
      "step": 50
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.3693,
      "step": 60
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.357,
      "step": 70
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.3456,
      "step": 80
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.3122,
      "step": 90
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.3131,
      "step": 100
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.3678,
      "step": 110
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.3364,
      "step": 120
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.324,
      "step": 130
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.353,
      "step": 140
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.3473,
      "step": 150
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.363,
      "step": 160
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.3192,
      "step": 170
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.3125,
      "step": 180
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.3001,
      "step": 190
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.3164,
      "step": 200
    },
    {
      "epoch": 0.69,
      "eval_loss": 0.32424649596214294,
      "eval_runtime": 166.2846,
      "eval_samples_per_second": 6.014,
      "eval_steps_per_second": 3.007,
      "step": 200
    },
    {
      "epoch": 0.69,
      "mmlu_eval_accuracy": 0.46206132856742516,
      "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 0.954094658943754,
      "step": 200
    }
  ],
  "max_steps": 5000,
  "num_train_epochs": 18,
  "total_flos": 9578945836154880.0,
  "trial_name": null,
  "trial_params": null
}