{
  "best_metric": 0.4259847402572632,
  "best_model_checkpoint": "experts/expert-3/checkpoint-10000",
  "epoch": 1.9743336623889438,
  "global_step": 10000,
  "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": 0.6315,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0002,
      "loss": 0.5934,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.626,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.5429,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.5681,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.5343,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.5877,
      "step": 70
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.5127,
      "step": 80
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.6177,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.5896,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.5248,
      "step": 110
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.5587,
      "step": 120
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.595,
      "step": 130
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.5844,
      "step": 140
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.5512,
      "step": 150
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.5369,
      "step": 160
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.538,
      "step": 170
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.5353,
      "step": 180
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.5467,
      "step": 190
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.5804,
      "step": 200
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.5388554930686951,
      "eval_runtime": 120.9678,
      "eval_samples_per_second": 8.267,
      "eval_steps_per_second": 4.133,
      "step": 200
    },
    {
      "epoch": 0.04,
      "mmlu_eval_accuracy": 0.48625561253335453,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "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": 0.36363636363636365,
      "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": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.6132891311969209,
      "step": 200
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.5957,
      "step": 210
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.5015,
      "step": 220
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.5642,
      "step": 230
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.5342,
      "step": 240
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.5747,
      "step": 250
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.5296,
      "step": 260
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.5372,
      "step": 270
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.5351,
      "step": 280
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.5149,
      "step": 290
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.506,
      "step": 300
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.537,
      "step": 310
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.5453,
      "step": 320
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.4446,
      "step": 330
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.57,
      "step": 340
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.5069,
      "step": 350
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.5427,
      "step": 360
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.548,
      "step": 370
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.512,
      "step": 380
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.5288,
      "step": 390
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.5621,
      "step": 400
    },
    {
      "epoch": 0.08,
      "eval_loss": 0.5241960287094116,
      "eval_runtime": 120.9955,
      "eval_samples_per_second": 8.265,
      "eval_steps_per_second": 4.132,
      "step": 400
    },
    {
      "epoch": 0.08,
      "mmlu_eval_accuracy": 0.49496500537399013,
      "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "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": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.6960247003378508,
      "step": 400
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.5103,
      "step": 410
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.5214,
      "step": 420
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.5224,
      "step": 430
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.5354,
      "step": 440
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.5348,
      "step": 450
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.57,
      "step": 460
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.4568,
      "step": 470
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.4849,
      "step": 480
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.4744,
      "step": 490
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.498,
      "step": 500
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.5287,
      "step": 510
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.5238,
      "step": 520
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.4976,
      "step": 530
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.5194,
      "step": 540
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.4604,
      "step": 550
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.515,
      "step": 560
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.5078,
      "step": 570
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.4964,
      "step": 580
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.562,
      "step": 590
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.468,
      "step": 600
    },
    {
      "epoch": 0.12,
      "eval_loss": 0.5107640624046326,
      "eval_runtime": 120.932,
      "eval_samples_per_second": 8.269,
      "eval_steps_per_second": 4.135,
      "step": 600
    },
    {
      "epoch": 0.12,
      "mmlu_eval_accuracy": 0.4755733813823238,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "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": 0.18181818181818182,
      "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": 0.38461538461538464,
      "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": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "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": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.5026965173977784,
      "step": 600
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.5165,
      "step": 610
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.4661,
      "step": 620
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.4972,
      "step": 630
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.4491,
      "step": 640
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.5422,
      "step": 650
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.509,
      "step": 660
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.4908,
      "step": 670
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.5095,
      "step": 680
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.5728,
      "step": 690
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.5072,
      "step": 700
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.4679,
      "step": 710
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.6002,
      "step": 720
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.5409,
      "step": 730
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.5171,
      "step": 740
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.487,
      "step": 750
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.4966,
      "step": 760
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.5075,
      "step": 770
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.4843,
      "step": 780
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.5621,
      "step": 790
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.4956,
      "step": 800
    },
    {
      "epoch": 0.16,
      "eval_loss": 0.5016384124755859,
      "eval_runtime": 120.9061,
      "eval_samples_per_second": 8.271,
      "eval_steps_per_second": 4.135,
      "step": 800
    },
    {
      "epoch": 0.16,
      "mmlu_eval_accuracy": 0.4932069529610808,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "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": 0.2727272727272727,
      "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": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "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": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "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": 0.7368421052631579,
      "mmlu_loss": 1.6144507999675393,
      "step": 800
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.5247,
      "step": 810
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.5195,
      "step": 820
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.5139,
      "step": 830
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.493,
      "step": 840
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.5231,
      "step": 850
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.4524,
      "step": 860
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.5345,
      "step": 870
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.4938,
      "step": 880
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.4413,
      "step": 890
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.4804,
      "step": 900
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.4931,
      "step": 910
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.5052,
      "step": 920
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.4568,
      "step": 930
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.5116,
      "step": 940
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.5283,
      "step": 950
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.4458,
      "step": 960
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.4363,
      "step": 970
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.52,
      "step": 980
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.4738,
      "step": 990
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.4412,
      "step": 1000
    },
    {
      "epoch": 0.2,
      "eval_loss": 0.4944809675216675,
      "eval_runtime": 120.9421,
      "eval_samples_per_second": 8.268,
      "eval_steps_per_second": 4.134,
      "step": 1000
    },
    {
      "epoch": 0.2,
      "mmlu_eval_accuracy": 0.4918118163660498,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "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": 0.4090909090909091,
      "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": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "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": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "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": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "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": 1.4865406374240357,
      "step": 1000
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.4602,
      "step": 1010
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.529,
      "step": 1020
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.5004,
      "step": 1030
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.4578,
      "step": 1040
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.5158,
      "step": 1050
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.4544,
      "step": 1060
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.5057,
      "step": 1070
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.5576,
      "step": 1080
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.5211,
      "step": 1090
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.5121,
      "step": 1100
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.5027,
      "step": 1110
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.5084,
      "step": 1120
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.5442,
      "step": 1130
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.5235,
      "step": 1140
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.511,
      "step": 1150
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.5256,
      "step": 1160
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.4597,
      "step": 1170
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.4232,
      "step": 1180
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.4606,
      "step": 1190
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.4949,
      "step": 1200
    },
    {
      "epoch": 0.24,
      "eval_loss": 0.4886409342288971,
      "eval_runtime": 120.9084,
      "eval_samples_per_second": 8.271,
      "eval_steps_per_second": 4.135,
      "step": 1200
    },
    {
      "epoch": 0.24,
      "mmlu_eval_accuracy": 0.49673794687932926,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "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": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4536286007330872,
      "step": 1200
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.5038,
      "step": 1210
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.4878,
      "step": 1220
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.5202,
      "step": 1230
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.4384,
      "step": 1240
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.4801,
      "step": 1250
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.4979,
      "step": 1260
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.4846,
      "step": 1270
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.4547,
      "step": 1280
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.4653,
      "step": 1290
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.4676,
      "step": 1300
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.4748,
      "step": 1310
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.4746,
      "step": 1320
