{
  "best_metric": 0.7607901692390442,
  "best_model_checkpoint": "experts/expert-21/checkpoint-8000",
  "epoch": 1.596806387225549,
  "global_step": 8000,
  "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": 0.9541,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0002,
      "loss": 0.8555,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8687,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8501,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.863,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.9269,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8752,
      "step": 70
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.9716,
      "step": 80
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.859,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.8005,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.9085,
      "step": 110
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.8209,
      "step": 120
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.9187,
      "step": 130
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8224,
      "step": 140
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8602,
      "step": 150
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.7885,
      "step": 160
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.821,
      "step": 170
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.8676,
      "step": 180
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.9202,
      "step": 190
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.8188,
      "step": 200
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.823621928691864,
      "eval_runtime": 189.8157,
      "eval_samples_per_second": 5.268,
      "eval_steps_per_second": 2.634,
      "step": 200
    },
    {
      "epoch": 0.04,
      "mmlu_eval_accuracy": 0.486383698866548,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "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": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "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": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "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": 0.7368421052631579,
      "mmlu_loss": 1.6085881815722343,
      "step": 200
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.859,
      "step": 210
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.9367,
      "step": 220
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.8414,
      "step": 230
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.8446,
      "step": 240
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.8391,
      "step": 250
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7457,
      "step": 260
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7707,
      "step": 270
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.7892,
      "step": 280
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8066,
      "step": 290
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8116,
      "step": 300
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8572,
      "step": 310
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.7915,
      "step": 320
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.9248,
      "step": 330
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8565,
      "step": 340
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8508,
      "step": 350
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8534,
      "step": 360
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8604,
      "step": 370
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.803,
      "step": 380
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.9193,
      "step": 390
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.8465,
      "step": 400
    },
    {
      "epoch": 0.08,
      "eval_loss": 0.8140458464622498,
      "eval_runtime": 187.0783,
      "eval_samples_per_second": 5.345,
      "eval_steps_per_second": 2.673,
      "step": 400
    },
    {
      "epoch": 0.08,
      "mmlu_eval_accuracy": 0.49107726483887787,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.5687837348594367,
      "step": 400
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.9084,
      "step": 410
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.8347,
      "step": 420
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.8994,
      "step": 430
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.8385,
      "step": 440
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.7844,
      "step": 450
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.7386,
      "step": 460
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.8053,
      "step": 470
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.8232,
      "step": 480
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.784,
      "step": 490
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.9117,
      "step": 500
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.8685,
      "step": 510
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.8039,
      "step": 520
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.7753,
      "step": 530
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.8229,
      "step": 540
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.8525,
      "step": 550
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.8225,
      "step": 560
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.8656,
      "step": 570
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.8633,
      "step": 580
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.7918,
      "step": 590
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.8414,
      "step": 600
    },
    {
      "epoch": 0.12,
      "eval_loss": 0.806236207485199,
      "eval_runtime": 187.2336,
      "eval_samples_per_second": 5.341,
      "eval_steps_per_second": 2.67,
      "step": 600
    },
    {
      "epoch": 0.12,
      "mmlu_eval_accuracy": 0.4972022603401932,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.21,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.5916847097188935,
      "step": 600
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.7856,
      "step": 610
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.7942,
      "step": 620
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.7534,
      "step": 630
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.7279,
      "step": 640
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.8715,
      "step": 650
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.8721,
      "step": 660
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.8938,
      "step": 670
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.85,
      "step": 680
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.8078,
      "step": 690
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.8544,
      "step": 700
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.7725,
      "step": 710
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.8107,
      "step": 720
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.83,
      "step": 730
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.8263,
      "step": 740
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.7419,
      "step": 750
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.7672,
      "step": 760
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.846,
      "step": 770
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8475,
      "step": 780
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.7958,
      "step": 790
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8119,
      "step": 800
    },
    {
      "epoch": 0.16,
      "eval_loss": 0.8003740906715393,
      "eval_runtime": 187.0872,
      "eval_samples_per_second": 5.345,
      "eval_steps_per_second": 2.673,
      "step": 800
    },
    {
      "epoch": 0.16,
      "mmlu_eval_accuracy": 0.48984142289466515,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "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": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.538144465759591,
      "step": 800
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8485,
      "step": 810
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8256,
      "step": 820
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.8165,
      "step": 830
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.7499,
      "step": 840
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.7682,
      "step": 850
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.8159,
      "step": 860
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.8239,
      "step": 870
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.8509,
      "step": 880
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.7904,
      "step": 890
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.8335,
      "step": 900
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.744,
      "step": 910
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.8277,
      "step": 920
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.7853,
      "step": 930
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.9298,
      "step": 940
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.8082,
      "step": 950
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.8226,
      "step": 960
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.7895,
      "step": 970
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.8203,
      "step": 980
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.7988,
      "step": 990
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.8613,
      "step": 1000
    },
    {
      "epoch": 0.2,
      "eval_loss": 0.7977481484413147,
      "eval_runtime": 188.4501,
      "eval_samples_per_second": 5.306,
      "eval_steps_per_second": 2.653,
      "step": 1000
    },
    {
      "epoch": 0.2,
      "mmlu_eval_accuracy": 0.5024144702256793,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.8833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.22,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4293943774295539,
      "step": 1000
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.7364,
