{
  "best_metric": 0.7293602228164673,
  "best_model_checkpoint": "experts/expert-16/checkpoint-6200",
  "epoch": 2.091254752851711,
  "global_step": 6600,
  "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": 0.8339,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8289,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.9041,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8491,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.8151,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.79,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.7835,
      "step": 70
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8831,
      "step": 80
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8607,
      "step": 90
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.7876,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8031,
      "step": 110
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.8207,
      "step": 120
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.807,
      "step": 130
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.9262,
      "step": 140
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7964,
      "step": 150
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7879,
      "step": 160
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7587,
      "step": 170
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8091,
      "step": 180
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8615,
      "step": 190
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8672,
      "step": 200
    },
    {
      "epoch": 0.06,
      "eval_loss": 0.7779108881950378,
      "eval_runtime": 110.9863,
      "eval_samples_per_second": 9.01,
      "eval_steps_per_second": 4.505,
      "step": 200
    },
    {
      "epoch": 0.06,
      "mmlu_eval_accuracy": 0.4744171116325413,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "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": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "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": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "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": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "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": 0.7368421052631579,
      "mmlu_loss": 1.5868234255450824,
      "step": 200
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8316,
      "step": 210
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8454,
      "step": 220
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8434,
      "step": 230
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.821,
      "step": 240
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.7893,
      "step": 250
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.8242,
      "step": 260
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.8128,
      "step": 270
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.8344,
      "step": 280
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.8338,
      "step": 290
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.7981,
      "step": 300
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.781,
      "step": 310
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.7717,
      "step": 320
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.767,
      "step": 330
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.7925,
      "step": 340
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.8226,
      "step": 350
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.7912,
      "step": 360
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.8093,
      "step": 370
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.7648,
      "step": 380
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.7866,
      "step": 390
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.7976,
      "step": 400
    },
    {
      "epoch": 0.13,
      "eval_loss": 0.7656086683273315,
      "eval_runtime": 110.9802,
      "eval_samples_per_second": 9.011,
      "eval_steps_per_second": 4.505,
      "step": 400
    },
    {
      "epoch": 0.13,
      "mmlu_eval_accuracy": 0.47124130233512024,
      "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": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "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": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.4339068503199297,
      "step": 400
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.8182,
      "step": 410
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.8438,
      "step": 420
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.8184,
      "step": 430
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.8202,
      "step": 440
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.8264,
      "step": 450
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.8384,
      "step": 460
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.8372,
      "step": 470
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.8072,
      "step": 480
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8214,
      "step": 490
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.814,
      "step": 500
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.847,
      "step": 510
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8444,
      "step": 520
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.8096,
      "step": 530
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.8496,
      "step": 540
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.7729,
      "step": 550
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.7826,
      "step": 560
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.7478,
      "step": 570
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.7953,
      "step": 580
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.7363,
      "step": 590
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.7971,
      "step": 600
    },
    {
      "epoch": 0.19,
      "eval_loss": 0.7616064548492432,
      "eval_runtime": 110.9404,
      "eval_samples_per_second": 9.014,
      "eval_steps_per_second": 4.507,
      "step": 600
    },
    {
      "epoch": 0.19,
      "mmlu_eval_accuracy": 0.4749850916074463,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "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": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "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": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "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": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.5555555555555556,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.5647042619341658,
      "step": 600
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.7936,
      "step": 610
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.7319,
      "step": 620
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.79,
      "step": 630
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.7806,
      "step": 640
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.8833,
      "step": 650
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.7711,
      "step": 660
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.8242,
      "step": 670
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.7948,
      "step": 680
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.7417,
      "step": 690
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.7275,
      "step": 700
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.8137,
      "step": 710
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.8568,
      "step": 720
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.802,
      "step": 730
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.8202,
      "step": 740
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.8077,
      "step": 750
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.814,
      "step": 760
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.7971,
      "step": 770
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.798,
      "step": 780
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.7806,
      "step": 790
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.8042,
      "step": 800
    },
    {
      "epoch": 0.25,
      "eval_loss": 0.7563537359237671,
      "eval_runtime": 111.023,
      "eval_samples_per_second": 9.007,
      "eval_steps_per_second": 4.504,
      "step": 800
    },
    {
      "epoch": 0.25,
      "mmlu_eval_accuracy": 0.4796267144005645,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "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": 0.2727272727272727,
      "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": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "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": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "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": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "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": 0.6842105263157895,
      "mmlu_loss": 1.4866046660796157,
      "step": 800
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.8119,
      "step": 810
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.8156,
      "step": 820
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.8288,
      "step": 830
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.8008,
      "step": 840
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.8649,
      "step": 850
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.8242,
      "step": 860
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.8255,
      "step": 870
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.8467,
      "step": 880
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.8264,
      "step": 890
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.7833,
      "step": 900
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.8338,
      "step": 910
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.8062,
      "step": 920
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.8112,
      "step": 930
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.7469,
      "step": 940
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.7897,