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.46,
      "step": 1330
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.4767,
      "step": 1340
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.531,
      "step": 1350
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.506,
      "step": 1360
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.5061,
      "step": 1370
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.5629,
      "step": 1380
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.4354,
      "step": 1390
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.488,
      "step": 1400
    },
    {
      "epoch": 0.28,
      "eval_loss": 0.4853256344795227,
      "eval_runtime": 120.9475,
      "eval_samples_per_second": 8.268,
      "eval_steps_per_second": 4.134,
      "step": 1400
    },
    {
      "epoch": 0.28,
      "mmlu_eval_accuracy": 0.48962734867617536,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "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": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "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": 0.34375,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "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": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.5658895157802821,
      "step": 1400
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.5051,
      "step": 1410
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.5125,
      "step": 1420
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.4898,
      "step": 1430
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.4969,
      "step": 1440
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.4657,
      "step": 1450
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.474,
      "step": 1460
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.4438,
      "step": 1470
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.4677,
      "step": 1480
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.4539,
      "step": 1490
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.4329,
      "step": 1500
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.5486,
      "step": 1510
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.4921,
      "step": 1520
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.4808,
      "step": 1530
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.4471,
      "step": 1540
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.4764,
      "step": 1550
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.4781,
      "step": 1560
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.4904,
      "step": 1570
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.4473,
      "step": 1580
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.5187,
      "step": 1590
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.4556,
      "step": 1600
    },
    {
      "epoch": 0.32,
      "eval_loss": 0.48241180181503296,
      "eval_runtime": 120.8643,
      "eval_samples_per_second": 8.274,
      "eval_steps_per_second": 4.137,
      "step": 1600
    },
    {
      "epoch": 0.32,
      "mmlu_eval_accuracy": 0.4883575864774028,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "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": 0.18181818181818182,
      "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": 0.45454545454545453,
      "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": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "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": 1.5009440863723855,
      "step": 1600
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.4643,
      "step": 1610
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.522,
      "step": 1620
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.4945,
      "step": 1630
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.4595,
      "step": 1640
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.462,
      "step": 1650
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.5106,
      "step": 1660
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.4827,
      "step": 1670
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.53,
      "step": 1680
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.4892,
      "step": 1690
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.4944,
      "step": 1700
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.4504,
      "step": 1710
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.4482,
      "step": 1720
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.4754,
      "step": 1730
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.4422,
      "step": 1740
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.4506,
      "step": 1750
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.4944,
      "step": 1760
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.48,
      "step": 1770
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.507,
      "step": 1780
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.5066,
      "step": 1790
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.5256,
      "step": 1800
    },
    {
      "epoch": 0.36,
      "eval_loss": 0.48142096400260925,
      "eval_runtime": 120.9534,
      "eval_samples_per_second": 8.268,
      "eval_steps_per_second": 4.134,
      "step": 1800
    },
    {
      "epoch": 0.36,
      "mmlu_eval_accuracy": 0.49656525776157384,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "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": 0.5,
      "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": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "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": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.5533446454036952,
      "step": 1800
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.487,
      "step": 1810
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.4907,
      "step": 1820
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.4508,
      "step": 1830
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.465,
      "step": 1840
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.468,
      "step": 1850
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.4473,
      "step": 1860
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.4871,
      "step": 1870
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.4548,
      "step": 1880
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.443,
      "step": 1890
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.4456,
      "step": 1900
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.4391,
      "step": 1910
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.4814,
      "step": 1920
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.4234,
      "step": 1930
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.475,
      "step": 1940
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.4788,
      "step": 1950
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.4003,
      "step": 1960
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.4877,
      "step": 1970
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.5129,
      "step": 1980
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.44,
      "step": 1990
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.5063,
      "step": 2000
    },
    {
      "epoch": 0.39,
      "eval_loss": 0.4745611250400543,
      "eval_runtime": 120.8642,
      "eval_samples_per_second": 8.274,
      "eval_steps_per_second": 4.137,
      "step": 2000
    },
    {
      "epoch": 0.39,
      "mmlu_eval_accuracy": 0.4888163204860966,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "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": 0.375,
      "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": 0.4230769230769231,
      "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": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "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": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.544741007817943,
      "step": 2000
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.4762,
      "step": 2010
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.4579,
      "step": 2020
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.4871,
      "step": 2030
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.485,
      "step": 2040
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.4452,
      "step": 2050
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.453,
      "step": 2060
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.4548,
      "step": 2070
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.4476,
      "step": 2080
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.4528,
      "step": 2090
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.4832,
      "step": 2100
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.5121,
      "step": 2110
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.5431,
      "step": 2120
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.5039,
      "step": 2130
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.4664,
      "step": 2140
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.5198,
      "step": 2150
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.5133,
      "step": 2160
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.473,
      "step": 2170
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.4378,
      "step": 2180
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.4392,
      "step": 2190
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.4974,
      "step": 2200
    },
    {
      "epoch": 0.43,
      "eval_loss": 0.4701443910598755,
      "eval_runtime": 121.0106,
      "eval_samples_per_second": 8.264,
      "eval_steps_per_second": 4.132,
      "step": 2200
    },
    {
      "epoch": 0.43,
      "mmlu_eval_accuracy": 0.49320092657219466,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "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": 0.21428571428571427,
      "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": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.4525446788293574,
      "step": 2200
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.4383,
      "step": 2210
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.438,
      "step": 2220
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.4679,
      "step": 2230
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.5195,
      "step": 2240
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.498,
      "step": 2250
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.4548,
      "step": 2260
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.4683,
      "step": 2270
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.4457,
      "step": 2280
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.4572,
      "step": 2290
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.4851,
      "step": 2300
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.508,
      "step": 2310
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.4427,
      "step": 2320
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.4894,
      "step": 2330
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.4747,
      "step": 2340
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.4248,
      "step": 2350
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.4464,
      "step": 2360
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.5028,
      "step": 2370
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.4596,
      "step": 2380
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.4853,
      "step": 2390
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.4449,
      "step": 2400
    },
    {
      "epoch": 0.47,
      "eval_loss": 0.4669264554977417,
      "eval_runtime": 120.8777,
      "eval_samples_per_second": 8.273,
      "eval_steps_per_second": 4.136,
      "step": 2400
    },
    {
      "epoch": 0.47,
      "mmlu_eval_accuracy": 0.4866731314157804,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "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": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "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": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.3380366984375462,
      "step": 2400
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.5165,
      "step": 2410
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.4427,
      "step": 2420
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.4659,
      "step": 2430
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.4092,
      "step": 2440
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.4962,
      "step": 2450
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.4625,
      "step": 2460
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.4924,
      "step": 2470
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.4835,
      "step": 2480
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.507,
      "step": 2490
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.4307,
      "step": 2500
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.5461,
      "step": 2510
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.4439,
      "step": 2520
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.5406,
      "step": 2530
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.4249,
      "step": 2540
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.436,
      "step": 2550
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.4839,
      "step": 2560
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.4224,
      "step": 2570
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.4361,
      "step": 2580
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.4657,
      "step": 2590
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.4279,
      "step": 2600
    },
    {
      "epoch": 0.51,
      "eval_loss": 0.46479299664497375,
      "eval_runtime": 120.926,
      "eval_samples_per_second": 8.27,
      "eval_steps_per_second": 4.135,
      "step": 2600