      "step": 1010
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.8136,
      "step": 1020
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.7847,
      "step": 1030
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.7593,
      "step": 1040
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.8041,
      "step": 1050
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.7329,
      "step": 1060
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.8265,
      "step": 1070
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.9067,
      "step": 1080
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.8517,
      "step": 1090
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.8549,
      "step": 1100
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.8212,
      "step": 1110
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.8496,
      "step": 1120
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.8076,
      "step": 1130
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.7832,
      "step": 1140
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.8627,
      "step": 1150
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.7884,
      "step": 1160
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.8796,
      "step": 1170
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.8647,
      "step": 1180
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.7339,
      "step": 1190
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.8742,
      "step": 1200
    },
    {
      "epoch": 0.24,
      "eval_loss": 0.7946733236312866,
      "eval_runtime": 187.11,
      "eval_samples_per_second": 5.344,
      "eval_steps_per_second": 2.672,
      "step": 1200
    },
    {
      "epoch": 0.24,
      "mmlu_eval_accuracy": 0.49322285882727945,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3325223162497926,
      "step": 1200
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.8767,
      "step": 1210
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.83,
      "step": 1220
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.8712,
      "step": 1230
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.7577,
      "step": 1240
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.7647,
      "step": 1250
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.9173,
      "step": 1260
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.8709,
      "step": 1270
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.8505,
      "step": 1280
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.8167,
      "step": 1290
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.8331,
      "step": 1300
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.9096,
      "step": 1310
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.7396,
      "step": 1320
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.78,
      "step": 1330
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.7839,
      "step": 1340
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.8077,
      "step": 1350
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.7606,
      "step": 1360
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.7988,
      "step": 1370
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.8694,
      "step": 1380
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.7781,
      "step": 1390
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.7748,
      "step": 1400
    },
    {
      "epoch": 0.28,
      "eval_loss": 0.7932688593864441,
      "eval_runtime": 187.13,
      "eval_samples_per_second": 5.344,
      "eval_steps_per_second": 2.672,
      "step": 1400
    },
    {
      "epoch": 0.28,
      "mmlu_eval_accuracy": 0.5079295799335525,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "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": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.4589004895550153,
      "step": 1400
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.8362,
      "step": 1410
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.8236,
      "step": 1420
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.8807,
      "step": 1430
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.8217,
      "step": 1440
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.813,
      "step": 1450
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.7989,
      "step": 1460
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.8456,
      "step": 1470
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.8417,
      "step": 1480
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.8405,
      "step": 1490
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.8141,
      "step": 1500
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.8759,
      "step": 1510
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.7641,
      "step": 1520
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.8134,
      "step": 1530
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.7795,
      "step": 1540
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.8466,
      "step": 1550
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.7989,
      "step": 1560
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.7817,
      "step": 1570
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.8158,
      "step": 1580
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.7773,
      "step": 1590
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.7918,
      "step": 1600
    },
    {
      "epoch": 0.32,
      "eval_loss": 0.7885581254959106,
      "eval_runtime": 187.1149,
      "eval_samples_per_second": 5.344,
      "eval_steps_per_second": 2.672,
      "step": 1600
    },
    {
      "epoch": 0.32,
      "mmlu_eval_accuracy": 0.5053062702770444,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "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": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4281705232341981,
      "step": 1600
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.8081,
      "step": 1610
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.7738,
      "step": 1620
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.651,
      "step": 1630
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7823,
      "step": 1640
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.8122,
      "step": 1650
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7635,
      "step": 1660
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.826,
      "step": 1670
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.7419,
      "step": 1680
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.7926,
      "step": 1690
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.8044,
      "step": 1700
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.7626,
      "step": 1710
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.8479,
      "step": 1720
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.8001,
      "step": 1730
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.7427,
      "step": 1740
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.8344,
      "step": 1750
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.814,
      "step": 1760
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.8108,
      "step": 1770
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7679,
      "step": 1780
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7813,
      "step": 1790
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.8284,
      "step": 1800
    },
    {
      "epoch": 0.36,
      "eval_loss": 0.7852458357810974,
      "eval_runtime": 187.1697,
      "eval_samples_per_second": 5.343,
      "eval_steps_per_second": 2.671,
      "step": 1800
    },
    {
      "epoch": 0.36,
      "mmlu_eval_accuracy": 0.5047854061909083,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "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": 0.3333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "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": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.565987547835233,
      "step": 1800
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.8043,
      "step": 1810
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7553,
      "step": 1820
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.8699,
      "step": 1830
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.7521,
      "step": 1840
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.8021,
      "step": 1850
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.8268,
      "step": 1860
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.8167,
      "step": 1870
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.8541,
      "step": 1880
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.8217,
      "step": 1890
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.7941,
      "step": 1900
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.7909,
      "step": 1910
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.7878,
      "step": 1920
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.8283,
      "step": 1930
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.7582,
      "step": 1940
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.8038,
      "step": 1950
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.7747,
      "step": 1960
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.8603,
      "step": 1970
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.7707,
      "step": 1980
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.8831,
      "step": 1990
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.8092,
      "step": 2000
    },
    {
      "epoch": 0.4,
      "eval_loss": 0.7837955355644226,
      "eval_runtime": 187.2749,
      "eval_samples_per_second": 5.34,
      "eval_steps_per_second": 2.67,
      "step": 2000
    },
    {
      "epoch": 0.4,
      "mmlu_eval_accuracy": 0.49875980342964915,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5384615384615384,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.4707468855863453,
      "step": 2000
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.7431,
      "step": 2010
