{
  "best_metric": 0.7169492244720459,
  "best_model_checkpoint": "experts/expert-13/checkpoint-8600",
  "epoch": 1.8074821353509878,
  "global_step": 8600,
  "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": 0.9187,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0002,
      "loss": 0.8703,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8582,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8973,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8188,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.8114,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 0.7796,
      "step": 70
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.8671,
      "step": 80
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.8863,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.815,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 0.8255,
      "step": 110
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8272,
      "step": 120
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8148,
      "step": 130
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8542,
      "step": 140
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8592,
      "step": 150
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002,
      "loss": 0.8743,
      "step": 160
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.7967,
      "step": 170
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.7863,
      "step": 180
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.8397,
      "step": 190
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.7508,
      "step": 200
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.8083679676055908,
      "eval_runtime": 68.466,
      "eval_samples_per_second": 14.606,
      "eval_steps_per_second": 7.303,
      "step": 200
    },
    {
      "epoch": 0.04,
      "mmlu_eval_accuracy": 0.4843923123312093,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "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": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.48628691881506,
      "step": 200
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 0.8043,
      "step": 210
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.8064,
      "step": 220
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7611,
      "step": 230
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.8212,
      "step": 240
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.8044,
      "step": 250
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002,
      "loss": 0.7862,
      "step": 260
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8106,
      "step": 270
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.7954,
      "step": 280
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8153,
      "step": 290
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002,
      "loss": 0.8235,
      "step": 300
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8087,
      "step": 310
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.7824,
      "step": 320
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8198,
      "step": 330
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.8288,
      "step": 340
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002,
      "loss": 0.797,
      "step": 350
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.7599,
      "step": 360
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.8156,
      "step": 370
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.8534,
      "step": 380
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.7712,
      "step": 390
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002,
      "loss": 0.7555,
      "step": 400
    },
    {
      "epoch": 0.08,
      "eval_loss": 0.7937927842140198,
      "eval_runtime": 68.5823,
      "eval_samples_per_second": 14.581,
      "eval_steps_per_second": 7.291,
      "step": 400
    },
    {
      "epoch": 0.08,
      "mmlu_eval_accuracy": 0.4820017142234064,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "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": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.21,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "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": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "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": 0.6842105263157895,
      "mmlu_loss": 1.500225733527938,
      "step": 400
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.8338,
      "step": 410
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.7197,
      "step": 420
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.8088,
      "step": 430
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.849,
      "step": 440
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 0.7808,
      "step": 450
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.8022,
      "step": 460
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.7923,
      "step": 470
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.8324,
      "step": 480
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002,
      "loss": 0.7889,
      "step": 490
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.821,
      "step": 500
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.8206,
      "step": 510
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.8063,
      "step": 520
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.7997,
      "step": 530
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002,
      "loss": 0.7582,
      "step": 540
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.9064,
      "step": 550
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.8074,
      "step": 560
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.8019,
      "step": 570
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.8125,
      "step": 580
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002,
      "loss": 0.7813,
      "step": 590
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.7835,
      "step": 600
    },
    {
      "epoch": 0.13,
      "eval_loss": 0.7827013731002808,
      "eval_runtime": 68.7211,
      "eval_samples_per_second": 14.552,
      "eval_steps_per_second": 7.276,
      "step": 600
    },
    {
      "epoch": 0.13,
      "mmlu_eval_accuracy": 0.4822273416458045,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.18181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.42857142857142855,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "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": 0.6363636363636364,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.21,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.3917966029818314,
      "step": 600
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.7644,
      "step": 610
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.8198,
      "step": 620
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.6895,
      "step": 630
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002,
      "loss": 0.7964,
      "step": 640
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.8632,
      "step": 650
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.7955,
      "step": 660
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.769,
      "step": 670
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0002,
      "loss": 0.8685,
      "step": 680
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.8812,
      "step": 690
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.8451,
      "step": 700
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.7944,
      "step": 710
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.7648,
      "step": 720
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0002,
      "loss": 0.7871,
      "step": 730
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.7852,
      "step": 740
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8073,
      "step": 750
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8123,
      "step": 760
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.769,
      "step": 770
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0002,
      "loss": 0.8062,
      "step": 780
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.7766,
      "step": 790
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.7507,
      "step": 800
    },
    {
      "epoch": 0.17,
      "eval_loss": 0.7727674841880798,
      "eval_runtime": 68.7464,
      "eval_samples_per_second": 14.546,
      "eval_steps_per_second": 7.273,
      "step": 800
    },
    {
      "epoch": 0.17,
      "mmlu_eval_accuracy": 0.4910291918783051,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "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": 0.36363636363636365,
      "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": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "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": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.3918722461782609,
      "step": 800
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.8315,
      "step": 810
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.7658,
      "step": 820
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002,
      "loss": 0.7532,
      "step": 830
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.746,
      "step": 840
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.7351,
      "step": 850
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.7981,
      "step": 860
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.7417,
      "step": 870
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 0.7517,
      "step": 880
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.8346,
      "step": 890
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.7979,
      "step": 900
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.787,
      "step": 910
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0002,
      "loss": 0.8434,
      "step": 920
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.7979,
      "step": 930
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.8029,
      "step": 940
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.768,
      "step": 950
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.7779,
      "step": 960
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002,
      "loss": 0.7882,
      "step": 970
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.7942,
      "step": 980
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.771,
      "step": 990
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.7953,
      "step": 1000
    },
    {
      "epoch": 0.21,
      "eval_loss": 0.7682561278343201,
      "eval_runtime": 68.3189,
      "eval_samples_per_second": 14.637,
      "eval_steps_per_second": 7.319,
      "step": 1000
    },
    {
      "epoch": 0.21,
      "mmlu_eval_accuracy": 0.49115442468360637,
      "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "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": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "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": 0.6923076923076923,
      "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": 0.45454545454545453,
      "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": 0.19,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5362318840579711,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.354807756315946,
      "step": 1000
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.8039,
      "step": 1010
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002,
      "loss": 0.7947,
      "step": 1020
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.8262,
      "step": 1030
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.7898,
      "step": 1040
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.7892,
      "step": 1050
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.7291,
      "step": 1060
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002,
      "loss": 0.7667,
      "step": 1070
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.7502,
      "step": 1080
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.778,
      "step": 1090
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.7508,
      "step": 1100
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002,
      "loss": 0.8401,
      "step": 1110
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.8034,
      "step": 1120
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.7376,
      "step": 1130
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.8119,
      "step": 1140
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.7607,
      "step": 1150
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002,
      "loss": 0.7693,
      "step": 1160
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.8165,
      "step": 1170