    },
    {
      "epoch": 0.51,
      "mmlu_eval_accuracy": 0.4754015400670814,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "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": 0.4090909090909091,
      "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": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "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": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3562013523230043,
      "step": 2600
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.4934,
      "step": 2610
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.5169,
      "step": 2620
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.408,
      "step": 2630
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.4046,
      "step": 2640
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.4291,
      "step": 2650
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.4555,
      "step": 2660
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.4633,
      "step": 2670
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.5286,
      "step": 2680
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.4853,
      "step": 2690
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.4444,
      "step": 2700
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.4377,
      "step": 2710
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.4823,
      "step": 2720
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.4756,
      "step": 2730
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.4809,
      "step": 2740
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.4461,
      "step": 2750
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.4633,
      "step": 2760
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.4764,
      "step": 2770
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.4446,
      "step": 2780
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.498,
      "step": 2790
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.4375,
      "step": 2800
    },
    {
      "epoch": 0.55,
      "eval_loss": 0.46188923716545105,
      "eval_runtime": 120.9443,
      "eval_samples_per_second": 8.268,
      "eval_steps_per_second": 4.134,
      "step": 2800
    },
    {
      "epoch": 0.55,
      "mmlu_eval_accuracy": 0.4847602790212367,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "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": 0.2727272727272727,
      "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": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "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": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.4064584104028135,
      "step": 2800
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.4356,
      "step": 2810
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.4214,
      "step": 2820
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.4708,
      "step": 2830
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.4852,
      "step": 2840
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.4497,
      "step": 2850
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.4408,
      "step": 2860
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.5293,
      "step": 2870
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.557,
      "step": 2880
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.5128,
      "step": 2890
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.4791,
      "step": 2900
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.4772,
      "step": 2910
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.431,
      "step": 2920
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.4155,
      "step": 2930
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.4177,
      "step": 2940
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.453,
      "step": 2950
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.4828,
      "step": 2960
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.4745,
      "step": 2970
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.4577,
      "step": 2980
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.4317,
      "step": 2990
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.5035,
      "step": 3000
    },
    {
      "epoch": 0.59,
      "eval_loss": 0.4610925018787384,
      "eval_runtime": 121.1229,
      "eval_samples_per_second": 8.256,
      "eval_steps_per_second": 4.128,
      "step": 3000
    },
    {
      "epoch": 0.59,
      "mmlu_eval_accuracy": 0.4919900164344259,
      "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": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "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": 0.3125,
      "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": 0.53125,
      "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": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "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": 0.6842105263157895,
      "mmlu_loss": 1.403657929937771,
      "step": 3000
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.4584,
      "step": 3010
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.4637,
      "step": 3020
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.4769,
      "step": 3030
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.4119,
      "step": 3040
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.4864,
      "step": 3050
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.4427,
      "step": 3060
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.424,
      "step": 3070
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.4701,
      "step": 3080
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.3826,
      "step": 3090
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.4408,
      "step": 3100
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.4834,
      "step": 3110
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.4982,
      "step": 3120
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.4404,
      "step": 3130
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.4507,
      "step": 3140
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.4693,
      "step": 3150
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.4584,
      "step": 3160
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.4698,
      "step": 3170
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.444,
      "step": 3180
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.4466,
      "step": 3190
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.4202,
      "step": 3200
    },
    {
      "epoch": 0.63,
      "eval_loss": 0.4561157524585724,
      "eval_runtime": 121.2847,
      "eval_samples_per_second": 8.245,
      "eval_steps_per_second": 4.123,
      "step": 3200
    },
    {
      "epoch": 0.63,
      "mmlu_eval_accuracy": 0.48376654691459353,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.09090909090909091,
      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "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": 0.6956521739130435,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.2846685934704838,
      "step": 3200
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.4429,
      "step": 3210
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.4753,
      "step": 3220
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.4564,
      "step": 3230
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.4141,
      "step": 3240
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.4408,
      "step": 3250
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.4264,
      "step": 3260
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.4858,
      "step": 3270
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.4386,
      "step": 3280
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.4695,
      "step": 3290
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.5033,
      "step": 3300
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.4331,
      "step": 3310
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.4349,
      "step": 3320
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.4408,
      "step": 3330
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.5011,
      "step": 3340
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.5235,
      "step": 3350
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.4472,
      "step": 3360
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.4597,
      "step": 3370
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.4219,
      "step": 3380
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.4587,
      "step": 3390
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.4483,
      "step": 3400
    },
    {
      "epoch": 0.67,
      "eval_loss": 0.45556724071502686,
      "eval_runtime": 120.8953,
      "eval_samples_per_second": 8.272,
      "eval_steps_per_second": 4.136,
      "step": 3400
    },
    {
      "epoch": 0.67,
      "mmlu_eval_accuracy": 0.48764633442828353,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "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": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.1875,
      "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": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "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": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "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": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2232888696554747,
      "step": 3400
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.4424,
      "step": 3410
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.4878,
      "step": 3420
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.4628,
      "step": 3430
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.4114,
      "step": 3440
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.4322,
      "step": 3450
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.5294,
      "step": 3460
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.4533,
      "step": 3470
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.47,
      "step": 3480
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.448,
      "step": 3490
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.4776,
      "step": 3500
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.4215,
      "step": 3510
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.4664,
      "step": 3520
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.474,
      "step": 3530
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.5337,
      "step": 3540
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.4892,
      "step": 3550
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.5021,
      "step": 3560
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.4266,
      "step": 3570
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.494,
      "step": 3580
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.4819,
      "step": 3590
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.4495,
      "step": 3600
    },
    {
      "epoch": 0.71,
      "eval_loss": 0.45507949590682983,
      "eval_runtime": 120.8484,
      "eval_samples_per_second": 8.275,
      "eval_steps_per_second": 4.137,
      "step": 3600
    },
    {
      "epoch": 0.71,
      "mmlu_eval_accuracy": 0.4989502092980565,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
      "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": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.46875,
      "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": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "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": 0.6521739130434783,
      "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": 0.5454545454545454,
      "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": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.36470588235294116,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2834886478615182,
      "step": 3600
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.395,
      "step": 3610
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.4795,
      "step": 3620
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.5049,
      "step": 3630
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.4186,
      "step": 3640
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.4121,
      "step": 3650
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.4233,
      "step": 3660
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.4311,
      "step": 3670
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.4896,
      "step": 3680
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.4856,
      "step": 3690
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.4353,
      "step": 3700
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.429,
      "step": 3710
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.4444,
      "step": 3720
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.4704,
      "step": 3730
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.4574,
      "step": 3740
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.5103,
      "step": 3750
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.4304,
      "step": 3760
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.4187,
      "step": 3770
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.4918,
      "step": 3780
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.435,
      "step": 3790
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.4882,
      "step": 3800
    },
    {
      "epoch": 0.75,
      "eval_loss": 0.45296812057495117,
      "eval_runtime": 120.8669,
      "eval_samples_per_second": 8.274,
      "eval_steps_per_second": 4.137,
      "step": 3800
    },
    {
      "epoch": 0.75,
      "mmlu_eval_accuracy": 0.4916316804272356,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.18181818181818182,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "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": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "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": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5384615384615384,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1725275041073482,