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.8298,
      "step": 2020
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.7946,
      "step": 2030
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.8223,
      "step": 2040
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.8185,
      "step": 2050
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.8171,
      "step": 2060
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.7983,
      "step": 2070
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.8583,
      "step": 2080
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.7071,
      "step": 2090
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.8242,
      "step": 2100
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.8486,
      "step": 2110
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.6957,
      "step": 2120
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.8166,
      "step": 2130
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.8096,
      "step": 2140
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.7978,
      "step": 2150
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.8999,
      "step": 2160
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.8312,
      "step": 2170
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.7875,
      "step": 2180
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.819,
      "step": 2190
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.8239,
      "step": 2200
    },
    {
      "epoch": 0.44,
      "eval_loss": 0.7816848158836365,
      "eval_runtime": 187.3461,
      "eval_samples_per_second": 5.338,
      "eval_steps_per_second": 2.669,
      "step": 2200
    },
    {
      "epoch": 0.44,
      "mmlu_eval_accuracy": 0.4955980019742852,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "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": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.22,
      "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.4989356849118873,
      "step": 2200
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.8019,
      "step": 2210
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.8282,
      "step": 2220
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.7252,
      "step": 2230
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.8095,
      "step": 2240
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.8038,
      "step": 2250
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.769,
      "step": 2260
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.8117,
      "step": 2270
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.8,
      "step": 2280
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.774,
      "step": 2290
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.8632,
      "step": 2300
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.8727,
      "step": 2310
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.7947,
      "step": 2320
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.8335,
      "step": 2330
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7785,
      "step": 2340
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.8432,
      "step": 2350
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7994,
      "step": 2360
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.8162,
      "step": 2370
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.8022,
      "step": 2380
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.7795,
      "step": 2390
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.7224,
      "step": 2400
    },
    {
      "epoch": 0.48,
      "eval_loss": 0.780394434928894,
      "eval_runtime": 187.2883,
      "eval_samples_per_second": 5.339,
      "eval_steps_per_second": 2.67,
      "step": 2400
    },
    {
      "epoch": 0.48,
      "mmlu_eval_accuracy": 0.5026329344652324,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "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": 0.4090909090909091,
      "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": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.5,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.4715563680966903,
      "step": 2400
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.8695,
      "step": 2410
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.7877,
      "step": 2420
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.8548,
      "step": 2430
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.8298,
      "step": 2440
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.8107,
      "step": 2450
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.7209,
      "step": 2460
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.7808,
      "step": 2470
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.7566,
      "step": 2480
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.8135,
      "step": 2490
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.8477,
      "step": 2500
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.758,
      "step": 2510
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.8049,
      "step": 2520
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.8744,
      "step": 2530
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7463,
      "step": 2540
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.82,
      "step": 2550
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.8455,
      "step": 2560
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7105,
      "step": 2570
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.6995,
      "step": 2580
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.8515,
      "step": 2590
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.8331,
      "step": 2600
    },
    {
      "epoch": 0.52,
      "eval_loss": 0.7785139083862305,
      "eval_runtime": 186.9246,
      "eval_samples_per_second": 5.35,
      "eval_steps_per_second": 2.675,
      "step": 2600
    },
    {
      "epoch": 0.52,
      "mmlu_eval_accuracy": 0.4939589560188444,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.16129032258064516,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.4832373786688473,
      "step": 2600
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.7971,
      "step": 2610
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.7771,
      "step": 2620
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.8641,
      "step": 2630
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.8352,
      "step": 2640
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7577,
      "step": 2650
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7748,
      "step": 2660
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.8012,
      "step": 2670
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.8073,
      "step": 2680
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7752,
      "step": 2690
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.8225,
      "step": 2700
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7684,
      "step": 2710
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7912,
      "step": 2720
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7707,
      "step": 2730
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.8007,
      "step": 2740
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.8413,
      "step": 2750
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.8439,
      "step": 2760
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.7243,
      "step": 2770
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.8112,
      "step": 2780
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.874,
      "step": 2790
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.8683,
      "step": 2800
    },
    {
      "epoch": 0.56,
      "eval_loss": 0.7752290964126587,
      "eval_runtime": 187.3026,
      "eval_samples_per_second": 5.339,
      "eval_steps_per_second": 2.669,
      "step": 2800
    },
    {
      "epoch": 0.56,
      "mmlu_eval_accuracy": 0.4937077937991497,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "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": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4585867143642186,
      "step": 2800
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.8228,
      "step": 2810
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.8865,
      "step": 2820
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.7906,
      "step": 2830
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.7763,
      "step": 2840
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.7077,
      "step": 2850
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.8119,
      "step": 2860
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.8127,
      "step": 2870
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.743,
      "step": 2880
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.8813,
      "step": 2890
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.7725,
      "step": 2900
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.715,
      "step": 2910
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.8224,
      "step": 2920
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.7893,
      "step": 2930
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.808,
      "step": 2940
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.816,
      "step": 2950
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.7345,
      "step": 2960
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.8386,
      "step": 2970
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.818,
      "step": 2980
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.8089,
      "step": 2990
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.9051,
      "step": 3000
    },
    {
      "epoch": 0.6,
      "eval_loss": 0.7735825181007385,
      "eval_runtime": 187.1525,
      "eval_samples_per_second": 5.343,
      "eval_steps_per_second": 2.672,
      "step": 3000
    },
    {
      "epoch": 0.6,
      "mmlu_eval_accuracy": 0.49383161171900514,
      "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3979128985118616,
      "step": 3000
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.7805,