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.8235,
      "step": 1180
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.76,
      "step": 1190
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.7601,
      "step": 1200
    },
    {
      "epoch": 0.25,
      "eval_loss": 0.7659562826156616,
      "eval_runtime": 68.3632,
      "eval_samples_per_second": 14.628,
      "eval_steps_per_second": 7.314,
      "step": 1200
    },
    {
      "epoch": 0.25,
      "mmlu_eval_accuracy": 0.49680947680174564,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "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": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "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": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.2,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4655385180412945,
      "step": 1200
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.8263,
      "step": 1210
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.7299,
      "step": 1220
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.7767,
      "step": 1230
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.7466,
      "step": 1240
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.7512,
      "step": 1250
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002,
      "loss": 0.8219,
      "step": 1260
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.7257,
      "step": 1270
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.7797,
      "step": 1280
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.7849,
      "step": 1290
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0002,
      "loss": 0.7739,
      "step": 1300
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.7542,
      "step": 1310
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.8234,
      "step": 1320
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.7476,
      "step": 1330
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.8598,
      "step": 1340
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0002,
      "loss": 0.7603,
      "step": 1350
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.772,
      "step": 1360
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.761,
      "step": 1370
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.7371,
      "step": 1380
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.7722,
      "step": 1390
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0002,
      "loss": 0.7693,
      "step": 1400
    },
    {
      "epoch": 0.29,
      "eval_loss": 0.7587528824806213,
      "eval_runtime": 68.3623,
      "eval_samples_per_second": 14.628,
      "eval_steps_per_second": 7.314,
      "step": 1400
    },
    {
      "epoch": 0.29,
      "mmlu_eval_accuracy": 0.4867351915777268,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.1928723966873034,
      "step": 1400
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.7502,
      "step": 1410
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.7828,
      "step": 1420
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.8183,
      "step": 1430
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.7737,
      "step": 1440
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 0.7304,
      "step": 1450
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.771,
      "step": 1460
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.8128,
      "step": 1470
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.7552,
      "step": 1480
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 0.7292,
      "step": 1490
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.7328,
      "step": 1500
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.7341,
      "step": 1510
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.8096,
      "step": 1520
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.7517,
      "step": 1530
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0002,
      "loss": 0.7468,
      "step": 1540
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7329,
      "step": 1550
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7753,
      "step": 1560
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7043,
      "step": 1570
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7866,
      "step": 1580
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002,
      "loss": 0.7029,
      "step": 1590
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.731,
      "step": 1600
    },
    {
      "epoch": 0.34,
      "eval_loss": 0.7554829716682434,
      "eval_runtime": 68.5087,
      "eval_samples_per_second": 14.597,
      "eval_steps_per_second": 7.298,
      "step": 1600
    },
    {
      "epoch": 0.34,
      "mmlu_eval_accuracy": 0.4845186912752572,
      "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": 0.5,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "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": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.22,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.361777840177322,
      "step": 1600
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.7678,
      "step": 1610
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.792,
      "step": 1620
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.7282,
      "step": 1630
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002,
      "loss": 0.7891,
      "step": 1640
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.7668,
      "step": 1650
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.7327,
      "step": 1660
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.7622,
      "step": 1670
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002,
      "loss": 0.7191,
      "step": 1680
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7678,
      "step": 1690
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7589,
      "step": 1700
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.7348,
      "step": 1710
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.8115,
      "step": 1720
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002,
      "loss": 0.784,
      "step": 1730
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.7988,
      "step": 1740
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.7881,
      "step": 1750
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.7729,
      "step": 1760
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.7251,
      "step": 1770
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002,
      "loss": 0.7676,
      "step": 1780
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.7724,
      "step": 1790
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.7396,
      "step": 1800
    },
    {
      "epoch": 0.38,
      "eval_loss": 0.7520008087158203,
      "eval_runtime": 68.9848,
      "eval_samples_per_second": 14.496,
      "eval_steps_per_second": 7.248,
      "step": 1800
    },
    {
      "epoch": 0.38,
      "mmlu_eval_accuracy": 0.4994070115210143,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.53125,
      "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": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.9,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "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": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "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": 1.2931012510475228,
      "step": 1800
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.8021,
      "step": 1810
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.8055,
      "step": 1820
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002,
      "loss": 0.7902,
      "step": 1830
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.7743,
      "step": 1840
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.7817,
      "step": 1850
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.755,
      "step": 1860
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002,
      "loss": 0.7329,
      "step": 1870
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.794,
      "step": 1880
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.7722,
      "step": 1890
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.7399,
      "step": 1900
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.7755,
      "step": 1910
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0002,
      "loss": 0.7996,
      "step": 1920
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.8096,
      "step": 1930
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.7545,
      "step": 1940
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.8003,
      "step": 1950
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.7321,
      "step": 1960
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002,
      "loss": 0.7846,
      "step": 1970
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.7836,
      "step": 1980
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.7531,
      "step": 1990
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.7589,
      "step": 2000
    },
    {
      "epoch": 0.42,
      "eval_loss": 0.7525200247764587,
      "eval_runtime": 68.3555,
      "eval_samples_per_second": 14.629,
      "eval_steps_per_second": 7.315,
      "step": 2000
    },
    {
      "epoch": 0.42,
      "mmlu_eval_accuracy": 0.4947166849858733,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "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": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
      "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": 0.27586206896551724,
      "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": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.6153846153846154,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.92,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "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": 0.48148148148148145,
      "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": 0.8421052631578947,
      "mmlu_loss": 1.3360929966906654,
      "step": 2000
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.6887,
      "step": 2010
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002,
      "loss": 0.7519,
      "step": 2020
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.7638,
      "step": 2030
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.7756,
      "step": 2040
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.7841,
      "step": 2050
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002,
      "loss": 0.7531,
      "step": 2060
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.7482,
      "step": 2070
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.7248,
      "step": 2080
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.7435,
      "step": 2090
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.7893,
      "step": 2100
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0002,
      "loss": 0.7783,
      "step": 2110
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.7628,
      "step": 2120
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.7156,
      "step": 2130
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.7376,
      "step": 2140
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.7321,
      "step": 2150
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0002,
      "loss": 0.8105,
      "step": 2160
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.801,
      "step": 2170
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.7729,
      "step": 2180
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.7929,
      "step": 2190
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.7701,
      "step": 2200
    },
    {
      "epoch": 0.46,
      "eval_loss": 0.7463650107383728,
      "eval_runtime": 68.307,
      "eval_samples_per_second": 14.64,
      "eval_steps_per_second": 7.32,
      "step": 2200
    },
    {
      "epoch": 0.46,
      "mmlu_eval_accuracy": 0.4908882886669515,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "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": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "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": 0.3125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "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": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.92,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.4074467631450833,
      "step": 2200
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002,
      "loss": 0.7942,
      "step": 2210
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7582,
      "step": 2220
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7207,
      "step": 2230
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7677,
      "step": 2240
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7617,
      "step": 2250