      "step": 3800
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.4354,
      "step": 3810
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.4431,
      "step": 3820
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.413,
      "step": 3830
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.4054,
      "step": 3840
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.4726,
      "step": 3850
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.4502,
      "step": 3860
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.4204,
      "step": 3870
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.4897,
      "step": 3880
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.4396,
      "step": 3890
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.4791,
      "step": 3900
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.3822,
      "step": 3910
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.417,
      "step": 3920
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.4249,
      "step": 3930
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.438,
      "step": 3940
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.4507,
      "step": 3950
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.4021,
      "step": 3960
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.4033,
      "step": 3970
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.4321,
      "step": 3980
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.4361,
      "step": 3990
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.491,
      "step": 4000
    },
    {
      "epoch": 0.79,
      "eval_loss": 0.44980093836784363,
      "eval_runtime": 120.9775,
      "eval_samples_per_second": 8.266,
      "eval_steps_per_second": 4.133,
      "step": 4000
    },
    {
      "epoch": 0.79,
      "mmlu_eval_accuracy": 0.4921530332624636,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "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": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.5,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2724952631612982,
      "step": 4000
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.4019,
      "step": 4010
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.4406,
      "step": 4020
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.4176,
      "step": 4030
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.4995,
      "step": 4040
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.4523,
      "step": 4050
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.4259,
      "step": 4060
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.4595,
      "step": 4070
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.4739,
      "step": 4080
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.3921,
      "step": 4090
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.3994,
      "step": 4100
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.3855,
      "step": 4110
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.4727,
      "step": 4120
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.459,
      "step": 4130
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.46,
      "step": 4140
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.4552,
      "step": 4150
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.4409,
      "step": 4160
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.4298,
      "step": 4170
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.4585,
      "step": 4180
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.435,
      "step": 4190
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.4208,
      "step": 4200
    },
    {
      "epoch": 0.83,
      "eval_loss": 0.44829413294792175,
      "eval_runtime": 120.9046,
      "eval_samples_per_second": 8.271,
      "eval_steps_per_second": 4.135,
      "step": 4200
    },
    {
      "epoch": 0.83,
      "mmlu_eval_accuracy": 0.4820310798605208,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.09090909090909091,
      "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": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "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": 0.8461538461538461,
      "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": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.30945631651592,
      "step": 4200
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.4431,
      "step": 4210
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.4237,
      "step": 4220
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.463,
      "step": 4230
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.451,
      "step": 4240
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.4268,
      "step": 4250
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.4588,
      "step": 4260
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.4043,
      "step": 4270
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.4268,
      "step": 4280
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.4681,
      "step": 4290
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.473,
      "step": 4300
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.4786,
      "step": 4310
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.455,
      "step": 4320
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.5098,
      "step": 4330
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.402,
      "step": 4340
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.399,
      "step": 4350
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.4216,
      "step": 4360
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.5107,
      "step": 4370
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.4393,
      "step": 4380
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.428,
      "step": 4390
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.4164,
      "step": 4400
    },
    {
      "epoch": 0.87,
      "eval_loss": 0.4475518465042114,
      "eval_runtime": 120.9091,
      "eval_samples_per_second": 8.271,
      "eval_steps_per_second": 4.135,
      "step": 4400
    },
    {
      "epoch": 0.87,
      "mmlu_eval_accuracy": 0.48807648877929727,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "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": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.5,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3588235294117647,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2211916058010595,
      "step": 4400
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.4431,
      "step": 4410
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.4319,
      "step": 4420
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.4553,
      "step": 4430
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.4337,
      "step": 4440
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.4904,
      "step": 4450
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.4788,
      "step": 4460
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.4917,
      "step": 4470
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.4325,
      "step": 4480
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.4707,
      "step": 4490
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.4678,
      "step": 4500
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.4225,
      "step": 4510
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.4345,
      "step": 4520
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.4318,
      "step": 4530
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.4122,
      "step": 4540
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.4812,
      "step": 4550
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.409,
      "step": 4560
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.4247,
      "step": 4570
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.4015,
      "step": 4580
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.4,
      "step": 4590
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.446,
      "step": 4600
    },
    {
      "epoch": 0.91,
      "eval_loss": 0.44572535157203674,
      "eval_runtime": 120.9107,
      "eval_samples_per_second": 8.271,
      "eval_steps_per_second": 4.135,
      "step": 4600
    },
    {
      "epoch": 0.91,
      "mmlu_eval_accuracy": 0.49677704119320965,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "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": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "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": 0.375,
      "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": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.407179198672815,
      "step": 4600
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.4454,
      "step": 4610
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.4219,
      "step": 4620
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.4271,
      "step": 4630
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.4304,
      "step": 4640
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.4175,
      "step": 4650
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.5122,
      "step": 4660
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.4369,
      "step": 4670
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.4295,
      "step": 4680
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.4607,
      "step": 4690
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.4701,
      "step": 4700
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.4415,
      "step": 4710
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.4734,
      "step": 4720
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.4721,
      "step": 4730
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.4545,
      "step": 4740
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.4625,
      "step": 4750
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.4169,
      "step": 4760
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.4053,
      "step": 4770
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.4444,
      "step": 4780
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.4249,
      "step": 4790
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.4892,
      "step": 4800
    },
    {
      "epoch": 0.95,
      "eval_loss": 0.4429953396320343,
      "eval_runtime": 120.8501,
      "eval_samples_per_second": 8.275,
      "eval_steps_per_second": 4.137,
      "step": 4800
    },
    {
      "epoch": 0.95,
      "mmlu_eval_accuracy": 0.42348733384019926,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.3125,
      "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": 0.09090909090909091,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.1875,
      "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": 0.25,
      "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": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.7,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6153846153846154,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6162790697674418,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.33,
      "mmlu_eval_accuracy_nutrition": 0.5454545454545454,
      "mmlu_eval_accuracy_philosophy": 0.38235294117647056,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.28823529411764703,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.75,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.47368421052631576,
      "mmlu_loss": 1.344029222588315,
      "step": 4800
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.4325,
      "step": 4810
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.4688,
      "step": 4820
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.4895,
      "step": 4830
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.3888,
      "step": 4840
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.4396,
      "step": 4850
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.4343,
      "step": 4860
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.5208,
      "step": 4870
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.49,
      "step": 4880
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.4216,
      "step": 4890
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.4288,
      "step": 4900
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.4957,
      "step": 4910
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.4002,
      "step": 4920
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.4209,
      "step": 4930
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.427,
      "step": 4940
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.5133,
      "step": 4950
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.4303,
      "step": 4960
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.3961,
      "step": 4970
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.4162,
      "step": 4980
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.4666,
      "step": 4990
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.448,
      "step": 5000
    },
    {
      "epoch": 0.99,
      "eval_loss": 0.44189587235450745,
      "eval_runtime": 120.8455,
      "eval_samples_per_second": 8.275,
      "eval_steps_per_second": 4.138,
      "step": 5000
    },
    {
      "epoch": 0.99,
      "mmlu_eval_accuracy": 0.4528310916721535,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.09090909090909091,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "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": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.34,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.29411764705882354,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.248610243395788,
      "step": 5000
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.4619,
      "step": 5010