      "step": 3010
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.8022,
      "step": 3020
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.8525,
      "step": 3030
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.8669,
      "step": 3040
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.7706,
      "step": 3050
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.7221,
      "step": 3060
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.8801,
      "step": 3070
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.6372,
      "step": 3080
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.881,
      "step": 3090
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.8141,
      "step": 3100
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.8995,
      "step": 3110
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.8088,
      "step": 3120
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.8005,
      "step": 3130
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.6721,
      "step": 3140
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.8626,
      "step": 3150
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.7597,
      "step": 3160
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.7343,
      "step": 3170
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.7306,
      "step": 3180
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.803,
      "step": 3190
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.7498,
      "step": 3200
    },
    {
      "epoch": 0.64,
      "eval_loss": 0.7715298533439636,
      "eval_runtime": 187.1506,
      "eval_samples_per_second": 5.343,
      "eval_steps_per_second": 2.672,
      "step": 3200
    },
    {
      "epoch": 0.64,
      "mmlu_eval_accuracy": 0.49445891887306603,
      "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": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3431594996243794,
      "step": 3200
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.7824,
      "step": 3210
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.7516,
      "step": 3220
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.8491,
      "step": 3230
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7538,
      "step": 3240
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7618,
      "step": 3250
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.8253,
      "step": 3260
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7631,
      "step": 3270
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.8339,
      "step": 3280
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.6797,
      "step": 3290
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.797,
      "step": 3300
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.8091,
      "step": 3310
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.8087,
      "step": 3320
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.8036,
      "step": 3330
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.9307,
      "step": 3340
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.7692,
      "step": 3350
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.8246,
      "step": 3360
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.7682,
      "step": 3370
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.7337,
      "step": 3380
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.7281,
      "step": 3390
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.8049,
      "step": 3400
    },
    {
      "epoch": 0.68,
      "eval_loss": 0.7722234725952148,
      "eval_runtime": 187.0936,
      "eval_samples_per_second": 5.345,
      "eval_steps_per_second": 2.672,
      "step": 3400
    },
    {
      "epoch": 0.68,
      "mmlu_eval_accuracy": 0.4885770768529298,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.36470588235294116,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3427304202987382,
      "step": 3400
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.7956,
      "step": 3410
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.6748,
      "step": 3420
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.7901,
      "step": 3430
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7958,
      "step": 3440
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.8239,
      "step": 3450
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7486,
      "step": 3460
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7866,
      "step": 3470
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.762,
      "step": 3480
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.8271,
      "step": 3490
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7367,
      "step": 3500
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7989,
      "step": 3510
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7986,
      "step": 3520
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7712,
      "step": 3530
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.8605,
      "step": 3540
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.7465,
      "step": 3550
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.8129,
      "step": 3560
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.8049,
      "step": 3570
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.8579,
      "step": 3580
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.8272,
      "step": 3590
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.6876,
      "step": 3600
    },
    {
      "epoch": 0.72,
      "eval_loss": 0.7700828909873962,
      "eval_runtime": 187.1333,
      "eval_samples_per_second": 5.344,
      "eval_steps_per_second": 2.672,
      "step": 3600
    },
    {
      "epoch": 0.72,
      "mmlu_eval_accuracy": 0.4956696905096371,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "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": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "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": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.278034775126385,
      "step": 3600
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.7432,
      "step": 3610
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.7804,
      "step": 3620
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.8165,
      "step": 3630
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.7751,
      "step": 3640
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.6947,
      "step": 3650
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.7461,
      "step": 3660
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.7294,
      "step": 3670
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.6992,
      "step": 3680
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.8659,
      "step": 3690
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.8452,
      "step": 3700
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.7714,
      "step": 3710
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.7721,
      "step": 3720
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.7642,
      "step": 3730
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.8614,
      "step": 3740
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7979,
      "step": 3750
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7714,
      "step": 3760
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.8341,
      "step": 3770
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7559,
      "step": 3780
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7893,
      "step": 3790
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7029,
      "step": 3800
    },
    {
      "epoch": 0.76,
      "eval_loss": 0.7690045833587646,
      "eval_runtime": 187.1512,
      "eval_samples_per_second": 5.343,
      "eval_steps_per_second": 2.672,
      "step": 3800
    },
    {
      "epoch": 0.76,
      "mmlu_eval_accuracy": 0.4986652426405072,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.09090909090909091,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.2290096188027304,
      "step": 3800
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7765,
      "step": 3810
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7342,
      "step": 3820
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.8179,
      "step": 3830
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7516,
      "step": 3840
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7402,
      "step": 3850
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7458,
      "step": 3860
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7975,
      "step": 3870
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7606,
      "step": 3880
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.7841,
      "step": 3890
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.8398,
      "step": 3900
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.8645,
      "step": 3910
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.8749,
      "step": 3920
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.7782,
      "step": 3930
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7548,
      "step": 3940
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7762,
      "step": 3950
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.6853,
      "step": 3960
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7392,
      "step": 3970
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.8183,
      "step": 3980
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7711,
      "step": 3990
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7947,
      "step": 4000
    },
    {
      "epoch": 0.8,
      "eval_loss": 0.7685290575027466,
      "eval_runtime": 187.1502,
      "eval_samples_per_second": 5.343,
      "eval_steps_per_second": 2.672,
      "step": 4000
    },
    {
      "epoch": 0.8,
      "mmlu_eval_accuracy": 0.49482145136513617,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "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": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "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": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "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": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4409859191344239,
