checkpoint-200/trainer_state.json

{
  "best_metric": 0.531296968460083,
  "best_model_checkpoint": "./output_v2/7b_cluster05_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_05/checkpoint-200",
  "epoch": 0.5947955390334573,
  "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": 0.6832,
      "step": 10
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.5389,
      "step": 20
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.5072,
      "step": 30
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.5211,
      "step": 40
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.5888,
      "step": 50
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.534,
      "step": 60
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.5684,
      "step": 70
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.514,
      "step": 80
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.5518,
      "step": 90
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.5034,
      "step": 100
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.542,
      "step": 110
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.4847,
      "step": 120
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.4772,
      "step": 130
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.5196,
      "step": 140
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.4672,
      "step": 150
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.4913,
      "step": 160
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.5498,
      "step": 170
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.5328,
      "step": 180
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.5313,
      "step": 190
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.515,
      "step": 200
    },
    {
      "epoch": 0.59,
      "eval_loss": 0.531296968460083,
      "eval_runtime": 174.4771,
      "eval_samples_per_second": 5.731,
      "eval_steps_per_second": 2.866,
      "step": 200
    },
    {
      "epoch": 0.59,
      "mmlu_eval_accuracy": 0.46292469330066577,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.5,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "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": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.65,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "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": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.5454545454545454,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3857738420406147,
      "step": 200
    }
  ],
  "max_steps": 5000,
  "num_train_epochs": 15,
  "total_flos": 4.043892293546803e+16,
  "trial_name": null,
  "trial_params": null
}