      "step": 950
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.8081,
      "step": 960
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.7571,
      "step": 970
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.8161,
      "step": 980
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.7759,
      "step": 990
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.7417,
      "step": 1000
    },
    {
      "epoch": 0.32,
      "eval_loss": 0.754473865032196,
      "eval_runtime": 111.0233,
      "eval_samples_per_second": 9.007,
      "eval_steps_per_second": 4.504,
      "step": 1000
    },
    {
      "epoch": 0.32,
      "mmlu_eval_accuracy": 0.4749030525395577,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "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": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.88,
      "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": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.29411764705882354,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.5362318840579711,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.596783688734468,
      "step": 1000
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.827,
      "step": 1010
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.8345,
      "step": 1020
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7883,
      "step": 1030
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7774,
      "step": 1040
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.8175,
      "step": 1050
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.8,
      "step": 1060
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.8049,
      "step": 1070
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.8116,
      "step": 1080
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.7852,
      "step": 1090
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.7429,
      "step": 1100
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.794,
      "step": 1110
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.7549,
      "step": 1120
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7347,
      "step": 1130
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7482,
      "step": 1140
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7393,
      "step": 1150
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.8103,
      "step": 1160
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.8075,
      "step": 1170
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.7831,
      "step": 1180
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.792,
      "step": 1190
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.7955,
      "step": 1200
    },
    {
      "epoch": 0.38,
      "eval_loss": 0.7498393654823303,
      "eval_runtime": 110.9719,
      "eval_samples_per_second": 9.011,
      "eval_steps_per_second": 4.506,
      "step": 1200
    },
    {
      "epoch": 0.38,
      "mmlu_eval_accuracy": 0.4769718071089565,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "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": 0.8833333333333333,
      "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": 0.3333333333333333,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.6994269322166244,
      "step": 1200
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.7821,
      "step": 1210
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.8138,
      "step": 1220
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.717,
      "step": 1230
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.7406,
      "step": 1240
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.8031,
      "step": 1250
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.7974,
      "step": 1260
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.8001,
      "step": 1270
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.766,
      "step": 1280
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.7679,
      "step": 1290
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.8017,
      "step": 1300
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.8027,
      "step": 1310
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.7819,
      "step": 1320
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.7558,
      "step": 1330
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.8363,
      "step": 1340
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.7809,
      "step": 1350
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.8114,
      "step": 1360
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.8446,
      "step": 1370
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.7877,
      "step": 1380
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.8309,
      "step": 1390
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.8131,
      "step": 1400
    },
    {
      "epoch": 0.44,
      "eval_loss": 0.7500243186950684,
      "eval_runtime": 111.0409,
      "eval_samples_per_second": 9.006,
      "eval_steps_per_second": 4.503,
      "step": 1400
    },
    {
      "epoch": 0.44,
      "mmlu_eval_accuracy": 0.4758070782649682,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.42857142857142855,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "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": 0.8833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "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": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "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": 1.4974891513041355,
      "step": 1400
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.8122,
      "step": 1410
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.7754,
      "step": 1420
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.8116,
      "step": 1430
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.7442,
      "step": 1440
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.7638,
      "step": 1450
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.7746,
      "step": 1460
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7616,
      "step": 1470
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.8144,
      "step": 1480
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7924,
      "step": 1490
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.8075,
      "step": 1500
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.769,
      "step": 1510
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.7296,
      "step": 1520
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.8284,
      "step": 1530
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.82,
      "step": 1540
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.7619,
      "step": 1550
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.7862,
      "step": 1560
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.7835,
      "step": 1570
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.7624,
      "step": 1580
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.8021,
      "step": 1590
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.793,
      "step": 1600
    },
    {
      "epoch": 0.51,
      "eval_loss": 0.7459111213684082,
      "eval_runtime": 111.062,
      "eval_samples_per_second": 9.004,
      "eval_steps_per_second": 4.502,
      "step": 1600
    },
    {
      "epoch": 0.51,
      "mmlu_eval_accuracy": 0.46888485292306403,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "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": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.88,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "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": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.6951039852112453,
      "step": 1600
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7999,
      "step": 1610
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7959,
      "step": 1620
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.7887,
      "step": 1630
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.7186,
      "step": 1640
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.8049,
      "step": 1650
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7934,
      "step": 1660
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.8369,
      "step": 1670
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7567,
      "step": 1680
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.8058,
      "step": 1690
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7818,
      "step": 1700
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7115,
      "step": 1710
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7434,
      "step": 1720
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.7788,
      "step": 1730
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.7824,
      "step": 1740
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.7198,
      "step": 1750
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.8059,
      "step": 1760
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.7892,
      "step": 1770
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.8048,
      "step": 1780
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.7938,
      "step": 1790
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.791,
      "step": 1800
    },
    {
      "epoch": 0.57,
      "eval_loss": 0.744739830493927,
      "eval_runtime": 111.1326,
      "eval_samples_per_second": 8.998,
      "eval_steps_per_second": 4.499,
      "step": 1800
    },
    {
      "epoch": 0.57,
      "mmlu_eval_accuracy": 0.4764276491893982,
      "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": 0.4482758620689655,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "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": 0.8636363636363636,
      "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": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.9,
      "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": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "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": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.7224162761754218,
      "step": 1800
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.7335,