      "step": 4000
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.8158,
      "step": 4010
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7975,
      "step": 4020
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.8133,
      "step": 4030
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.7486,
      "step": 4040
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.8209,
      "step": 4050
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.8269,
      "step": 4060
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.8395,
      "step": 4070
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.7845,
      "step": 4080
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7609,
      "step": 4090
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.8267,
      "step": 4100
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.6963,
      "step": 4110
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.8173,
      "step": 4120
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7732,
      "step": 4130
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.8267,
      "step": 4140
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7965,
      "step": 4150
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7632,
      "step": 4160
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7595,
      "step": 4170
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.8642,
      "step": 4180
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.8094,
      "step": 4190
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.7524,
      "step": 4200
    },
    {
      "epoch": 0.84,
      "eval_loss": 0.7668033838272095,
      "eval_runtime": 187.1363,
      "eval_samples_per_second": 5.344,
      "eval_steps_per_second": 2.672,
      "step": 4200
    },
    {
      "epoch": 0.84,
      "mmlu_eval_accuracy": 0.5033620342118746,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5769230769230769,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.475248734882544,
      "step": 4200
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.7157,
      "step": 4210
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.758,
      "step": 4220
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.7123,
      "step": 4230
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.8488,
      "step": 4240
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.7902,
      "step": 4250
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.7876,
      "step": 4260
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.8171,
      "step": 4270
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.8478,
      "step": 4280
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.7818,
      "step": 4290
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.6867,
      "step": 4300
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.8269,
      "step": 4310
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.6898,
      "step": 4320
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.8375,
      "step": 4330
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.7823,
      "step": 4340
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.8884,
      "step": 4350
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.8705,
      "step": 4360
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.7379,
      "step": 4370
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.6632,
      "step": 4380
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7939,
      "step": 4390
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7487,
      "step": 4400
    },
    {
      "epoch": 0.88,
      "eval_loss": 0.7660865187644958,
      "eval_runtime": 187.2239,
      "eval_samples_per_second": 5.341,
      "eval_steps_per_second": 2.671,
      "step": 4400
    },
    {
      "epoch": 0.88,
      "mmlu_eval_accuracy": 0.49666140703774375,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.5,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.7209302325581395,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.5290965444114126,
      "step": 4400
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.8308,
      "step": 4410
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.8165,
      "step": 4420
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.8891,
      "step": 4430
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.8207,
      "step": 4440
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7602,
      "step": 4450
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7541,
      "step": 4460
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7926,
      "step": 4470
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7693,
      "step": 4480
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.7225,
      "step": 4490
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.9039,
      "step": 4500
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.87,
      "step": 4510
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.8597,
      "step": 4520
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.8691,
      "step": 4530
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7924,
      "step": 4540
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.8612,
      "step": 4550
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7521,
      "step": 4560
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7845,
      "step": 4570
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7594,
      "step": 4580
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.703,
      "step": 4590
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7429,
      "step": 4600
    },
    {
      "epoch": 0.92,
      "eval_loss": 0.7656491994857788,
      "eval_runtime": 187.0859,
      "eval_samples_per_second": 5.345,
      "eval_steps_per_second": 2.673,
      "step": 4600
    },
    {
      "epoch": 0.92,
      "mmlu_eval_accuracy": 0.49214003939461304,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "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": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.473437037815002,
      "step": 4600
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7824,
      "step": 4610
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.8594,
      "step": 4620
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7766,
      "step": 4630
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7882,
      "step": 4640
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7788,
      "step": 4650
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7974,
      "step": 4660
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.887,
      "step": 4670
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7948,
      "step": 4680
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.8338,
      "step": 4690
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.8122,
      "step": 4700
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7172,
      "step": 4710
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.8443,
      "step": 4720
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.753,
      "step": 4730
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.8285,
      "step": 4740
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.8201,
      "step": 4750
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.7593,
      "step": 4760
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.8614,
      "step": 4770
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.7837,
      "step": 4780
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.8251,
      "step": 4790
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.8472,
      "step": 4800
    },
    {
      "epoch": 0.96,
      "eval_loss": 0.7640769481658936,
      "eval_runtime": 187.1184,
      "eval_samples_per_second": 5.344,
      "eval_steps_per_second": 2.672,
      "step": 4800
    },
    {
      "epoch": 0.96,
      "mmlu_eval_accuracy": 0.4917410322972955,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.5,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.43259734448347,
      "step": 4800
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.6779,
      "step": 4810
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.7722,
      "step": 4820
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.6602,
      "step": 4830
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.9011,
      "step": 4840
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.8638,
      "step": 4850
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.8419,
      "step": 4860
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.6895,
      "step": 4870
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.7256,
      "step": 4880
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.8401,
      "step": 4890
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.8183,
      "step": 4900
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.74,
      "step": 4910
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.7646,
      "step": 4920
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.6902,
      "step": 4930
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7008,
      "step": 4940
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7706,
      "step": 4950
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7198,
      "step": 4960
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.8393,
      "step": 4970
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.879,
      "step": 4980
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.7068,
      "step": 4990
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.8118,
      "step": 5000
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.7644715309143066,
      "eval_runtime": 187.1418,
      "eval_samples_per_second": 5.344,
      "eval_steps_per_second": 2.672,
      "step": 5000
    },
    {
      "epoch": 1.0,
      "mmlu_eval_accuracy": 0.4897235305372632,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.6559623529643366,
      "step": 5000
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.8291,
      "step": 5010