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002,
      "loss": 0.7722,
      "step": 2260
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.7898,
      "step": 2270
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.7669,
      "step": 2280
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.7539,
      "step": 2290
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0002,
      "loss": 0.7699,
      "step": 2300
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.782,
      "step": 2310
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.8049,
      "step": 2320
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.7646,
      "step": 2330
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.7589,
      "step": 2340
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0002,
      "loss": 0.7535,
      "step": 2350
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.6765,
      "step": 2360
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.7514,
      "step": 2370
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.7501,
      "step": 2380
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.7796,
      "step": 2390
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.7273,
      "step": 2400
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.7440771460533142,
      "eval_runtime": 68.807,
      "eval_samples_per_second": 14.533,
      "eval_steps_per_second": 7.267,
      "step": 2400
    },
    {
      "epoch": 0.5,
      "mmlu_eval_accuracy": 0.4687542096734408,
      "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": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "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": 0.7727272727272727,
      "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": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.31,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.166977745434634,
      "step": 2400
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7149,
      "step": 2410
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7221,
      "step": 2420
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7302,
      "step": 2430
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7229,
      "step": 2440
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002,
      "loss": 0.7638,
      "step": 2450
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.7229,
      "step": 2460
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.7439,
      "step": 2470
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.798,
      "step": 2480
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0002,
      "loss": 0.7164,
      "step": 2490
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7805,
      "step": 2500
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7655,
      "step": 2510
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7431,
      "step": 2520
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7858,
      "step": 2530
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0002,
      "loss": 0.7224,
      "step": 2540
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7378,
      "step": 2550
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7583,
      "step": 2560
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.6881,
      "step": 2570
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.7767,
      "step": 2580
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0002,
      "loss": 0.8014,
      "step": 2590
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.6781,
      "step": 2600
    },
    {
      "epoch": 0.55,
      "eval_loss": 0.7418100833892822,
      "eval_runtime": 68.452,
      "eval_samples_per_second": 14.609,
      "eval_steps_per_second": 7.304,
      "step": 2600
    },
    {
      "epoch": 0.55,
      "mmlu_eval_accuracy": 0.4912259056745587,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "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": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "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": 0.5555555555555556,
      "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": 0.2413793103448276,
      "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": 0.7692307692307693,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.1756191353651624,
      "step": 2600
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.7993,
      "step": 2610
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.7163,
      "step": 2620
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.7388,
      "step": 2630
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002,
      "loss": 0.7624,
      "step": 2640
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.8116,
      "step": 2650
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.7596,
      "step": 2660
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.7791,
      "step": 2670
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0002,
      "loss": 0.7477,
      "step": 2680
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.7417,
      "step": 2690
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.8245,
      "step": 2700
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.7862,
      "step": 2710
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.8016,
      "step": 2720
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0002,
      "loss": 0.752,
      "step": 2730
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.7339,
      "step": 2740
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.7639,
      "step": 2750
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.7765,
      "step": 2760
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.7844,
      "step": 2770
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0002,
      "loss": 0.8474,
      "step": 2780
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.7535,
      "step": 2790
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.7621,
      "step": 2800
    },
    {
      "epoch": 0.59,
      "eval_loss": 0.7420905828475952,
      "eval_runtime": 68.6133,
      "eval_samples_per_second": 14.574,
      "eval_steps_per_second": 7.287,
      "step": 2800
    },
    {
      "epoch": 0.59,
      "mmlu_eval_accuracy": 0.49914105602923814,
      "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": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.18181818181818182,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "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": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.187204834569839,
      "step": 2800
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.7477,
      "step": 2810
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.773,
      "step": 2820
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0002,
      "loss": 0.753,
      "step": 2830
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.7134,
      "step": 2840
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.7879,
      "step": 2850
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.8331,
      "step": 2860
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0002,
      "loss": 0.7552,
      "step": 2870
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.7215,
      "step": 2880
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.8461,
      "step": 2890
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.7281,
      "step": 2900
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.7571,
      "step": 2910
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0002,
      "loss": 0.7475,
      "step": 2920
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.753,
      "step": 2930
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.7219,
      "step": 2940
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.7309,
      "step": 2950
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.7802,
      "step": 2960
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0002,
      "loss": 0.7753,
      "step": 2970
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.7558,
      "step": 2980
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.7807,
      "step": 2990
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.7497,
      "step": 3000
    },
    {
      "epoch": 0.63,
      "eval_loss": 0.7386950850486755,
      "eval_runtime": 68.4662,
      "eval_samples_per_second": 14.606,
      "eval_steps_per_second": 7.303,
      "step": 3000
    },
    {
      "epoch": 0.63,
      "mmlu_eval_accuracy": 0.4985309813978678,
      "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": 0.4375,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "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": 0.7209302325581395,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "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": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.220315750522962,
      "step": 3000
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.6852,
      "step": 3010
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0002,
      "loss": 0.8119,
      "step": 3020
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.7423,
      "step": 3030
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.7891,
      "step": 3040
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.7448,
      "step": 3050
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002,
      "loss": 0.7504,
      "step": 3060
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7363,
      "step": 3070
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7909,
      "step": 3080
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7771,
      "step": 3090
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7287,
      "step": 3100
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0002,
      "loss": 0.7834,
      "step": 3110
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.7491,
      "step": 3120
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.7779,
      "step": 3130
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.7952,
      "step": 3140
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.7795,
      "step": 3150
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0002,
      "loss": 0.6761,
      "step": 3160
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.8223,
      "step": 3170
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.8088,
      "step": 3180
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.7314,
      "step": 3190
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.777,
      "step": 3200
    },
    {
      "epoch": 0.67,
      "eval_loss": 0.7370312213897705,
      "eval_runtime": 69.6561,
      "eval_samples_per_second": 14.356,
      "eval_steps_per_second": 7.178,
      "step": 3200
    },
    {
      "epoch": 0.67,
      "mmlu_eval_accuracy": 0.49695078178783586,
      "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": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.18181818181818182,
      "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": 0.5,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "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": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "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": 0.5217391304347826,
      "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": 0.5,
      "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": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "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": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.0152867997828725,
      "step": 3200
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002,
      "loss": 0.7559,
      "step": 3210
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.743,
      "step": 3220
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.7068,
      "step": 3230
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.7715,
      "step": 3240
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0002,
      "loss": 0.7383,
      "step": 3250
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7618,
      "step": 3260
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7182,
      "step": 3270
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7465,
      "step": 3280
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7275,
      "step": 3290
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0002,
      "loss": 0.7232,
      "step": 3300
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7317,
      "step": 3310
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7468,
      "step": 3320
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7292,
      "step": 3330
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7183,
      "step": 3340
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0002,
      "loss": 0.7464,
      "step": 3350
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.8278,