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.4159,
      "step": 5020
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.4595,
      "step": 5030
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.4805,
      "step": 5040
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.4142,
      "step": 5050
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.4738,
      "step": 5060
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.4312,
      "step": 5070
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.3751,
      "step": 5080
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.3727,
      "step": 5090
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.4004,
      "step": 5100
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.321,
      "step": 5110
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.3732,
      "step": 5120
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.4085,
      "step": 5130
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.375,
      "step": 5140
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.3404,
      "step": 5150
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.3563,
      "step": 5160
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.3942,
      "step": 5170
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.3983,
      "step": 5180
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.3584,
      "step": 5190
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.4188,
      "step": 5200
    },
    {
      "epoch": 1.03,
      "eval_loss": 0.4447983503341675,
      "eval_runtime": 120.9555,
      "eval_samples_per_second": 8.268,
      "eval_steps_per_second": 4.134,
      "step": 5200
    },
    {
      "epoch": 1.03,
      "mmlu_eval_accuracy": 0.4518212506875742,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "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": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.09090909090909091,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "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": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
      "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": 0.6086956521739131,
      "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": 0.5454545454545454,
      "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": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.29411764705882354,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.4218421375315744,
      "step": 5200
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.3856,
      "step": 5210
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.3671,
      "step": 5220
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.4026,
      "step": 5230
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.3901,
      "step": 5240
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.3971,
      "step": 5250
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.399,
      "step": 5260
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.3545,
      "step": 5270
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.3822,
      "step": 5280
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.3674,
      "step": 5290
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.4135,
      "step": 5300
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.3672,
      "step": 5310
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.3764,
      "step": 5320
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.3323,
      "step": 5330
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.3272,
      "step": 5340
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.3951,
      "step": 5350
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.3789,
      "step": 5360
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.3835,
      "step": 5370
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.3936,
      "step": 5380
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.3412,
      "step": 5390
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.3681,
      "step": 5400
    },
    {
      "epoch": 1.07,
      "eval_loss": 0.444974809885025,
      "eval_runtime": 120.8685,
      "eval_samples_per_second": 8.273,
      "eval_steps_per_second": 4.137,
      "step": 5400
    },
    {
      "epoch": 1.07,
      "mmlu_eval_accuracy": 0.4654486875039308,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.09090909090909091,
      "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": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "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": 0.7833333333333333,
      "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": 0.6956521739130435,
      "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": 0.5454545454545454,
      "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": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "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": 0.6842105263157895,
      "mmlu_loss": 1.3117361959682743,
      "step": 5400
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.4009,
      "step": 5410
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.4481,
      "step": 5420
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.4177,
      "step": 5430
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.3689,
      "step": 5440
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.3541,
      "step": 5450
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.3723,
      "step": 5460
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.3283,
      "step": 5470
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.3009,
      "step": 5480
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.4244,
      "step": 5490
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.3518,
      "step": 5500
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.333,
      "step": 5510
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.3731,
      "step": 5520
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.3436,
      "step": 5530
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.3466,
      "step": 5540
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.3746,
      "step": 5550
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.377,
      "step": 5560
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.3285,
      "step": 5570
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.3921,
      "step": 5580
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.3245,
      "step": 5590
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.3772,
      "step": 5600
    },
    {
      "epoch": 1.11,
      "eval_loss": 0.4458780288696289,
      "eval_runtime": 120.9607,
      "eval_samples_per_second": 8.267,
      "eval_steps_per_second": 4.134,
      "step": 5600
    },
    {
      "epoch": 1.11,
      "mmlu_eval_accuracy": 0.4782084359627476,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "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": 0.2727272727272727,
      "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": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.84,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1528637884257669,
      "step": 5600
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.407,
      "step": 5610
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.3313,
      "step": 5620
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.3899,
      "step": 5630
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.3699,
      "step": 5640
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.3993,
      "step": 5650
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.4171,
      "step": 5660
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.4057,
      "step": 5670
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.43,
      "step": 5680
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.3676,
      "step": 5690
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.3765,
      "step": 5700
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.3842,
      "step": 5710
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.4444,
      "step": 5720
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.3708,
      "step": 5730
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.3741,
      "step": 5740
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.3913,
      "step": 5750
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.3688,
      "step": 5760
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.3822,
      "step": 5770
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.3647,
      "step": 5780
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.3952,
      "step": 5790
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.4012,
      "step": 5800
    },
    {
      "epoch": 1.15,
      "eval_loss": 0.448940634727478,
      "eval_runtime": 120.8993,
      "eval_samples_per_second": 8.271,
      "eval_steps_per_second": 4.136,
      "step": 5800
    },
    {
      "epoch": 1.15,
      "mmlu_eval_accuracy": 0.47228054356366644,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "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": 0.2727272727272727,
      "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": 0.34615384615384615,
      "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": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.21875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1274879405143678,
      "step": 5800
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.3832,
      "step": 5810
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.3735,
      "step": 5820
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.3377,
      "step": 5830
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.3043,
      "step": 5840
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.4161,
      "step": 5850
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.3896,
      "step": 5860
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.3816,
      "step": 5870
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.4081,
      "step": 5880
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.393,
      "step": 5890
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.3588,
      "step": 5900
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.3957,
      "step": 5910
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.3748,
      "step": 5920
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.3642,
      "step": 5930
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.417,
      "step": 5940
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.3838,
      "step": 5950
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.3685,
      "step": 5960
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.3826,
      "step": 5970
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.3439,
      "step": 5980
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.3645,
      "step": 5990
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.3679,
      "step": 6000
    },
    {
      "epoch": 1.18,
      "eval_loss": 0.4468071758747101,
      "eval_runtime": 120.9853,
      "eval_samples_per_second": 8.265,
      "eval_steps_per_second": 4.133,
      "step": 6000
    },
    {
      "epoch": 1.18,
      "mmlu_eval_accuracy": 0.47902924053887225,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "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": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.3537822210524164,
      "step": 6000
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.3572,
      "step": 6010
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.4093,
      "step": 6020
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.4071,
      "step": 6030
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.4245,
      "step": 6040
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.3755,
      "step": 6050
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.3693,
      "step": 6060
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.3972,
      "step": 6070
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.39,
      "step": 6080
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.4284,
      "step": 6090
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.3538,
      "step": 6100
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.3566,
      "step": 6110
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.3398,
      "step": 6120
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.4341,
      "step": 6130
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.3967,
      "step": 6140
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.3533,
      "step": 6150
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.3826,
      "step": 6160
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.387,
      "step": 6170
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.4174,
      "step": 6180
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.4021,
      "step": 6190
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.3649,
      "step": 6200
    },
    {
      "epoch": 1.22,
      "eval_loss": 0.4450535178184509,
      "eval_runtime": 120.974,
      "eval_samples_per_second": 8.266,
      "eval_steps_per_second": 4.133,
      "step": 6200
    },
    {
      "epoch": 1.22,
      "mmlu_eval_accuracy": 0.49171108594535207,
      "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_law": 0.3588235294117647,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1403136445554678,
      "step": 6200
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.3599,
      "step": 6210
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.3859,
      "step": 6220
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.3573,
      "step": 6230
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.3245,
      "step": 6240
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.4243,
      "step": 6250
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.3974,
      "step": 6260
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.4257,
      "step": 6270
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.3506,
      "step": 6280