      "step": 1810
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.7762,
      "step": 1820
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.75,
      "step": 1830
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.7875,
      "step": 1840
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.7749,
      "step": 1850
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.8516,
      "step": 1860
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.7729,
      "step": 1870
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.7664,
      "step": 1880
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.802,
      "step": 1890
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.7791,
      "step": 1900
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.8041,
      "step": 1910
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.7671,
      "step": 1920
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.7785,
      "step": 1930
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.782,
      "step": 1940
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.8032,
      "step": 1950
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.8065,
      "step": 1960
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.7713,
      "step": 1970
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.7709,
      "step": 1980
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.8036,
      "step": 1990
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.7614,
      "step": 2000
    },
    {
      "epoch": 0.63,
      "eval_loss": 0.7417653799057007,
      "eval_runtime": 111.078,
      "eval_samples_per_second": 9.003,
      "eval_steps_per_second": 4.501,
      "step": 2000
    },
    {
      "epoch": 0.63,
      "mmlu_eval_accuracy": 0.4656871532254676,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "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": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "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": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.76,
      "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": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "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": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.632609389158204,
      "step": 2000
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.8459,
      "step": 2010
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.7348,
      "step": 2020
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.811,
      "step": 2030
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7091,
      "step": 2040
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7715,
      "step": 2050
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.8017,
      "step": 2060
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.7734,
      "step": 2070
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.8292,
      "step": 2080
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.7873,
      "step": 2090
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.757,
      "step": 2100
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.7986,
      "step": 2110
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.7848,
      "step": 2120
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.7579,
      "step": 2130
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.7683,
      "step": 2140
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.7958,
      "step": 2150
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.8009,
      "step": 2160
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7504,
      "step": 2170
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7558,
      "step": 2180
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7143,
      "step": 2190
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7767,
      "step": 2200
    },
    {
      "epoch": 0.7,
      "eval_loss": 0.7396783232688904,
      "eval_runtime": 111.0434,
      "eval_samples_per_second": 9.005,
      "eval_steps_per_second": 4.503,
      "step": 2200
    },
    {
      "epoch": 0.7,
      "mmlu_eval_accuracy": 0.48937488654796385,
      "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": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.18181818181818182,
      "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": 0.4883720930232558,
      "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": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "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": 0.48148148148148145,
      "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": 0.7894736842105263,
      "mmlu_loss": 1.464440327400327,
      "step": 2200
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7848,
      "step": 2210
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7864,
      "step": 2220
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.7609,
      "step": 2230
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.7782,
      "step": 2240
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.7825,
      "step": 2250
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.85,
      "step": 2260
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.7802,
      "step": 2270
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.7715,
      "step": 2280
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.8032,
      "step": 2290
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.854,
      "step": 2300
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.8123,
      "step": 2310
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.8101,
      "step": 2320
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.8075,
      "step": 2330
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.817,
      "step": 2340
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.7747,
      "step": 2350
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.8012,
      "step": 2360
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7893,
      "step": 2370
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7661,
      "step": 2380
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7711,
      "step": 2390
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.8136,
      "step": 2400
    },
    {
      "epoch": 0.76,
      "eval_loss": 0.7395493388175964,
      "eval_runtime": 110.7923,
      "eval_samples_per_second": 9.026,
      "eval_steps_per_second": 4.513,
      "step": 2400
    },
    {
      "epoch": 0.76,
      "mmlu_eval_accuracy": 0.4873047408851529,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "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": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "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": 0.058823529411764705,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "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": 0.8461538461538461,
      "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": 0.88,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "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": 0.6666666666666666,
      "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": 0.7368421052631579,
      "mmlu_loss": 1.3917097181237397,
      "step": 2400
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7579,
      "step": 2410
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.8421,
      "step": 2420
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7957,
      "step": 2430
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7452,
      "step": 2440
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.8478,
      "step": 2450
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.8443,
      "step": 2460
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.8409,
      "step": 2470
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.8168,
      "step": 2480
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7648,
      "step": 2490
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7938,
      "step": 2500
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.791,
      "step": 2510
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7691,
      "step": 2520
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7648,
      "step": 2530
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7575,
      "step": 2540
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.7797,
      "step": 2550
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.7742,
      "step": 2560
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.8391,
      "step": 2570
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7746,
      "step": 2580
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7534,
      "step": 2590
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7395,
      "step": 2600
    },
    {
      "epoch": 0.82,
      "eval_loss": 0.7380212545394897,
      "eval_runtime": 111.0553,
      "eval_samples_per_second": 9.005,
      "eval_steps_per_second": 4.502,
      "step": 2600
    },
    {
      "epoch": 0.82,
      "mmlu_eval_accuracy": 0.4979448031756729,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "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": 0.16666666666666666,
      "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": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.058823529411764705,
      "mmlu_eval_accuracy_high_school_psychology": 0.8833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.2727272727272727,
      "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": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "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": 1.3912735815614696,
      "step": 2600
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7792,
      "step": 2610
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7228,
      "step": 2620
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7294,
      "step": 2630
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.6968,
      "step": 2640
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.7463,
      "step": 2650
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.7588,
      "step": 2660
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.7406,
      "step": 2670
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.7817,