      "step": 3360
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.7716,
      "step": 3370
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.7769,
      "step": 3380
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.7401,
      "step": 3390
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002,
      "loss": 0.736,
      "step": 3400
    },
    {
      "epoch": 0.71,
      "eval_loss": 0.7355515956878662,
      "eval_runtime": 68.3178,
      "eval_samples_per_second": 14.637,
      "eval_steps_per_second": 7.319,
      "step": 3400
    },
    {
      "epoch": 0.71,
      "mmlu_eval_accuracy": 0.4917868745651109,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.18181818181818182,
      "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": 0.5,
      "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": 0.375,
      "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": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "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": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "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": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.1342425108267185,
      "step": 3400
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.7237,
      "step": 3410
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.7239,
      "step": 3420
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.7474,
      "step": 3430
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0002,
      "loss": 0.6993,
      "step": 3440
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.7271,
      "step": 3450
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.7153,
      "step": 3460
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.7559,
      "step": 3470
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.7884,
      "step": 3480
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0002,
      "loss": 0.7234,
      "step": 3490
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.7523,
      "step": 3500
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.7698,
      "step": 3510
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.6896,
      "step": 3520
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.7721,
      "step": 3530
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0002,
      "loss": 0.7825,
      "step": 3540
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7049,
      "step": 3550
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7407,
      "step": 3560
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.6962,
      "step": 3570
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7708,
      "step": 3580
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.7487,
      "step": 3590
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.749,
      "step": 3600
    },
    {
      "epoch": 0.76,
      "eval_loss": 0.7301719188690186,
      "eval_runtime": 68.3192,
      "eval_samples_per_second": 14.637,
      "eval_steps_per_second": 7.319,
      "step": 3600
    },
    {
      "epoch": 0.76,
      "mmlu_eval_accuracy": 0.49994964792517205,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "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": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "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": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "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": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.5454545454545454,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.1752422991994156,
      "step": 3600
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7499,
      "step": 3610
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7554,
      "step": 3620
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0002,
      "loss": 0.7251,
      "step": 3630
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7855,
      "step": 3640
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.6997,
      "step": 3650
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.823,
      "step": 3660
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7773,
      "step": 3670
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0002,
      "loss": 0.7832,
      "step": 3680
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.7751,
      "step": 3690
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.6956,
      "step": 3700
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.6991,
      "step": 3710
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.7148,
      "step": 3720
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0002,
      "loss": 0.7846,
      "step": 3730
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7343,
      "step": 3740
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7888,
      "step": 3750
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7191,
      "step": 3760
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7248,
      "step": 3770
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0002,
      "loss": 0.7719,
      "step": 3780
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7563,
      "step": 3790
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.718,
      "step": 3800
    },
    {
      "epoch": 0.8,
      "eval_loss": 0.7296046018600464,
      "eval_runtime": 68.2783,
      "eval_samples_per_second": 14.646,
      "eval_steps_per_second": 7.323,
      "step": 3800
    },
    {
      "epoch": 0.8,
      "mmlu_eval_accuracy": 0.4966949296455417,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "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": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "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": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "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": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "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": 0.7391304347826086,
      "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": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.7093023255813954,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "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": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1436357539409756,
      "step": 3800
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7208,
      "step": 3810
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7545,
      "step": 3820
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002,
      "loss": 0.7129,
      "step": 3830
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.7213,
      "step": 3840
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.7133,
      "step": 3850
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.7585,
      "step": 3860
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002,
      "loss": 0.7214,
      "step": 3870
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7562,
      "step": 3880
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7784,
      "step": 3890
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7414,
      "step": 3900
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.739,
      "step": 3910
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0002,
      "loss": 0.7614,
      "step": 3920
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7796,
      "step": 3930
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7426,
      "step": 3940
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.6766,
      "step": 3950
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7508,
      "step": 3960
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002,
      "loss": 0.7139,
      "step": 3970
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.7233,
      "step": 3980
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.8255,
      "step": 3990
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.7567,
      "step": 4000
    },
    {
      "epoch": 0.84,
      "eval_loss": 0.7322874665260315,
      "eval_runtime": 68.7066,
      "eval_samples_per_second": 14.555,
      "eval_steps_per_second": 7.277,
      "step": 4000
    },
    {
      "epoch": 0.84,
      "mmlu_eval_accuracy": 0.49341287103342835,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "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": 0.25,
      "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": 0.3,
      "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": 0.5555555555555556,
      "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": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "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": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.181115123956384,
      "step": 4000
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.6492,
      "step": 4010
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0002,
      "loss": 0.7133,
      "step": 4020
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.7843,
      "step": 4030
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.7574,
      "step": 4040
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.6927,
      "step": 4050
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002,
      "loss": 0.781,
      "step": 4060
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.7418,
      "step": 4070
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.7751,
      "step": 4080
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.7064,
      "step": 4090
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.7273,
      "step": 4100
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0002,
      "loss": 0.6985,
      "step": 4110
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.8219,
      "step": 4120
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.7222,
      "step": 4130
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.7343,
      "step": 4140
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.7616,
      "step": 4150
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002,
      "loss": 0.7021,
      "step": 4160
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7408,
      "step": 4170
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7244,
      "step": 4180
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7527,
      "step": 4190
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.8307,
      "step": 4200
    },
    {
      "epoch": 0.88,
      "eval_loss": 0.7286710143089294,
      "eval_runtime": 68.5382,
      "eval_samples_per_second": 14.59,
      "eval_steps_per_second": 7.295,
      "step": 4200
    },
    {
      "epoch": 0.88,
      "mmlu_eval_accuracy": 0.48789929357470085,
      "mmlu_eval_accuracy_abstract_algebra": 0.5454545454545454,
      "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "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": 0.45454545454545453,
      "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": 0.36585365853658536,
      "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": 0.13636363636363635,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
      "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": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1633820450881753,
      "step": 4200
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0002,
      "loss": 0.7526,
      "step": 4210
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7527,
      "step": 4220
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7768,
      "step": 4230
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7082,
      "step": 4240
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0002,
      "loss": 0.7187,
      "step": 4250
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.7434,
      "step": 4260
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.7279,
      "step": 4270
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.7039,
      "step": 4280
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.8235,
      "step": 4290
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0002,
      "loss": 0.7082,
      "step": 4300
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7735,
      "step": 4310
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7585,
      "step": 4320
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7315,
      "step": 4330
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7654,
      "step": 4340
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002,
      "loss": 0.7662,
      "step": 4350
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7443,
      "step": 4360
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7114,
      "step": 4370
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7486,
      "step": 4380
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.8516,
      "step": 4390
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002,