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.6988,
      "step": 5020
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.788,
      "step": 5030
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.6987,
      "step": 5040
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.7461,
      "step": 5050
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.6927,
      "step": 5060
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.8146,
      "step": 5070
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.7272,
      "step": 5080
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.6673,
      "step": 5090
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.7504,
      "step": 5100
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.741,
      "step": 5110
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.6656,
      "step": 5120
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.7639,
      "step": 5130
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.6899,
      "step": 5140
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.78,
      "step": 5150
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.7191,
      "step": 5160
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.7769,
      "step": 5170
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.7312,
      "step": 5180
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.7441,
      "step": 5190
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.621,
      "step": 5200
    },
    {
      "epoch": 1.04,
      "eval_loss": 0.7662388682365417,
      "eval_runtime": 187.1909,
      "eval_samples_per_second": 5.342,
      "eval_steps_per_second": 2.671,
      "step": 5200
    },
    {
      "epoch": 1.04,
      "mmlu_eval_accuracy": 0.5020650841220764,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5384615384615384,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.5350869625184305,
      "step": 5200
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.6897,
      "step": 5210
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.8237,
      "step": 5220
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.7928,
      "step": 5230
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.677,
      "step": 5240
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.691,
      "step": 5250
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.7186,
      "step": 5260
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.7162,
      "step": 5270
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.7113,
      "step": 5280
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.7728,
      "step": 5290
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.7765,
      "step": 5300
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.7582,
      "step": 5310
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.6322,
      "step": 5320
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.6756,
      "step": 5330
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.7367,
      "step": 5340
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.7243,
      "step": 5350
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.7407,
      "step": 5360
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.7641,
      "step": 5370
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.6907,
      "step": 5380
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.7651,
      "step": 5390
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.7084,
      "step": 5400
    },
    {
      "epoch": 1.08,
      "eval_loss": 0.7681072950363159,
      "eval_runtime": 187.1791,
      "eval_samples_per_second": 5.342,
      "eval_steps_per_second": 2.671,
      "step": 5400
    },
    {
      "epoch": 1.08,
      "mmlu_eval_accuracy": 0.5015341575680828,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "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": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4263136799279454,
      "step": 5400
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.7107,
      "step": 5410
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.6685,
      "step": 5420
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.681,
      "step": 5430
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.7014,
      "step": 5440
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.7075,
      "step": 5450
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.6847,
      "step": 5460
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.7239,
      "step": 5470
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.7613,
      "step": 5480
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.7472,
      "step": 5490
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.6243,
      "step": 5500
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.7922,
      "step": 5510
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.6214,
      "step": 5520
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.6899,
      "step": 5530
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.7722,
      "step": 5540
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.67,
      "step": 5550
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.7355,
      "step": 5560
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.7009,
      "step": 5570
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.6981,
      "step": 5580
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6678,
      "step": 5590
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6483,
      "step": 5600
    },
    {
      "epoch": 1.12,
      "eval_loss": 0.7694364786148071,
      "eval_runtime": 187.1233,
      "eval_samples_per_second": 5.344,
      "eval_steps_per_second": 2.672,
      "step": 5600
    },
    {
      "epoch": 1.12,
      "mmlu_eval_accuracy": 0.4959247642842794,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.09090909090909091,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "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": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3448996527932642,
      "step": 5600
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.7608,
      "step": 5610
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.7287,
      "step": 5620
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.8166,
      "step": 5630
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.7145,
      "step": 5640
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.5995,
      "step": 5650
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.7108,
      "step": 5660
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.7644,
      "step": 5670
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.6972,
      "step": 5680
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.656,
      "step": 5690
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.6991,
      "step": 5700
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.7643,
      "step": 5710
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.6859,
      "step": 5720
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.7445,
      "step": 5730
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.6089,
      "step": 5740
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.7375,
      "step": 5750
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.6874,
      "step": 5760
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.7839,
      "step": 5770
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.7691,
      "step": 5780
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.8125,
      "step": 5790
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.7102,
      "step": 5800
    },
    {
      "epoch": 1.16,
      "eval_loss": 0.7666054368019104,
      "eval_runtime": 187.1926,
      "eval_samples_per_second": 5.342,
      "eval_steps_per_second": 2.671,
      "step": 5800
    },
    {
      "epoch": 1.16,
      "mmlu_eval_accuracy": 0.4953397086939019,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "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": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3072250700572763,
      "step": 5800
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.7693,
      "step": 5810
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.752,
      "step": 5820
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.7062,
      "step": 5830
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.6499,
      "step": 5840
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.7847,
      "step": 5850
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.7177,
      "step": 5860
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.8077,
      "step": 5870
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.7185,
      "step": 5880
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.6927,
      "step": 5890
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.6748,
      "step": 5900
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.7682,
      "step": 5910
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.7313,
      "step": 5920
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.657,
      "step": 5930
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6988,
      "step": 5940
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.7192,
      "step": 5950
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.7366,
      "step": 5960
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6799,
      "step": 5970
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6884,
      "step": 5980
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.7824,
      "step": 5990
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.7324,
      "step": 6000
    },
    {
      "epoch": 1.2,
      "eval_loss": 0.7687702178955078,
      "eval_runtime": 187.2259,
      "eval_samples_per_second": 5.341,
      "eval_steps_per_second": 2.671,
      "step": 6000
    },
    {
      "epoch": 1.2,
      "mmlu_eval_accuracy": 0.4887805391479623,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.2987300734140232,
      "step": 6000
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.694,
      "step": 6010
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.671,
      "step": 6020
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.7151,