      "step": 2680
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.808,
      "step": 2690
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.771,
      "step": 2700
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.7678,
      "step": 2710
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.7885,
      "step": 2720
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.8297,
      "step": 2730
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.8218,
      "step": 2740
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.7742,
      "step": 2750
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.7512,
      "step": 2760
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7508,
      "step": 2770
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7947,
      "step": 2780
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7399,
      "step": 2790
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7589,
      "step": 2800
    },
    {
      "epoch": 0.89,
      "eval_loss": 0.7355720400810242,
      "eval_runtime": 110.8718,
      "eval_samples_per_second": 9.019,
      "eval_steps_per_second": 4.51,
      "step": 2800
    },
    {
      "epoch": 0.89,
      "mmlu_eval_accuracy": 0.48346048137181885,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "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": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "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": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "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": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "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": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "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": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.5362318840579711,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3572583458831353,
      "step": 2800
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.8099,
      "step": 2810
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7303,
      "step": 2820
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.8154,
      "step": 2830
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.8166,
      "step": 2840
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.7425,
      "step": 2850
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.8223,
      "step": 2860
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7443,
      "step": 2870
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7733,
      "step": 2880
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.8092,
      "step": 2890
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7371,
      "step": 2900
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7323,
      "step": 2910
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7716,
      "step": 2920
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7824,
      "step": 2930
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7373,
      "step": 2940
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7384,
      "step": 2950
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7598,
      "step": 2960
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7211,
      "step": 2970
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7886,
      "step": 2980
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.8107,
      "step": 2990
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.8389,
      "step": 3000
    },
    {
      "epoch": 0.95,
      "eval_loss": 0.7343361377716064,
      "eval_runtime": 110.9061,
      "eval_samples_per_second": 9.017,
      "eval_steps_per_second": 4.508,
      "step": 3000
    },
    {
      "epoch": 0.95,
      "mmlu_eval_accuracy": 0.5003901788212859,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "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": 0.4375,
      "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": 0.5454545454545454,
      "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": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.53125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.18181818181818182,
      "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": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8833333333333333,
      "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": 0.8461538461538461,
      "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": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.22,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.217419229584014,
      "step": 3000
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.7964,
      "step": 3010
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.7841,
      "step": 3020
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.7951,
      "step": 3030
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.7523,
      "step": 3040
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.7729,
      "step": 3050
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.705,
      "step": 3060
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.7745,
      "step": 3070
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.7992,
      "step": 3080
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.7836,
      "step": 3090
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.7347,
      "step": 3100
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7213,
      "step": 3110
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7427,
      "step": 3120
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7799,
      "step": 3130
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.825,
      "step": 3140
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.7389,
      "step": 3150
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.8275,
      "step": 3160
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.7484,
      "step": 3170
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.7419,
      "step": 3180
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.6543,
      "step": 3190
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.6952,
      "step": 3200
    },
    {
      "epoch": 1.01,
      "eval_loss": 0.7377473711967468,
      "eval_runtime": 111.1786,
      "eval_samples_per_second": 8.995,
      "eval_steps_per_second": 4.497,
      "step": 3200
    },
    {
      "epoch": 1.01,
      "mmlu_eval_accuracy": 0.48409598343583005,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "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": 0.36363636363636365,
      "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": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
      "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": 0.18181818181818182,
      "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": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.058823529411764705,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "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": 0.88,
      "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": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "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": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.4613330938486144,
      "step": 3200
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.664,
      "step": 3210
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.6673,
      "step": 3220
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.703,
      "step": 3230
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.763,
      "step": 3240
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.6587,
      "step": 3250
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.6725,
      "step": 3260
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.7518,
      "step": 3270
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.7182,
      "step": 3280
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.6655,
      "step": 3290
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.6333,
      "step": 3300
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.6699,
      "step": 3310
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.659,
      "step": 3320
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.7138,
      "step": 3330
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.7309,
      "step": 3340
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.7251,
      "step": 3350
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.6712,
      "step": 3360
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.6527,
      "step": 3370
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.7752,
      "step": 3380
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.6896,
      "step": 3390
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.7441,
      "step": 3400
    },
    {
      "epoch": 1.08,
      "eval_loss": 0.7388539910316467,
      "eval_runtime": 111.0879,
      "eval_samples_per_second": 9.002,
      "eval_steps_per_second": 4.501,
      "step": 3400
    },
    {
      "epoch": 1.08,
      "mmlu_eval_accuracy": 0.49153955280819217,
      "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": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "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": 0.3125,
      "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": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "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": 0.2413793103448276,
      "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": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.3683368821990707,
      "step": 3400
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.723,
      "step": 3410
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.7545,
      "step": 3420
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.6885,
      "step": 3430
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.7021,
      "step": 3440
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.7284,
      "step": 3450
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.6811,
      "step": 3460
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.7076,
      "step": 3470
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.7074,
      "step": 3480