      "loss": 0.7194,
      "step": 4400
    },
    {
      "epoch": 0.92,
      "eval_loss": 0.727415919303894,
      "eval_runtime": 68.5594,
      "eval_samples_per_second": 14.586,
      "eval_steps_per_second": 7.293,
      "step": 4400
    },
    {
      "epoch": 0.92,
      "mmlu_eval_accuracy": 0.4721171345219018,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.18181818181818182,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.13636363636363635,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "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": 0.627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "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": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.032352588623373,
      "step": 4400
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7648,
      "step": 4410
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7777,
      "step": 4420
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.783,
      "step": 4430
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002,
      "loss": 0.7583,
      "step": 4440
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7611,
      "step": 4450
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7573,
      "step": 4460
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7341,
      "step": 4470
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7138,
      "step": 4480
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002,
      "loss": 0.7643,
      "step": 4490
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.7093,
      "step": 4500
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.7204,
      "step": 4510
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.7699,
      "step": 4520
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.7058,
      "step": 4530
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002,
      "loss": 0.7258,
      "step": 4540
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.7174,
      "step": 4550
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.7622,
      "step": 4560
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.7799,
      "step": 4570
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.7048,
      "step": 4580
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002,
      "loss": 0.8384,
      "step": 4590
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.6833,
      "step": 4600
    },
    {
      "epoch": 0.97,
      "eval_loss": 0.726478636264801,
      "eval_runtime": 68.7096,
      "eval_samples_per_second": 14.554,
      "eval_steps_per_second": 7.277,
      "step": 4600
    },
    {
      "epoch": 0.97,
      "mmlu_eval_accuracy": 0.48058763626721557,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.18181818181818182,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.6176470588235294,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.4927536231884058,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.1677384916936138,
      "step": 4600
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.8717,
      "step": 4610
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.7288,
      "step": 4620
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002,
      "loss": 0.7779,
      "step": 4630
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.7677,
      "step": 4640
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.6299,
      "step": 4650
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.8307,
      "step": 4660
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.8258,
      "step": 4670
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002,
      "loss": 0.6958,
      "step": 4680
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7695,
      "step": 4690
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7034,
      "step": 4700
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7368,
      "step": 4710
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7405,
      "step": 4720
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002,
      "loss": 0.7453,
      "step": 4730
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.7834,
      "step": 4740
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.7321,
      "step": 4750
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.7151,
      "step": 4760
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.7174,
      "step": 4770
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.6821,
      "step": 4780
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.6354,
      "step": 4790
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.6674,
      "step": 4800
    },
    {
      "epoch": 1.01,
      "eval_loss": 0.7293053269386292,
      "eval_runtime": 68.8076,
      "eval_samples_per_second": 14.533,
      "eval_steps_per_second": 7.267,
      "step": 4800
    },
    {
      "epoch": 1.01,
      "mmlu_eval_accuracy": 0.49507036825080164,
      "mmlu_eval_accuracy_abstract_algebra": 0.5454545454545454,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
      "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": 0.2926829268292683,
      "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": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "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": 0.30434782608695654,
      "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": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.182025539423716,
      "step": 4800
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.676,
      "step": 4810
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002,
      "loss": 0.7102,
      "step": 4820
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.6785,
      "step": 4830
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.6647,
      "step": 4840
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.7103,
      "step": 4850
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.6349,
      "step": 4860
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002,
      "loss": 0.6633,
      "step": 4870
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.7152,
      "step": 4880
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.6288,
      "step": 4890
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.65,
      "step": 4900
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.6463,
      "step": 4910
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002,
      "loss": 0.7215,
      "step": 4920
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.6978,
      "step": 4930
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.6607,
      "step": 4940
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.6723,
      "step": 4950
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.6597,
      "step": 4960
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002,
      "loss": 0.6398,
      "step": 4970
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.6516,
      "step": 4980
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.6944,
      "step": 4990
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.6111,
      "step": 5000
    },
    {
      "epoch": 1.05,
      "eval_loss": 0.7314338088035583,
      "eval_runtime": 68.873,
      "eval_samples_per_second": 14.519,
      "eval_steps_per_second": 7.26,
      "step": 5000
    },
    {
      "epoch": 1.05,
      "mmlu_eval_accuracy": 0.5150028793145787,
      "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.3333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "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": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.5,
      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.7093023255813954,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.1535632668680684,
      "step": 5000
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002,
      "loss": 0.6553,
      "step": 5010
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.6758,
      "step": 5020
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.7309,
      "step": 5030
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.6594,
      "step": 5040
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.7451,
      "step": 5050
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002,
      "loss": 0.6627,
      "step": 5060
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.6498,
      "step": 5070
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.6743,
      "step": 5080
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.6609,
      "step": 5090
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.7177,
      "step": 5100
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002,
      "loss": 0.6704,
      "step": 5110
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.7194,
      "step": 5120
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.7,
      "step": 5130
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.6061,
      "step": 5140
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.6658,
      "step": 5150
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002,
      "loss": 0.6595,
      "step": 5160
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.6354,
      "step": 5170
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.6885,
      "step": 5180
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.6428,
      "step": 5190
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.6798,
      "step": 5200
    },
    {
      "epoch": 1.09,
      "eval_loss": 0.7294297218322754,
      "eval_runtime": 68.5437,
      "eval_samples_per_second": 14.589,
      "eval_steps_per_second": 7.295,
      "step": 5200
    },
    {
      "epoch": 1.09,
      "mmlu_eval_accuracy": 0.4950820370235212,
      "mmlu_eval_accuracy_abstract_algebra": 0.5454545454545454,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.09090909090909091,
      "mmlu_eval_accuracy_college_medicine": 0.5454545454545454,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "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": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "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": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.21,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.36470588235294116,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "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": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.2110077123651306,
      "step": 5200
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002,
      "loss": 0.692,
      "step": 5210
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.6375,
      "step": 5220
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.6255,
      "step": 5230
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.7116,
      "step": 5240
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002,
      "loss": 0.6671,
      "step": 5250
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.6606,
      "step": 5260
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.6779,
      "step": 5270
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.6881,
      "step": 5280
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.6726,
      "step": 5290
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0002,
      "loss": 0.6664,
      "step": 5300
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6281,
      "step": 5310
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6438,
      "step": 5320
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6577,
      "step": 5330
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6926,
      "step": 5340
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002,
      "loss": 0.6395,
      "step": 5350
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.6935,
      "step": 5360
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.6912,
      "step": 5370
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.5957,
      "step": 5380
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.7,
      "step": 5390
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002,
      "loss": 0.671,
      "step": 5400
    },
    {
      "epoch": 1.13,
      "eval_loss": 0.7315531969070435,
      "eval_runtime": 68.2697,
      "eval_samples_per_second": 14.648,
      "eval_steps_per_second": 7.324,
      "step": 5400
    },
    {
      "epoch": 1.13,
      "mmlu_eval_accuracy": 0.4863032195722604,
      "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": 0.5862068965517241,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "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": 0.4,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "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": 0.5909090909090909,
      "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": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.7272727272727273,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.463768115942029,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.0932272243126253,