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.4198,
      "step": 6290
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.3741,
      "step": 6300
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.3465,
      "step": 6310
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.3988,
      "step": 6320
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.3865,
      "step": 6330
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.3699,
      "step": 6340
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.3522,
      "step": 6350
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.4129,
      "step": 6360
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.358,
      "step": 6370
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.4189,
      "step": 6380
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.399,
      "step": 6390
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.3877,
      "step": 6400
    },
    {
      "epoch": 1.26,
      "eval_loss": 0.4440629780292511,
      "eval_runtime": 120.9414,
      "eval_samples_per_second": 8.268,
      "eval_steps_per_second": 4.134,
      "step": 6400
    },
    {
      "epoch": 1.26,
      "mmlu_eval_accuracy": 0.4867509959392518,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "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": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.1875,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.3,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "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": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.0828519074863927,
      "step": 6400
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.3989,
      "step": 6410
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.3786,
      "step": 6420
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.3326,
      "step": 6430
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.3602,
      "step": 6440
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.414,
      "step": 6450
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.3756,
      "step": 6460
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.3688,
      "step": 6470
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.3732,
      "step": 6480
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.3914,
      "step": 6490
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.357,
      "step": 6500
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.343,
      "step": 6510
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.3942,
      "step": 6520
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.373,
      "step": 6530
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.3962,
      "step": 6540
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.407,
      "step": 6550
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.3611,
      "step": 6560
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.3557,
      "step": 6570
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.4006,
      "step": 6580
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.3899,
      "step": 6590
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.4637,
      "step": 6600
    },
    {
      "epoch": 1.3,
      "eval_loss": 0.44296130537986755,
      "eval_runtime": 120.9444,
      "eval_samples_per_second": 8.268,
      "eval_steps_per_second": 4.134,
      "step": 6600
    },
    {
      "epoch": 1.3,
      "mmlu_eval_accuracy": 0.5059018732754555,
      "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": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.5,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.178827084542566,
      "step": 6600
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.3946,
      "step": 6610
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.3958,
      "step": 6620
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.3441,
      "step": 6630
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.3368,
      "step": 6640
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.3992,
      "step": 6650
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.359,
      "step": 6660
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.4192,
      "step": 6670
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.3531,
      "step": 6680
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.3698,
      "step": 6690
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.4178,
      "step": 6700
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.3839,
      "step": 6710
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.3901,
      "step": 6720
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.4016,
      "step": 6730
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.4134,
      "step": 6740
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.3701,
      "step": 6750
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.3915,
      "step": 6760
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.312,
      "step": 6770
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.4279,
      "step": 6780
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.4226,
      "step": 6790
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.415,
      "step": 6800
    },
    {
      "epoch": 1.34,
      "eval_loss": 0.44151541590690613,
      "eval_runtime": 120.9028,
      "eval_samples_per_second": 8.271,
      "eval_steps_per_second": 4.136,
      "step": 6800
    },
    {
      "epoch": 1.34,
      "mmlu_eval_accuracy": 0.5084804815045342,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.8181818181818182,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.193184396266626,
      "step": 6800
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.4017,
      "step": 6810
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.3976,
      "step": 6820
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.3451,
      "step": 6830
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.3789,
      "step": 6840
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.3654,
      "step": 6850
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.4088,
      "step": 6860
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.3614,
      "step": 6870
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.4376,
      "step": 6880
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.4113,
      "step": 6890
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.384,
      "step": 6900
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.3689,
      "step": 6910
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.3565,
      "step": 6920
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.3899,
      "step": 6930
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.392,
      "step": 6940
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.3805,
      "step": 6950
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.3245,
      "step": 6960
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.3815,
      "step": 6970
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.353,
      "step": 6980
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.3542,
      "step": 6990
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.4175,
      "step": 7000
    },
    {
      "epoch": 1.38,
      "eval_loss": 0.44231361150741577,
      "eval_runtime": 120.8967,
      "eval_samples_per_second": 8.272,
      "eval_steps_per_second": 4.136,
      "step": 7000
    },
    {
      "epoch": 1.38,
      "mmlu_eval_accuracy": 0.5054198185434623,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.625,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "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": 0.6956521739130435,
      "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": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.1865613444200072,
      "step": 7000
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.4169,
      "step": 7010
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.3836,
      "step": 7020
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.3449,
      "step": 7030
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.4059,
      "step": 7040
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.3668,
      "step": 7050
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.4043,
      "step": 7060
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.3529,
      "step": 7070
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.3659,
      "step": 7080
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.4007,
      "step": 7090
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.4162,
      "step": 7100
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.3846,
      "step": 7110
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.4277,
      "step": 7120
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.4338,
      "step": 7130
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.3412,
      "step": 7140
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.4108,
      "step": 7150
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.4078,
      "step": 7160
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.3698,
      "step": 7170
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.4155,
      "step": 7180
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.3653,
      "step": 7190
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.3598,
      "step": 7200
    },
    {
      "epoch": 1.42,
      "eval_loss": 0.43925899267196655,
      "eval_runtime": 120.9998,
      "eval_samples_per_second": 8.264,
      "eval_steps_per_second": 4.132,
      "step": 7200
    },
    {
      "epoch": 1.42,
      "mmlu_eval_accuracy": 0.48394951792577023,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "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": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.5454545454545454,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "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": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.127036721486958,
      "step": 7200
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.3588,
      "step": 7210
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.3902,
      "step": 7220
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.3806,
      "step": 7230
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.3985,
      "step": 7240
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.3945,
      "step": 7250
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.4605,
      "step": 7260
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.3761,
      "step": 7270
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.3667,
      "step": 7280
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.3682,
      "step": 7290
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.3361,
      "step": 7300
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.3685,
      "step": 7310
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.3448,
      "step": 7320
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.3498,
      "step": 7330
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.3714,
      "step": 7340
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.3915,
      "step": 7350
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.3867,
      "step": 7360
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.3838,
      "step": 7370
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.3923,
      "step": 7380
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.3739,
      "step": 7390
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.4029,
      "step": 7400
    },
    {
      "epoch": 1.46,
      "eval_loss": 0.43981724977493286,
      "eval_runtime": 121.0098,
      "eval_samples_per_second": 8.264,
      "eval_steps_per_second": 4.132,
      "step": 7400
    },
    {
      "epoch": 1.46,
      "mmlu_eval_accuracy": 0.47466190250303497,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "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": 0.36363636363636365,
      "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": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1439847480846137,
      "step": 7400
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.3562,
      "step": 7410
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.3582,
      "step": 7420
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.4188,
      "step": 7430
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.3889,
      "step": 7440
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.3712,
      "step": 7450
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.4247,
      "step": 7460
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.3805,
      "step": 7470
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.3322,
      "step": 7480
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.3859,
      "step": 7490
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.3529,
      "step": 7500
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.3412,
      "step": 7510
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.4411,
      "step": 7520
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.3807,
      "step": 7530
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.3794,
      "step": 7540
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.355,
      "step": 7550
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.404,
      "step": 7560
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.4042,
      "step": 7570
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.3696,
      "step": 7580
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.3807,
      "step": 7590
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.4191,
      "step": 7600
    },
    {
      "epoch": 1.5,
      "eval_loss": 0.43704837560653687,
      "eval_runtime": 120.9943,
      "eval_samples_per_second": 8.265,
      "eval_steps_per_second": 4.132,
      "step": 7600
    },
    {
      "epoch": 1.5,