      "step": 6030
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.7579,
      "step": 6040
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.7156,
      "step": 6050
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.6539,
      "step": 6060
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.7516,
      "step": 6070
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.7031,
      "step": 6080
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.6637,
      "step": 6090
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.7157,
      "step": 6100
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.7545,
      "step": 6110
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.7356,
      "step": 6120
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.7113,
      "step": 6130
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.6886,
      "step": 6140
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.7,
      "step": 6150
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.7022,
      "step": 6160
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.7017,
      "step": 6170
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.7389,
      "step": 6180
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.772,
      "step": 6190
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.7928,
      "step": 6200
    },
    {
      "epoch": 1.24,
      "eval_loss": 0.7661731839179993,
      "eval_runtime": 187.2036,
      "eval_samples_per_second": 5.342,
      "eval_steps_per_second": 2.671,
      "step": 6200
    },
    {
      "epoch": 1.24,
      "mmlu_eval_accuracy": 0.4945041167073197,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.09090909090909091,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.058823529411764705,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4803929476762876,
      "step": 6200
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.6577,
      "step": 6210
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.6931,
      "step": 6220
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.697,
      "step": 6230
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.7056,
      "step": 6240
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.692,
      "step": 6250
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.7575,
      "step": 6260
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.7492,
      "step": 6270
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.8006,
      "step": 6280
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.6728,
      "step": 6290
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.7139,
      "step": 6300
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.7133,
      "step": 6310
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.7331,
      "step": 6320
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.7015,
      "step": 6330
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.7085,
      "step": 6340
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.707,
      "step": 6350
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.813,
      "step": 6360
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.6732,
      "step": 6370
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.6956,
      "step": 6380
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.6881,
      "step": 6390
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.7021,
      "step": 6400
    },
    {
      "epoch": 1.28,
      "eval_loss": 0.765486478805542,
      "eval_runtime": 187.2431,
      "eval_samples_per_second": 5.341,
      "eval_steps_per_second": 2.67,
      "step": 6400
    },
    {
      "epoch": 1.28,
      "mmlu_eval_accuracy": 0.48791791045653865,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.058823529411764705,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3861944245918612,
      "step": 6400
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.7624,
      "step": 6410
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.6935,
      "step": 6420
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.662,
      "step": 6430
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.7622,
      "step": 6440
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.6905,
      "step": 6450
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.7081,
      "step": 6460
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.6841,
      "step": 6470
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.7377,
      "step": 6480
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.7325,
      "step": 6490
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.7371,
      "step": 6500
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.6946,
      "step": 6510
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.7349,
      "step": 6520
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.7146,
      "step": 6530
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.6544,
      "step": 6540
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.7181,
      "step": 6550
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.7402,
      "step": 6560
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.6383,
      "step": 6570
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.7457,
      "step": 6580
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.6756,
      "step": 6590
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.6816,
      "step": 6600
    },
    {
      "epoch": 1.32,
      "eval_loss": 0.7636769413948059,
      "eval_runtime": 187.168,
      "eval_samples_per_second": 5.343,
      "eval_steps_per_second": 2.671,
      "step": 6600
    },
    {
      "epoch": 1.32,
      "mmlu_eval_accuracy": 0.4920059183708129,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.4239774859527383,
      "step": 6600
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.7748,
      "step": 6610
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.6706,
      "step": 6620
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.7519,
      "step": 6630
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.7019,
      "step": 6640
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.6951,
      "step": 6650
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.6859,
      "step": 6660
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.7121,
      "step": 6670
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.7487,
      "step": 6680
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.6306,
      "step": 6690
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.6431,
      "step": 6700
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.7545,
      "step": 6710
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.7246,
      "step": 6720
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.7913,
      "step": 6730
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.7761,
      "step": 6740
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.7073,
      "step": 6750
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.754,
      "step": 6760
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.7558,
      "step": 6770
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.7042,
      "step": 6780
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.6929,
      "step": 6790
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.7915,
      "step": 6800
    },
    {
      "epoch": 1.36,
      "eval_loss": 0.7638018131256104,
      "eval_runtime": 187.2002,
      "eval_samples_per_second": 5.342,
      "eval_steps_per_second": 2.671,
      "step": 6800
    },
    {
      "epoch": 1.36,
      "mmlu_eval_accuracy": 0.4960624186488206,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3617745879113519,
      "step": 6800
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.8217,
      "step": 6810
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.6967,
      "step": 6820
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.6943,
      "step": 6830
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.6951,
      "step": 6840
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.6993,
      "step": 6850
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.713,
      "step": 6860
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.7332,
      "step": 6870
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.6572,
      "step": 6880
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.6926,
      "step": 6890
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.6644,
      "step": 6900
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.7057,
      "step": 6910
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.6722,
      "step": 6920
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.7249,
      "step": 6930
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.7689,
      "step": 6940
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.6632,
      "step": 6950
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.7049,
      "step": 6960
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.6287,
      "step": 6970
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.7653,
      "step": 6980
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.6594,
      "step": 6990
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.7705,
      "step": 7000
    },
    {
      "epoch": 1.4,
      "eval_loss": 0.7635005116462708,
      "eval_runtime": 187.2323,
      "eval_samples_per_second": 5.341,
      "eval_steps_per_second": 2.67,
      "step": 7000
    },
    {
      "epoch": 1.4,
      "mmlu_eval_accuracy": 0.499593792435714,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3253327236941215,
      "step": 7000
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.7633,
      "step": 7010
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.6171,
      "step": 7020
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.6963,
      "step": 7030
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.7332,
      "step": 7040
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.6447,
      "step": 7050
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.7448,
      "step": 7060
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.644,
      "step": 7070