      "step": 5400
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.667,
      "step": 5410
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.6323,
      "step": 5420
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.7363,
      "step": 5430
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002,
      "loss": 0.6725,
      "step": 5440
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.6841,
      "step": 5450
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.6762,
      "step": 5460
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.6889,
      "step": 5470
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.5807,
      "step": 5480
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002,
      "loss": 0.6927,
      "step": 5490
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.6158,
      "step": 5500
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.6813,
      "step": 5510
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.6609,
      "step": 5520
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.6817,
      "step": 5530
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0002,
      "loss": 0.6494,
      "step": 5540
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.6779,
      "step": 5550
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.7426,
      "step": 5560
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.7299,
      "step": 5570
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.6463,
      "step": 5580
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0002,
      "loss": 0.6247,
      "step": 5590
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.6909,
      "step": 5600
    },
    {
      "epoch": 1.18,
      "eval_loss": 0.7314475178718567,
      "eval_runtime": 68.6796,
      "eval_samples_per_second": 14.56,
      "eval_steps_per_second": 7.28,
      "step": 5600
    },
    {
      "epoch": 1.18,
      "mmlu_eval_accuracy": 0.4993167974782687,
      "mmlu_eval_accuracy_abstract_algebra": 0.5454545454545454,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "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": 0.09090909090909091,
      "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": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.5,
      "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": 0.8181818181818182,
      "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": 0.38461538461538464,
      "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": 0.5454545454545454,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3764705882352941,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "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": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.0935832330514494,
      "step": 5600
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.7122,
      "step": 5610
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.6432,
      "step": 5620
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0002,
      "loss": 0.643,
      "step": 5630
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6434,
      "step": 5640
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6964,
      "step": 5650
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6695,
      "step": 5660
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6788,
      "step": 5670
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0002,
      "loss": 0.6747,
      "step": 5680
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.6737,
      "step": 5690
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.6911,
      "step": 5700
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.6305,
      "step": 5710
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.6244,
      "step": 5720
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0002,
      "loss": 0.6454,
      "step": 5730
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.6063,
      "step": 5740
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.6425,
      "step": 5750
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.6865,
      "step": 5760
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.6819,
      "step": 5770
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0002,
      "loss": 0.7132,
      "step": 5780
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.6937,
      "step": 5790
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.6646,
      "step": 5800
    },
    {
      "epoch": 1.22,
      "eval_loss": 0.7308344841003418,
      "eval_runtime": 68.9985,
      "eval_samples_per_second": 14.493,
      "eval_steps_per_second": 7.247,
      "step": 5800
    },
    {
      "epoch": 1.22,
      "mmlu_eval_accuracy": 0.4891768767603638,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "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": 0.45454545454545453,
      "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": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "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": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "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": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.1318859975036069,
      "step": 5800
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.636,
      "step": 5810
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0002,
      "loss": 0.6791,
      "step": 5820
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.6103,
      "step": 5830
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.6495,
      "step": 5840
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.6703,
      "step": 5850
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.6721,
      "step": 5860
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0002,
      "loss": 0.7038,
      "step": 5870
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.6557,
      "step": 5880
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.7128,
      "step": 5890
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.6581,
      "step": 5900
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.6406,
      "step": 5910
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0002,
      "loss": 0.6997,
      "step": 5920
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.6823,
      "step": 5930
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.7109,
      "step": 5940
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.7779,
      "step": 5950
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.7092,
      "step": 5960
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.6788,
      "step": 5970
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.6307,
      "step": 5980
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.6547,
      "step": 5990
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.6924,
      "step": 6000
    },
    {
      "epoch": 1.26,
      "eval_loss": 0.7297111749649048,
      "eval_runtime": 68.7437,
      "eval_samples_per_second": 14.547,
      "eval_steps_per_second": 7.273,
      "step": 6000
    },
    {
      "epoch": 1.26,
      "mmlu_eval_accuracy": 0.4915754992456156,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "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": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "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": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "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": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.6176470588235294,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "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": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.1507751187868902,
      "step": 6000
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0002,
      "loss": 0.6461,
      "step": 6010
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.6775,
      "step": 6020
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.6672,
      "step": 6030
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.702,
      "step": 6040
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.6896,
      "step": 6050
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002,
      "loss": 0.6629,
      "step": 6060
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.7265,
      "step": 6070
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.652,
      "step": 6080
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.6366,
      "step": 6090
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.679,
      "step": 6100
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0002,
      "loss": 0.6509,
      "step": 6110
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.7165,
      "step": 6120
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.668,
      "step": 6130
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.6805,
      "step": 6140
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.6402,
      "step": 6150
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0002,
      "loss": 0.6672,
      "step": 6160
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.6282,
      "step": 6170
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.7047,
      "step": 6180
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.6396,
      "step": 6190
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0002,
      "loss": 0.6649,
      "step": 6200
    },
    {
      "epoch": 1.3,
      "eval_loss": 0.7309466600418091,
      "eval_runtime": 68.6834,
      "eval_samples_per_second": 14.56,
      "eval_steps_per_second": 7.28,
      "step": 6200
    },
    {
      "epoch": 1.3,
      "mmlu_eval_accuracy": 0.49088395397340523,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "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": 0.18181818181818182,
      "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": 0.34615384615384615,
      "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": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "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": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.5,
      "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": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.33,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.0723085504793624,
      "step": 6200
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.7416,
      "step": 6210
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.6194,
      "step": 6220
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.5743,
      "step": 6230
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.7051,
      "step": 6240
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0002,
      "loss": 0.6916,
      "step": 6250
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.6563,
      "step": 6260
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.6698,
      "step": 6270
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.624,
      "step": 6280
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.6657,
      "step": 6290
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0002,
      "loss": 0.667,
      "step": 6300
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.6484,
      "step": 6310
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.6839,
      "step": 6320
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.6612,
      "step": 6330
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.686,
      "step": 6340
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0002,
      "loss": 0.6924,
      "step": 6350
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.6502,
      "step": 6360
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.6912,
      "step": 6370
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.7249,
      "step": 6380
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0002,
      "loss": 0.6714,
      "step": 6390
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.6447,
      "step": 6400
    },
    {
      "epoch": 1.35,
      "eval_loss": 0.7249250411987305,
      "eval_runtime": 68.525,
      "eval_samples_per_second": 14.593,
      "eval_steps_per_second": 7.297,
      "step": 6400
    },
    {
      "epoch": 1.35,
      "mmlu_eval_accuracy": 0.48839285984652686,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
      "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": 0.2727272727272727,
      "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": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "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": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "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": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.6470588235294118,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.066172287840756,
      "step": 6400