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.6734,
      "step": 3490
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.7243,
      "step": 3500
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.7347,
      "step": 3510
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6888,
      "step": 3520
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.7332,
      "step": 3530
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.7117,
      "step": 3540
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6575,
      "step": 3550
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.729,
      "step": 3560
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.6825,
      "step": 3570
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.6935,
      "step": 3580
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.7004,
      "step": 3590
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.7237,
      "step": 3600
    },
    {
      "epoch": 1.14,
      "eval_loss": 0.7381147742271423,
      "eval_runtime": 111.0101,
      "eval_samples_per_second": 9.008,
      "eval_steps_per_second": 4.504,
      "step": 3600
    },
    {
      "epoch": 1.14,
      "mmlu_eval_accuracy": 0.49167050353968145,
      "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": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "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": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "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": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "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": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.7209302325581395,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5362318840579711,
      "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": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.5044772917545806,
      "step": 3600
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.7361,
      "step": 3610
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.7179,
      "step": 3620
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.7499,
      "step": 3630
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.7319,
      "step": 3640
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.7104,
      "step": 3650
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.6892,
      "step": 3660
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.7666,
      "step": 3670
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.632,
      "step": 3680
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.713,
      "step": 3690
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.6958,
      "step": 3700
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.7253,
      "step": 3710
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.7608,
      "step": 3720
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.7277,
      "step": 3730
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.7346,
      "step": 3740
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.7075,
      "step": 3750
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6278,
      "step": 3760
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.7088,
      "step": 3770
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.7667,
      "step": 3780
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.7051,
      "step": 3790
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.699,
      "step": 3800
    },
    {
      "epoch": 1.2,
      "eval_loss": 0.7395787239074707,
      "eval_runtime": 110.7949,
      "eval_samples_per_second": 9.026,
      "eval_steps_per_second": 4.513,
      "step": 3800
    },
    {
      "epoch": 1.2,
      "mmlu_eval_accuracy": 0.48410138439418055,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "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": 0.3181818181818182,
      "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": 0.16666666666666666,
      "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": 0.18181818181818182,
      "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": 0.7619047619047619,
      "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": 0.058823529411764705,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.414894336547615,
      "step": 3800
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.6892,
      "step": 3810
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.6753,
      "step": 3820
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.6998,
      "step": 3830
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.686,
      "step": 3840
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.7254,
      "step": 3850
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.6942,
      "step": 3860
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.6729,
      "step": 3870
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.7486,
      "step": 3880
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.6997,
      "step": 3890
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.7308,
      "step": 3900
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.7214,
      "step": 3910
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.6879,
      "step": 3920
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.6662,
      "step": 3930
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.7045,
      "step": 3940
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.7908,
      "step": 3950
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.72,
      "step": 3960
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.6646,
      "step": 3970
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.7421,
      "step": 3980
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.7489,
      "step": 3990
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.7082,
      "step": 4000
    },
    {
      "epoch": 1.27,
      "eval_loss": 0.7381725907325745,
      "eval_runtime": 111.1345,
      "eval_samples_per_second": 8.998,
      "eval_steps_per_second": 4.499,
      "step": 4000
    },
    {
      "epoch": 1.27,
      "mmlu_eval_accuracy": 0.48533511185669687,
      "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": 0.4827586206896552,
      "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": 0.18181818181818182,
      "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": 0.16666666666666666,
      "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": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.13636363636363635,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "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": 0.34782608695652173,
      "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": 0.6111111111111112,
      "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": 0.7093023255813954,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "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": 0.6842105263157895,
      "mmlu_loss": 1.3075970206652858,
      "step": 4000
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.6578,
      "step": 4010
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.7462,
      "step": 4020
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.699,
      "step": 4030
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.7144,
      "step": 4040
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.6771,
      "step": 4050
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.7198,
      "step": 4060
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.6848,
      "step": 4070
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.762,
      "step": 4080
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.7354,
      "step": 4090
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.6529,
      "step": 4100
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.6373,
      "step": 4110
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.7415,
      "step": 4120
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.6646,
      "step": 4130
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.6904,
      "step": 4140
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.7462,
      "step": 4150
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.7261,
      "step": 4160
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.6866,
      "step": 4170
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.6789,
      "step": 4180
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.6943,
      "step": 4190
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.6644,
      "step": 4200
    },
    {
      "epoch": 1.33,
      "eval_loss": 0.7391716241836548,
      "eval_runtime": 111.1279,
      "eval_samples_per_second": 8.999,
      "eval_steps_per_second": 4.499,
      "step": 4200
    },
    {
      "epoch": 1.33,
      "mmlu_eval_accuracy": 0.48595716946128337,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.1875,
      "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": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.18181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.72,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.2436322906776758,
      "step": 4200
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.7427,
      "step": 4210
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.628,
      "step": 4220
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.6656,
      "step": 4230
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.6631,
      "step": 4240
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.7031,
      "step": 4250
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.7102,
      "step": 4260
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.7077,
      "step": 4270
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.7679,
      "step": 4280
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.6569,
      "step": 4290
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.6911,
      "step": 4300
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.7468,
      "step": 4310
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.6641,
      "step": 4320
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.7248,
      "step": 4330
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.706,
      "step": 4340
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.717,