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.7323,
      "step": 7080
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.66,
      "step": 7090
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.6856,
      "step": 7100
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.8723,
      "step": 7110
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.7134,
      "step": 7120
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.7198,
      "step": 7130
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.7441,
      "step": 7140
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.644,
      "step": 7150
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.624,
      "step": 7160
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.6627,
      "step": 7170
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.7031,
      "step": 7180
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.653,
      "step": 7190
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.7555,
      "step": 7200
    },
    {
      "epoch": 1.44,
      "eval_loss": 0.7617677450180054,
      "eval_runtime": 186.9164,
      "eval_samples_per_second": 5.35,
      "eval_steps_per_second": 2.675,
      "step": 7200
    },
    {
      "epoch": 1.44,
      "mmlu_eval_accuracy": 0.48545216173800443,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.2410852103141519,
      "step": 7200
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.7148,
      "step": 7210
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.776,
      "step": 7220
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.6795,
      "step": 7230
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.7791,
      "step": 7240
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.7769,
      "step": 7250
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.6923,
      "step": 7260
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.7276,
      "step": 7270
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.749,
      "step": 7280
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.6771,
      "step": 7290
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.7031,
      "step": 7300
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.6358,
      "step": 7310
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.6835,
      "step": 7320
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.645,
      "step": 7330
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.729,
      "step": 7340
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.757,
      "step": 7350
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.7158,
      "step": 7360
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.6721,
      "step": 7370
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.5802,
      "step": 7380
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.776,
      "step": 7390
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.7365,
      "step": 7400
    },
    {
      "epoch": 1.48,
      "eval_loss": 0.7627587914466858,
      "eval_runtime": 187.0956,
      "eval_samples_per_second": 5.345,
      "eval_steps_per_second": 2.672,
      "step": 7400
    },
    {
      "epoch": 1.48,
      "mmlu_eval_accuracy": 0.48769321030050283,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.2603290581749873,
      "step": 7400
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.7099,
      "step": 7410
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.6696,
      "step": 7420
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.7407,
      "step": 7430
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6573,
      "step": 7440
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6826,
      "step": 7450
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6908,
      "step": 7460
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.7449,
      "step": 7470
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6686,
      "step": 7480
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.6168,
      "step": 7490
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.7281,
      "step": 7500
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.7463,
      "step": 7510
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.7347,
      "step": 7520
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.6648,
      "step": 7530
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.6236,
      "step": 7540
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.7377,
      "step": 7550
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.7758,
      "step": 7560
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.7311,
      "step": 7570
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.714,
      "step": 7580
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.7496,
      "step": 7590
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.7344,
      "step": 7600
    },
    {
      "epoch": 1.52,
      "eval_loss": 0.7611469626426697,
      "eval_runtime": 187.1908,
      "eval_samples_per_second": 5.342,
      "eval_steps_per_second": 2.671,
      "step": 7600
    },
    {
      "epoch": 1.52,
      "mmlu_eval_accuracy": 0.4936462441359661,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.09090909090909091,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.10344827586206896,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.2213753110901806,
      "step": 7600
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.6931,
      "step": 7610
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.7574,
      "step": 7620
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.6972,
      "step": 7630
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.6504,
      "step": 7640
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.7215,
      "step": 7650
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.7029,
      "step": 7660
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.7427,
      "step": 7670
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.6842,
      "step": 7680
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.7486,
      "step": 7690
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.7101,
      "step": 7700
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.6632,
      "step": 7710
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.8137,
      "step": 7720
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.7323,
      "step": 7730
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.7583,
      "step": 7740
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.7165,
      "step": 7750
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.6297,
      "step": 7760
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.687,
      "step": 7770
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.6872,
      "step": 7780
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.7573,
      "step": 7790
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.8051,
      "step": 7800
    },
    {
      "epoch": 1.56,
      "eval_loss": 0.7618192434310913,
      "eval_runtime": 187.2153,
      "eval_samples_per_second": 5.341,
      "eval_steps_per_second": 2.671,
      "step": 7800
    },
    {
      "epoch": 1.56,
      "mmlu_eval_accuracy": 0.5004751739502825,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.5,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.10344827586206896,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5384615384615384,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.8181818181818182,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1960963902717783,
      "step": 7800
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.7347,
      "step": 7810
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.7379,
      "step": 7820
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.7469,
      "step": 7830
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.7462,
      "step": 7840
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.6847,
      "step": 7850
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.735,
      "step": 7860
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.727,
      "step": 7870
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.7069,
      "step": 7880
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.664,
      "step": 7890
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.7443,
      "step": 7900
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.6675,
      "step": 7910
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.7169,
      "step": 7920
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.6731,
      "step": 7930
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.6599,
      "step": 7940
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.6804,
      "step": 7950
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.6881,
      "step": 7960
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.7459,
      "step": 7970
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.7848,
      "step": 7980
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.6993,
      "step": 7990
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.6961,
      "step": 8000
    },
    {
      "epoch": 1.6,
      "eval_loss": 0.7607901692390442,
      "eval_runtime": 187.1646,
      "eval_samples_per_second": 5.343,
      "eval_steps_per_second": 2.671,
      "step": 8000
    },
    {
      "epoch": 1.6,
      "mmlu_eval_accuracy": 0.5047661692655837,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.5,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.09090909090909091,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "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": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "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": 0.10344827586206896,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
      "mmlu_eval_accuracy_management": 0.8181818181818182,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3588235294117647,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.395272626422404,
      "step": 8000
    }
  ],
  "max_steps": 10000,
  "num_train_epochs": 2,
  "total_flos": 1.1501069235904512e+18,
  "trial_name": null,
  "trial_params": null
}