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.6692,
      "step": 6410
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.6763,
      "step": 6420
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.7303,
      "step": 6430
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0002,
      "loss": 0.679,
      "step": 6440
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.7092,
      "step": 6450
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.6678,
      "step": 6460
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.6672,
      "step": 6470
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.7287,
      "step": 6480
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0002,
      "loss": 0.647,
      "step": 6490
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.7539,
      "step": 6500
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.7182,
      "step": 6510
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.7114,
      "step": 6520
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.6535,
      "step": 6530
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0002,
      "loss": 0.5957,
      "step": 6540
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.6658,
      "step": 6550
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.7061,
      "step": 6560
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.6461,
      "step": 6570
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0002,
      "loss": 0.6066,
      "step": 6580
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.6782,
      "step": 6590
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.6702,
      "step": 6600
    },
    {
      "epoch": 1.39,
      "eval_loss": 0.7293462157249451,
      "eval_runtime": 68.5486,
      "eval_samples_per_second": 14.588,
      "eval_steps_per_second": 7.294,
      "step": 6600
    },
    {
      "epoch": 1.39,
      "mmlu_eval_accuracy": 0.48383698013409976,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "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": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.5454545454545454,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
      "mmlu_loss": 1.0967509123659631,
      "step": 6600
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.6836,
      "step": 6610
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.693,
      "step": 6620
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0002,
      "loss": 0.6858,
      "step": 6630
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.6364,
      "step": 6640
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.6714,
      "step": 6650
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.6115,
      "step": 6660
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.6549,
      "step": 6670
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0002,
      "loss": 0.6451,
      "step": 6680
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.6558,
      "step": 6690
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.68,
      "step": 6700
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.683,
      "step": 6710
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.6502,
      "step": 6720
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0002,
      "loss": 0.6376,
      "step": 6730
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.6441,
      "step": 6740
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.6165,
      "step": 6750
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.6539,
      "step": 6760
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.654,
      "step": 6770
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0002,
      "loss": 0.692,
      "step": 6780
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.6395,
      "step": 6790
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.7853,
      "step": 6800
    },
    {
      "epoch": 1.43,
      "eval_loss": 0.727479100227356,
      "eval_runtime": 69.0526,
      "eval_samples_per_second": 14.482,
      "eval_steps_per_second": 7.241,
      "step": 6800
    },
    {
      "epoch": 1.43,
      "mmlu_eval_accuracy": 0.4903361112929509,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "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": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "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": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.8,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6153846153846154,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.88,
      "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": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "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": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.110004111385221,
      "step": 6800
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.6608,
      "step": 6810
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0002,
      "loss": 0.6936,
      "step": 6820
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.7166,
      "step": 6830
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.6418,
      "step": 6840
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.6742,
      "step": 6850
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.6629,
      "step": 6860
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0002,
      "loss": 0.6409,
      "step": 6870
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.6627,
      "step": 6880
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.7074,
      "step": 6890
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.7161,
      "step": 6900
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.6387,
      "step": 6910
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0002,
      "loss": 0.6847,
      "step": 6920
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.7103,
      "step": 6930
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.6514,
      "step": 6940
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.7131,
      "step": 6950
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.7141,
      "step": 6960
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0002,
      "loss": 0.6462,
      "step": 6970
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.6674,
      "step": 6980
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.668,
      "step": 6990
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.6397,
      "step": 7000
    },
    {
      "epoch": 1.47,
      "eval_loss": 0.7244565486907959,
      "eval_runtime": 68.6993,
      "eval_samples_per_second": 14.556,
      "eval_steps_per_second": 7.278,
      "step": 7000
    },
    {
      "epoch": 1.47,
      "mmlu_eval_accuracy": 0.49890965953062405,
      "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": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.40625,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.5217391304347826,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "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": 0.7894736842105263,
      "mmlu_loss": 1.215826900291692,
      "step": 7000
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0002,
      "loss": 0.6485,
      "step": 7010
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.6665,
      "step": 7020
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.7524,
      "step": 7030
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.6048,
      "step": 7040
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.6882,
      "step": 7050
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0002,
      "loss": 0.7154,
      "step": 7060
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6451,
      "step": 7070
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6682,
      "step": 7080
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.676,
      "step": 7090
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6611,
      "step": 7100
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0002,
      "loss": 0.6424,
      "step": 7110
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.6539,
      "step": 7120
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.7308,
      "step": 7130
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.6522,
      "step": 7140
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.6578,
      "step": 7150
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0002,
      "loss": 0.7486,
      "step": 7160
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.6696,
      "step": 7170
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.6158,
      "step": 7180
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.654,
      "step": 7190
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0002,
      "loss": 0.6651,
      "step": 7200
    },
    {
      "epoch": 1.51,
      "eval_loss": 0.7256432175636292,
      "eval_runtime": 68.7275,
      "eval_samples_per_second": 14.55,
      "eval_steps_per_second": 7.275,
      "step": 7200
    },
    {
      "epoch": 1.51,
      "mmlu_eval_accuracy": 0.5042163217177772,
      "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": 0.18181818181818182,
      "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": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.6470588235294118,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.4782608695652174,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.2178749454161517,
      "step": 7200
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.6572,
      "step": 7210
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.7088,
      "step": 7220
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.6629,
      "step": 7230
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.7313,
      "step": 7240
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002,
      "loss": 0.6697,
      "step": 7250
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.651,
      "step": 7260
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.659,
      "step": 7270
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.6818,
      "step": 7280
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.6666,
      "step": 7290
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0002,
      "loss": 0.6644,
      "step": 7300
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.6221,
      "step": 7310
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.6159,
      "step": 7320
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.6928,
      "step": 7330
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.7239,
      "step": 7340
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0002,
      "loss": 0.6607,
      "step": 7350
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.6846,
      "step": 7360
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.683,
      "step": 7370
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.6558,
      "step": 7380
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002,
      "loss": 0.6602,
      "step": 7390
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.716,
      "step": 7400
    },
    {
      "epoch": 1.56,
      "eval_loss": 0.7221587896347046,
      "eval_runtime": 68.7417,
      "eval_samples_per_second": 14.547,
      "eval_steps_per_second": 7.274,
      "step": 7400
    },
    {
      "epoch": 1.56,
      "mmlu_eval_accuracy": 0.4963567622571148,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
      "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": 0.18181818181818182,
      "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": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "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": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "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": 0.36363636363636365,
      "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": 0.6976744186046512,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.6176470588235294,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.1572000444246646,
      "step": 7400
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.658,
      "step": 7410
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.689,
      "step": 7420
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.7234,
      "step": 7430
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002,
      "loss": 0.6593,
      "step": 7440
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.6837,
      "step": 7450
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.6942,
      "step": 7460
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.6963,
      "step": 7470
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.6399,
      "step": 7480
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0002,
      "loss": 0.6236,
      "step": 7490
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.7154,
      "step": 7500
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.6572,