      "step": 4350
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.6462,
      "step": 4360
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.6752,
      "step": 4370
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.7239,
      "step": 4380
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.6665,
      "step": 4390
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.7077,
      "step": 4400
    },
    {
      "epoch": 1.39,
      "eval_loss": 0.7374858260154724,
      "eval_runtime": 111.3021,
      "eval_samples_per_second": 8.985,
      "eval_steps_per_second": 4.492,
      "step": 4400
    },
    {
      "epoch": 1.39,
      "mmlu_eval_accuracy": 0.49250240895964725,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
      "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": 0.25,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.3333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "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": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "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": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "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": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.2793094400637455,
      "step": 4400
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.7042,
      "step": 4410
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.7554,
      "step": 4420
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.757,
      "step": 4430
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.7173,
      "step": 4440
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.6655,
      "step": 4450
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.6991,
      "step": 4460
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.7148,
      "step": 4470
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.7085,
      "step": 4480
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.6955,
      "step": 4490
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.7139,
      "step": 4500
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.7262,
      "step": 4510
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.7705,
      "step": 4520
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.7028,
      "step": 4530
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.7146,
      "step": 4540
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.6868,
      "step": 4550
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.6591,
      "step": 4560
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.7019,
      "step": 4570
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.6676,
      "step": 4580
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.7085,
      "step": 4590
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.664,
      "step": 4600
    },
    {
      "epoch": 1.46,
      "eval_loss": 0.7358158230781555,
      "eval_runtime": 111.2934,
      "eval_samples_per_second": 8.985,
      "eval_steps_per_second": 4.493,
      "step": 4600
    },
    {
      "epoch": 1.46,
      "mmlu_eval_accuracy": 0.4861300261183481,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "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": 0.2727272727272727,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.3333333333333333,
      "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": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.18181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "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": 0.1724137931034483,
      "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": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "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": 0.6842105263157895,
      "mmlu_loss": 1.3095654961567946,
      "step": 4600
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.6486,
      "step": 4610
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.6999,
      "step": 4620
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.6458,
      "step": 4630
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.6762,
      "step": 4640
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.6924,
      "step": 4650
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.682,
      "step": 4660
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.7081,
      "step": 4670
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.7506,
      "step": 4680
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.7311,
      "step": 4690
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6463,
      "step": 4700
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6741,
      "step": 4710
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.6626,
      "step": 4720
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.712,
      "step": 4730
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.6676,
      "step": 4740
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.7193,
      "step": 4750
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.6699,
      "step": 4760
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.6718,
      "step": 4770
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.6899,
      "step": 4780
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.6954,
      "step": 4790
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.7187,
      "step": 4800
    },
    {
      "epoch": 1.52,
      "eval_loss": 0.7387924790382385,
      "eval_runtime": 111.1141,
      "eval_samples_per_second": 9.0,
      "eval_steps_per_second": 4.5,
      "step": 4800
    },
    {
      "epoch": 1.52,
      "mmlu_eval_accuracy": 0.4879926358283337,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "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": 0.3125,
      "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": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.18181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5384615384615384,
      "mmlu_eval_accuracy_high_school_physics": 0.058823529411764705,
      "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": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "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": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "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": 0.5483870967741935,
      "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": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.4884750641901874,
      "step": 4800
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.6733,
      "step": 4810
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.6607,
      "step": 4820
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.6933,
      "step": 4830
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.7517,
      "step": 4840
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.7391,
      "step": 4850
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.6636,
      "step": 4860
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.7221,
      "step": 4870
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.6967,
      "step": 4880
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.7117,
      "step": 4890
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.6256,
      "step": 4900
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.7923,
      "step": 4910
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.7151,
      "step": 4920
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.7119,
      "step": 4930
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.7105,
      "step": 4940
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.6653,
      "step": 4950
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.7084,
      "step": 4960
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.6644,
      "step": 4970
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.6665,
      "step": 4980
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.6746,
      "step": 4990
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.7223,
      "step": 5000
    },
    {
      "epoch": 1.58,
      "eval_loss": 0.7373215556144714,
      "eval_runtime": 111.2649,
      "eval_samples_per_second": 8.988,
      "eval_steps_per_second": 4.494,
      "step": 5000
    },
    {
      "epoch": 1.58,
      "mmlu_eval_accuracy": 0.46701126611778865,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "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": 0.2727272727272727,
      "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": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "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": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.18181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "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": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.058823529411764705,
      "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": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.72,
      "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": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.4578603010031324,
      "step": 5000
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.6833,
      "step": 5010
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.7323,
      "step": 5020
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.7224,
      "step": 5030
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.734,
      "step": 5040
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.692,
      "step": 5050
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.7083,
      "step": 5060
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.6993,
      "step": 5070
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.755,
      "step": 5080
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.7323,
      "step": 5090
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.6725,
      "step": 5100
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.6989,
      "step": 5110
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.6938,
      "step": 5120
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.6895,
      "step": 5130
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.6915,
      "step": 5140