      "mmlu_eval_accuracy": 0.4864390912539841,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.625,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "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": 0.45454545454545453,
      "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": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "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": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.202333774018848,
      "step": 7600
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.4169,
      "step": 7610
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.4004,
      "step": 7620
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.3925,
      "step": 7630
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.3838,
      "step": 7640
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.339,
      "step": 7650
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.3929,
      "step": 7660
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.4526,
      "step": 7670
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.3797,
      "step": 7680
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.4072,
      "step": 7690
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.3355,
      "step": 7700
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.3736,
      "step": 7710
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.3589,
      "step": 7720
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.3284,
      "step": 7730
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.3473,
      "step": 7740
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.3735,
      "step": 7750
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.3869,
      "step": 7760
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.367,
      "step": 7770
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.4084,
      "step": 7780
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.3827,
      "step": 7790
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.4602,
      "step": 7800
    },
    {
      "epoch": 1.54,
      "eval_loss": 0.4351891577243805,
      "eval_runtime": 121.0193,
      "eval_samples_per_second": 8.263,
      "eval_steps_per_second": 4.132,
      "step": 7800
    },
    {
      "epoch": 1.54,
      "mmlu_eval_accuracy": 0.48877828979099663,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "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": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "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": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.27906976744186046,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.0244236340838686,
      "step": 7800
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.3771,
      "step": 7810
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.4082,
      "step": 7820
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.4122,
      "step": 7830
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.4006,
      "step": 7840
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.4035,
      "step": 7850
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.3887,
      "step": 7860
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.3624,
      "step": 7870
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.3508,
      "step": 7880
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.3463,
      "step": 7890
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.3644,
      "step": 7900
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.428,
      "step": 7910
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.3583,
      "step": 7920
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.3895,
      "step": 7930
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.379,
      "step": 7940
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.3231,
      "step": 7950
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.3399,
      "step": 7960
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.4171,
      "step": 7970
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.4399,
      "step": 7980
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.3888,
      "step": 7990
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.3381,
      "step": 8000
    },
    {
      "epoch": 1.58,
      "eval_loss": 0.43523940443992615,
      "eval_runtime": 120.9711,
      "eval_samples_per_second": 8.266,
      "eval_steps_per_second": 4.133,
      "step": 8000
    },
    {
      "epoch": 1.58,
      "mmlu_eval_accuracy": 0.5000652378894127,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "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": 0.32558139534883723,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
      "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": 0.7692307692307693,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "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": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "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": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.1098531644120229,
      "step": 8000
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.3125,
      "step": 8010
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.312,
      "step": 8020
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.3938,
      "step": 8030
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.363,
      "step": 8040
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.3951,
      "step": 8050
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.3938,
      "step": 8060
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.3905,
      "step": 8070
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.3564,
      "step": 8080
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.4336,
      "step": 8090
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.3662,
      "step": 8100
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.3913,
      "step": 8110
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.3552,
      "step": 8120
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.3672,
      "step": 8130
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.4189,
      "step": 8140
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.4258,
      "step": 8150
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.3944,
      "step": 8160
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.3819,
      "step": 8170
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.326,
      "step": 8180
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.3583,
      "step": 8190
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.3877,
      "step": 8200
    },
    {
      "epoch": 1.62,
      "eval_loss": 0.43449267745018005,
      "eval_runtime": 120.9418,
      "eval_samples_per_second": 8.268,
      "eval_steps_per_second": 4.134,
      "step": 8200
    },
    {
      "epoch": 1.62,
      "mmlu_eval_accuracy": 0.49162575918435,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.5217391304347826,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.1440130493970826,
      "step": 8200
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.399,
      "step": 8210
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.3728,
      "step": 8220
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.3855,
      "step": 8230
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.3794,
      "step": 8240
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.3679,
      "step": 8250
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.3555,
      "step": 8260
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.33,
      "step": 8270
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.3658,
      "step": 8280
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.3872,
      "step": 8290
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.3699,
      "step": 8300
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.3732,
      "step": 8310
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.3534,
      "step": 8320
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.3955,
      "step": 8330
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.3624,
      "step": 8340
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.3391,
      "step": 8350
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.3551,
      "step": 8360
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.3488,
      "step": 8370
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.39,
      "step": 8380
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.4008,
      "step": 8390
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.3572,
      "step": 8400
    },
    {
      "epoch": 1.66,
      "eval_loss": 0.43166613578796387,
      "eval_runtime": 121.0703,
      "eval_samples_per_second": 8.26,
      "eval_steps_per_second": 4.13,
      "step": 8400
    },
    {
      "epoch": 1.66,
      "mmlu_eval_accuracy": 0.4921472652780345,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.201037272071714,
      "step": 8400
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.3819,
      "step": 8410
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.4056,
      "step": 8420
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.4169,
      "step": 8430
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.4646,
      "step": 8440
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.3713,
      "step": 8450
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.3924,
      "step": 8460
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.3988,
      "step": 8470
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.4092,
      "step": 8480
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.3528,
      "step": 8490
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.3653,
      "step": 8500
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.3953,
      "step": 8510
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.3782,
      "step": 8520
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.3816,
      "step": 8530
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.4093,
      "step": 8540
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.3487,
      "step": 8550
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.4031,
      "step": 8560
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.3905,
      "step": 8570
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.3379,
      "step": 8580
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.3251,
      "step": 8590
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.3807,
      "step": 8600
    },
    {
      "epoch": 1.7,
      "eval_loss": 0.4307052195072174,
      "eval_runtime": 120.921,
      "eval_samples_per_second": 8.27,
      "eval_steps_per_second": 4.135,
      "step": 8600
    },
    {
      "epoch": 1.7,
      "mmlu_eval_accuracy": 0.48902984209003103,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "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": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "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": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1535750755908907,
      "step": 8600
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.3005,
      "step": 8610
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.3911,
      "step": 8620
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.3333,
      "step": 8630
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.3275,
      "step": 8640
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.3681,
      "step": 8650
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.365,
      "step": 8660
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.3958,
      "step": 8670
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.3902,
      "step": 8680
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.3519,
      "step": 8690
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.3752,
      "step": 8700
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.4066,
      "step": 8710
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.407,
      "step": 8720
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.3423,
      "step": 8730
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.3839,
      "step": 8740
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.4293,
      "step": 8750
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.3772,
      "step": 8760
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.3927,
      "step": 8770
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.3952,
      "step": 8780
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.3434,
      "step": 8790
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.3012,
      "step": 8800
    },
    {
      "epoch": 1.74,
      "eval_loss": 0.4287540912628174,
      "eval_runtime": 120.9927,
      "eval_samples_per_second": 8.265,
      "eval_steps_per_second": 4.132,
      "step": 8800
    },
    {
      "epoch": 1.74,
      "mmlu_eval_accuracy": 0.47671443145279063,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "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": 0.07142857142857142,
      "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": 0.7727272727272727,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.8,
      "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": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.2309164613407524,
      "step": 8800
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.3738,
      "step": 8810
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.3638,
      "step": 8820
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.406,
      "step": 8830
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.3632,
      "step": 8840
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.3826,
      "step": 8850
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.3817,
      "step": 8860
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.3471,
      "step": 8870
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.3944,
      "step": 8880
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.4503,
      "step": 8890
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.3523,