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.7672,
      "step": 5150
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.6413,
      "step": 5160
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.7195,
      "step": 5170
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.6783,
      "step": 5180
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.6457,
      "step": 5190
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.6959,
      "step": 5200
    },
    {
      "epoch": 1.65,
      "eval_loss": 0.736714243888855,
      "eval_runtime": 111.0389,
      "eval_samples_per_second": 9.006,
      "eval_steps_per_second": 4.503,
      "step": 5200
    },
    {
      "epoch": 1.65,
      "mmlu_eval_accuracy": 0.4835750759985151,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "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": 0.2727272727272727,
      "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": 0.5454545454545454,
      "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": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.18181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "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": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "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": 0.8461538461538461,
      "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": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "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": 0.631578947368421,
      "mmlu_loss": 1.2928575183564004,
      "step": 5200
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.654,
      "step": 5210
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.692,
      "step": 5220
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.6774,
      "step": 5230
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.6383,
      "step": 5240
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.6949,
      "step": 5250
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.6992,
      "step": 5260
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.6612,
      "step": 5270
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.7651,
      "step": 5280
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.6994,
      "step": 5290
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.7105,
      "step": 5300
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.6972,
      "step": 5310
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.7531,
      "step": 5320
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.7072,
      "step": 5330
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.6964,
      "step": 5340
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.7574,
      "step": 5350
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.7155,
      "step": 5360
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.7104,
      "step": 5370
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.7495,
      "step": 5380
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.7259,
      "step": 5390
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.7394,
      "step": 5400
    },
    {
      "epoch": 1.71,
      "eval_loss": 0.7311118245124817,
      "eval_runtime": 111.2623,
      "eval_samples_per_second": 8.988,
      "eval_steps_per_second": 4.494,
      "step": 5400
    },
    {
      "epoch": 1.71,
      "mmlu_eval_accuracy": 0.4848351410303045,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "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": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "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": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "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": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
      "mmlu_eval_accuracy_moral_scenarios": 0.22,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "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": 1.348453690776937,
      "step": 5400
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.7047,
      "step": 5410
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.7001,
      "step": 5420
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.6759,
      "step": 5430
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.707,
      "step": 5440
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.6648,
      "step": 5450
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.7223,
      "step": 5460
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.722,
      "step": 5470
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.7848,
      "step": 5480
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.6956,
      "step": 5490
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.6584,
      "step": 5500
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.7522,
      "step": 5510
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.7374,
      "step": 5520
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.635,
      "step": 5530
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.6947,
      "step": 5540
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.6948,
      "step": 5550
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.676,
      "step": 5560
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.7053,
      "step": 5570
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.6868,
      "step": 5580
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.7307,
      "step": 5590
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.6902,
      "step": 5600
    },
    {
      "epoch": 1.77,
      "eval_loss": 0.7314637899398804,
      "eval_runtime": 111.0487,
      "eval_samples_per_second": 9.005,
      "eval_steps_per_second": 4.503,
      "step": 5600
    },
    {
      "epoch": 1.77,
      "mmlu_eval_accuracy": 0.48467107795368586,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "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": 0.2727272727272727,
      "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": 0.5454545454545454,
      "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": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "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": 0.058823529411764705,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "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": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "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": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.3874250733665636,
      "step": 5600
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.6558,
      "step": 5610
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.714,
      "step": 5620
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.7019,
      "step": 5630
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.7084,
      "step": 5640
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.7184,
      "step": 5650
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.6524,
      "step": 5660
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.7265,
      "step": 5670
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.7164,
      "step": 5680
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.6825,
      "step": 5690
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.7427,
      "step": 5700
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.7416,
      "step": 5710
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.7027,
      "step": 5720
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002,
      "loss": 0.7039,
      "step": 5730
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002,
      "loss": 0.7108,
      "step": 5740
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002,
      "loss": 0.6257,
      "step": 5750
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.6665,
      "step": 5760
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.7371,
      "step": 5770
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.7194,
      "step": 5780
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.7164,
      "step": 5790
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0002,
      "loss": 0.6887,
      "step": 5800
    },
    {
      "epoch": 1.84,
      "eval_loss": 0.732559084892273,
      "eval_runtime": 111.5342,
      "eval_samples_per_second": 8.966,
      "eval_steps_per_second": 4.483,
      "step": 5800
    },
    {
      "epoch": 1.84,
      "mmlu_eval_accuracy": 0.4740066355704332,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "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": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "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": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.29411764705882354,
      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.3845557023822173,
      "step": 5800
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0002,
      "loss": 0.668,
      "step": 5810
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0002,
      "loss": 0.6993,
      "step": 5820
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002,
      "loss": 0.7418,
      "step": 5830
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002,
      "loss": 0.6916,
      "step": 5840
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002,
      "loss": 0.7564,
      "step": 5850
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002,
      "loss": 0.641,
      "step": 5860
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002,
      "loss": 0.7593,
      "step": 5870
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002,
      "loss": 0.6886,
      "step": 5880
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0002,
      "loss": 0.7053,
      "step": 5890
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0002,
      "loss": 0.6201,
      "step": 5900
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0002,
      "loss": 0.6998,
      "step": 5910
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0002,
      "loss": 0.6768,
      "step": 5920
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0002,
      "loss": 0.711,
      "step": 5930
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0002,
      "loss": 0.681,
      "step": 5940
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.7145,
      "step": 5950
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.7513,
      "step": 5960
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.6817,
      "step": 5970
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.6757,
      "step": 5980
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0002,
      "loss": 0.6899,
      "step": 5990
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0002,
      "loss": 0.6821,
      "step": 6000
    },
    {
      "epoch": 1.9,
      "eval_loss": 0.7302425503730774,
      "eval_runtime": 111.0525,
      "eval_samples_per_second": 9.005,
      "eval_steps_per_second": 4.502,
      "step": 6000
    },
    {
      "epoch": 1.9,
      "mmlu_eval_accuracy": 0.47023094937776666,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "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": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "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": 0.4230769230769231,
      "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": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "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": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "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": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "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": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.4106916200087525,
      "step": 6000
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0002,
      "loss": 0.7115,
      "step": 6010
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002,
      "loss": 0.6862,
      "step": 6020
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002,
      "loss": 0.6705,
      "step": 6030
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002,
      "loss": 0.6848,
      "step": 6040
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.7765,
      "step": 6050
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.6801,
      "step": 6060
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.6648,
      "step": 6070
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0002,
      "loss": 0.6847,
      "step": 6080
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0002,
      "loss": 0.665,
      "step": 6090
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0002,
      "loss": 0.7627,
      "step": 6100
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0002,
      "loss": 0.6874,
      "step": 6110
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0002,
      "loss": 0.6907,
      "step": 6120
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0002,
      "loss": 0.6369,
      "step": 6130
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0002,
      "loss": 0.7289,
      "step": 6140
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0002,
      "loss": 0.7233,
      "step": 6150
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0002,
      "loss": 0.68,
      "step": 6160
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.6842,
      "step": 6170
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.7125,
      "step": 6180
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.683,
      "step": 6190
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.7097,
      "step": 6200
    },
    {
      "epoch": 1.96,
      "eval_loss": 0.7293602228164673,
      "eval_runtime": 111.0579,
      "eval_samples_per_second": 9.004,
      "eval_steps_per_second": 4.502,
      "step": 6200
    },
    {
      "epoch": 1.96,
      "mmlu_eval_accuracy": 0.4704848103487601,
      "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": 0.4482758620689655,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "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": 0.34146341463414637,
      "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": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "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": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
      "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": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
      "mmlu_eval_accuracy_moral_scenarios": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "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": 0.631578947368421,
      "mmlu_loss": 1.374586288985011,
      "step": 6200
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0002,
      "loss": 0.7095,
      "step": 6210
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0002,
      "loss": 0.7681,
      "step": 6220
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0002,
      "loss": 0.7356,
      "step": 6230
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0002,
      "loss": 0.6956,
      "step": 6240
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0002,
      "loss": 0.7034,
      "step": 6250
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0002,
      "loss": 0.6532,
      "step": 6260
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0002,
      "loss": 0.6917,
      "step": 6270
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0002,
      "loss": 0.6392,
      "step": 6280
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0002,
      "loss": 0.6656,
      "step": 6290
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0002,
      "loss": 0.6829,
      "step": 6300
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0002,
      "loss": 0.675,
      "step": 6310
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0002,
      "loss": 0.6321,
      "step": 6320
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0002,
      "loss": 0.6109,
      "step": 6330
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0002,
      "loss": 0.6065,
      "step": 6340
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0002,
      "loss": 0.5912,
      "step": 6350
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0002,
      "loss": 0.613,
      "step": 6360
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0002,
      "loss": 0.586,
      "step": 6370
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0002,
      "loss": 0.6383,
      "step": 6380
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0002,
      "loss": 0.5629,
      "step": 6390
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0002,
      "loss": 0.6048,
      "step": 6400
    },
    {
      "epoch": 2.03,
      "eval_loss": 0.7574472427368164,
      "eval_runtime": 110.9511,
      "eval_samples_per_second": 9.013,
      "eval_steps_per_second": 4.506,
      "step": 6400
    },
    {
      "epoch": 2.03,
      "mmlu_eval_accuracy": 0.470592564742188,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "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": 0.125,
      "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": 0.4230769230769231,
      "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": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.0,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "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": 0.3333333333333333,
      "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": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "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": 0.5833333333333334,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.3004325469542422,
      "step": 6400
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0002,
      "loss": 0.5702,
      "step": 6410
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0002,
      "loss": 0.5957,
      "step": 6420
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0002,
      "loss": 0.5994,
      "step": 6430
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0002,
      "loss": 0.5922,
      "step": 6440
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0002,
      "loss": 0.5626,
      "step": 6450
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0002,
      "loss": 0.5912,
      "step": 6460
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0002,
      "loss": 0.5877,
      "step": 6470
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0002,
      "loss": 0.578,
      "step": 6480
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0002,
      "loss": 0.6207,
      "step": 6490
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0002,
      "loss": 0.5606,
      "step": 6500
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0002,
      "loss": 0.553,
      "step": 6510
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0002,
      "loss": 0.6092,
      "step": 6520
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0002,
      "loss": 0.6183,
      "step": 6530
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0002,
      "loss": 0.5825,
      "step": 6540
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0002,
      "loss": 0.5674,
      "step": 6550
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0002,
      "loss": 0.5587,
      "step": 6560
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0002,
      "loss": 0.5317,
      "step": 6570
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0002,
      "loss": 0.6731,
      "step": 6580
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0002,
      "loss": 0.6242,
      "step": 6590
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0002,
      "loss": 0.6332,
      "step": 6600
    },
    {
      "epoch": 2.09,
      "eval_loss": 0.7567528486251831,
      "eval_runtime": 111.0264,
      "eval_samples_per_second": 9.007,
      "eval_steps_per_second": 4.503,
      "step": 6600
    },
    {
      "epoch": 2.09,
      "mmlu_eval_accuracy": 0.47542707100737025,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "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": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "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": 0.058823529411764705,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "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": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "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": 0.5806451612903226,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.4275867019247448,
      "step": 6600
    }
  ],
  "max_steps": 10000,
  "num_train_epochs": 4,
  "total_flos": 2.0028582776394056e+18,
  "trial_name": null,
  "trial_params": null
}