      "step": 8900
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.3745,
      "step": 8910
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.4025,
      "step": 8920
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.3678,
      "step": 8930
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.3248,
      "step": 8940
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.3688,
      "step": 8950
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.3556,
      "step": 8960
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.3366,
      "step": 8970
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.3883,
      "step": 8980
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.3661,
      "step": 8990
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.3394,
      "step": 9000
    },
    {
      "epoch": 1.78,
      "eval_loss": 0.42867738008499146,
      "eval_runtime": 120.9321,
      "eval_samples_per_second": 8.269,
      "eval_steps_per_second": 4.135,
      "step": 9000
    },
    {
      "epoch": 1.78,
      "mmlu_eval_accuracy": 0.49017157423524627,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "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": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.5454545454545454,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 1.0,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.275885030666157,
      "step": 9000
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.3498,
      "step": 9010
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.3482,
      "step": 9020
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.3994,
      "step": 9030
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.3459,
      "step": 9040
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.4181,
      "step": 9050
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.4244,
      "step": 9060
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.3617,
      "step": 9070
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.3783,
      "step": 9080
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.3786,
      "step": 9090
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.3845,
      "step": 9100
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.388,
      "step": 9110
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.3617,
      "step": 9120
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.3468,
      "step": 9130
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.3784,
      "step": 9140
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.3768,
      "step": 9150
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.3638,
      "step": 9160
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.3638,
      "step": 9170
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.4116,
      "step": 9180
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.3764,
      "step": 9190
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002,
      "loss": 0.3771,
      "step": 9200
    },
    {
      "epoch": 1.82,
      "eval_loss": 0.4287903308868408,
      "eval_runtime": 120.9286,
      "eval_samples_per_second": 8.269,
      "eval_steps_per_second": 4.135,
      "step": 9200
    },
    {
      "epoch": 1.82,
      "mmlu_eval_accuracy": 0.48421812063771447,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "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": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "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": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "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": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.3082739823636436,
      "step": 9200
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002,
      "loss": 0.3359,
      "step": 9210
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002,
      "loss": 0.373,
      "step": 9220
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002,
      "loss": 0.3199,
      "step": 9230
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002,
      "loss": 0.3683,
      "step": 9240
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.4138,
      "step": 9250
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.3295,
      "step": 9260
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.4039,
      "step": 9270
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.3871,
      "step": 9280
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0002,
      "loss": 0.4008,
      "step": 9290
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0002,
      "loss": 0.3854,
      "step": 9300
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0002,
      "loss": 0.4233,
      "step": 9310
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0002,
      "loss": 0.3633,
      "step": 9320
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0002,
      "loss": 0.4346,
      "step": 9330
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0002,
      "loss": 0.3353,
      "step": 9340
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002,
      "loss": 0.3984,
      "step": 9350
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002,
      "loss": 0.3812,
      "step": 9360
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002,
      "loss": 0.3765,
      "step": 9370
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002,
      "loss": 0.3911,
      "step": 9380
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002,
      "loss": 0.401,
      "step": 9390
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002,
      "loss": 0.3898,
      "step": 9400
    },
    {
      "epoch": 1.86,
      "eval_loss": 0.4289686679840088,
      "eval_runtime": 120.9149,
      "eval_samples_per_second": 8.27,
      "eval_steps_per_second": 4.135,
      "step": 9400
    },
    {
      "epoch": 1.86,
      "mmlu_eval_accuracy": 0.49087852841885354,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "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": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.1834476133784178,
      "step": 9400
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002,
      "loss": 0.423,
      "step": 9410
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002,
      "loss": 0.3717,
      "step": 9420
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002,
      "loss": 0.3912,
      "step": 9430
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002,
      "loss": 0.3826,
      "step": 9440
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0002,
      "loss": 0.4101,
      "step": 9450
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0002,
      "loss": 0.3503,
      "step": 9460
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0002,
      "loss": 0.4014,
      "step": 9470
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0002,
      "loss": 0.3957,
      "step": 9480
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0002,
      "loss": 0.3602,
      "step": 9490
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0002,
      "loss": 0.3585,
      "step": 9500
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0002,
      "loss": 0.3648,
      "step": 9510
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0002,
      "loss": 0.4045,
      "step": 9520
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0002,
      "loss": 0.4431,
      "step": 9530
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0002,
      "loss": 0.352,
      "step": 9540
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.3462,
      "step": 9550
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.3775,
      "step": 9560
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.4117,
      "step": 9570
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.3857,
      "step": 9580
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002,
      "loss": 0.3965,
      "step": 9590
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0002,
      "loss": 0.3415,
      "step": 9600
    },
    {
      "epoch": 1.9,
      "eval_loss": 0.4280742108821869,
      "eval_runtime": 120.9619,
      "eval_samples_per_second": 8.267,
      "eval_steps_per_second": 4.134,
      "step": 9600
    },
    {
      "epoch": 1.9,
      "mmlu_eval_accuracy": 0.48945860845923983,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.75,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.332587436074999,
      "step": 9600
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0002,
      "loss": 0.4047,
      "step": 9610
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0002,
      "loss": 0.4208,
      "step": 9620
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0002,
      "loss": 0.3615,
      "step": 9630
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0002,
      "loss": 0.3888,
      "step": 9640
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002,
      "loss": 0.4164,
      "step": 9650
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002,
      "loss": 0.3407,
      "step": 9660
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002,
      "loss": 0.3701,
      "step": 9670
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002,
      "loss": 0.355,
      "step": 9680
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002,
      "loss": 0.3631,
      "step": 9690
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.3251,
      "step": 9700
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.3726,
      "step": 9710
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.4086,
      "step": 9720
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.3816,
      "step": 9730
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.3532,
      "step": 9740
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0002,
      "loss": 0.356,
      "step": 9750
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0002,
      "loss": 0.3322,
      "step": 9760
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0002,
      "loss": 0.3646,
      "step": 9770
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0002,
      "loss": 0.3488,
      "step": 9780
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0002,
      "loss": 0.3924,
      "step": 9790
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0002,
      "loss": 0.3421,
      "step": 9800
    },
    {
      "epoch": 1.93,
      "eval_loss": 0.4274959862232208,
      "eval_runtime": 120.9327,
      "eval_samples_per_second": 8.269,
      "eval_steps_per_second": 4.135,
      "step": 9800
    },
    {
      "epoch": 1.93,
      "mmlu_eval_accuracy": 0.4850290034468293,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "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": 0.14285714285714285,
      "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": 0.6666666666666666,
      "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": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.75,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6162790697674418,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_law": 0.3588235294117647,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.3235855651508734,
      "step": 9800
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0002,
      "loss": 0.353,
      "step": 9810
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0002,
      "loss": 0.3133,
      "step": 9820
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0002,
      "loss": 0.3535,
      "step": 9830
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0002,
      "loss": 0.3927,
      "step": 9840
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0002,
      "loss": 0.3617,
      "step": 9850
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0002,
      "loss": 0.3543,
      "step": 9860
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0002,
      "loss": 0.4171,
      "step": 9870
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0002,
      "loss": 0.3953,
      "step": 9880
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0002,
      "loss": 0.403,
      "step": 9890
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0002,
      "loss": 0.4396,
      "step": 9900
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.3313,
      "step": 9910
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.388,
      "step": 9920
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.3691,
      "step": 9930
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.3759,
      "step": 9940
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002,
      "loss": 0.3811,
      "step": 9950
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0002,
      "loss": 0.4153,
      "step": 9960
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0002,
      "loss": 0.3673,
      "step": 9970
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0002,
      "loss": 0.3998,
      "step": 9980
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0002,
      "loss": 0.4362,
      "step": 9990
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0002,
      "loss": 0.3691,
      "step": 10000
    },
    {
      "epoch": 1.97,
      "eval_loss": 0.4259847402572632,
      "eval_runtime": 120.9609,
      "eval_samples_per_second": 8.267,
      "eval_steps_per_second": 4.134,
      "step": 10000
    },
    {
      "epoch": 1.97,
      "mmlu_eval_accuracy": 0.47760434329067897,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "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": 0.4090909090909091,
      "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": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "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": 0.3125,
      "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": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "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": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 1.0,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "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": 0.7894736842105263,
      "mmlu_loss": 1.324429467748413,
      "step": 10000
    }
  ],
  "max_steps": 10000,
  "num_train_epochs": 2,
  "total_flos": 9.247330990328709e+17,
  "trial_name": null,
  "trial_params": null
}