      "step": 7510
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.6635,
      "step": 7520
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.7151,
      "step": 7530
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002,
      "loss": 0.7181,
      "step": 7540
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.6754,
      "step": 7550
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.7071,
      "step": 7560
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.6624,
      "step": 7570
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002,
      "loss": 0.6702,
      "step": 7580
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.6467,
      "step": 7590
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.6818,
      "step": 7600
    },
    {
      "epoch": 1.6,
      "eval_loss": 0.719873309135437,
      "eval_runtime": 68.54,
      "eval_samples_per_second": 14.59,
      "eval_steps_per_second": 7.295,
      "step": 7600
    },
    {
      "epoch": 1.6,
      "mmlu_eval_accuracy": 0.4938915553061816,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "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": 0.85,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.142687739497999,
      "step": 7600
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.675,
      "step": 7610
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.7186,
      "step": 7620
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0002,
      "loss": 0.7125,
      "step": 7630
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.5775,
      "step": 7640
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.6777,
      "step": 7650
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.6238,
      "step": 7660
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.6249,
      "step": 7670
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0002,
      "loss": 0.6576,
      "step": 7680
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.62,
      "step": 7690
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.6442,
      "step": 7700
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.6802,
      "step": 7710
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.6526,
      "step": 7720
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002,
      "loss": 0.606,
      "step": 7730
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.6409,
      "step": 7740
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.7021,
      "step": 7750
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.6751,
      "step": 7760
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0002,
      "loss": 0.703,
      "step": 7770
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.6279,
      "step": 7780
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.6784,
      "step": 7790
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.6603,
      "step": 7800
    },
    {
      "epoch": 1.64,
      "eval_loss": 0.7202572226524353,
      "eval_runtime": 68.9286,
      "eval_samples_per_second": 14.508,
      "eval_steps_per_second": 7.254,
      "step": 7800
    },
    {
      "epoch": 1.64,
      "mmlu_eval_accuracy": 0.49506372934866344,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "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": 0.46875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "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": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "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": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.84,
      "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": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.5072463768115942,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "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": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.191394856214212,
      "step": 7800
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.7018,
      "step": 7810
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0002,
      "loss": 0.6855,
      "step": 7820
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.6327,
      "step": 7830
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.6578,
      "step": 7840
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.6678,
      "step": 7850
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.67,
      "step": 7860
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002,
      "loss": 0.6898,
      "step": 7870
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.6412,
      "step": 7880
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.6896,
      "step": 7890
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.652,
      "step": 7900
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.5984,
      "step": 7910
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002,
      "loss": 0.6559,
      "step": 7920
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.6171,
      "step": 7930
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.7222,
      "step": 7940
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.6233,
      "step": 7950
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0002,
      "loss": 0.6479,
      "step": 7960
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.6898,
      "step": 7970
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.6229,
      "step": 7980
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.6232,
      "step": 7990
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.6723,
      "step": 8000
    },
    {
      "epoch": 1.68,
      "eval_loss": 0.7203255295753479,
      "eval_runtime": 68.5826,
      "eval_samples_per_second": 14.581,
      "eval_steps_per_second": 7.29,
      "step": 8000
    },
    {
      "epoch": 1.68,
      "mmlu_eval_accuracy": 0.5020970296577588,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.5,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "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": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "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": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "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": 0.84,
      "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": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5588235294117647,
      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.5362318840579711,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.179802088762388,
      "step": 8000
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002,
      "loss": 0.7006,
      "step": 8010
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.6832,
      "step": 8020
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.6369,
      "step": 8030
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.6072,
      "step": 8040
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.7377,
      "step": 8050
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002,
      "loss": 0.665,
      "step": 8060
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.7071,
      "step": 8070
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.6955,
      "step": 8080
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.6439,
      "step": 8090
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.6282,
      "step": 8100
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002,
      "loss": 0.6891,
      "step": 8110
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.6347,
      "step": 8120
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.6353,
      "step": 8130
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.6922,
      "step": 8140
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0002,
      "loss": 0.6889,
      "step": 8150
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.6686,
      "step": 8160
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.6963,
      "step": 8170
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.6383,
      "step": 8180
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.6832,
      "step": 8190
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0002,
      "loss": 0.6876,
      "step": 8200
    },
    {
      "epoch": 1.72,
      "eval_loss": 0.718351423740387,
      "eval_runtime": 68.543,
      "eval_samples_per_second": 14.589,
      "eval_steps_per_second": 7.295,
      "step": 8200
    },
    {
      "epoch": 1.72,
      "mmlu_eval_accuracy": 0.4837673768983555,
      "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": 0.4375,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "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": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.4375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
      "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": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
      "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": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.5362318840579711,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.1326499803103605,
      "step": 8200
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.6358,
      "step": 8210
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.6971,
      "step": 8220
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.659,
      "step": 8230
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.6953,
      "step": 8240
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002,
      "loss": 0.6798,
      "step": 8250
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.7388,
      "step": 8260
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.6326,
      "step": 8270
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.6421,
      "step": 8280
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.7034,
      "step": 8290
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0002,
      "loss": 0.6446,
      "step": 8300
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.6453,
      "step": 8310
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.6467,
      "step": 8320
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.682,
      "step": 8330
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.6742,
      "step": 8340
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002,
      "loss": 0.6804,
      "step": 8350
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.6372,
      "step": 8360
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.6415,
      "step": 8370
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.6883,
      "step": 8380
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0002,
      "loss": 0.6623,
      "step": 8390
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.6615,
      "step": 8400
    },
    {
      "epoch": 1.77,
      "eval_loss": 0.7216736078262329,
      "eval_runtime": 69.5248,
      "eval_samples_per_second": 14.383,
      "eval_steps_per_second": 7.192,
      "step": 8400
    },
    {
      "epoch": 1.77,
      "mmlu_eval_accuracy": 0.4797165293839395,
      "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": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "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": 0.8636363636363636,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.6363636363636364,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.34,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.5652173913043478,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.8181818181818182,
      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
      "mmlu_loss": 1.0919515004007063,
      "step": 8400
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.582,
      "step": 8410
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.6839,
      "step": 8420
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.6713,
      "step": 8430
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0002,
      "loss": 0.7167,
      "step": 8440
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.5882,
      "step": 8450
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.6557,
      "step": 8460
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.6893,
      "step": 8470
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.6681,
      "step": 8480
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0002,
      "loss": 0.6147,
      "step": 8490
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.698,
      "step": 8500
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.6464,
      "step": 8510
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.7075,
      "step": 8520
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.6394,
      "step": 8530
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002,
      "loss": 0.6622,
      "step": 8540
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.7166,
      "step": 8550
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.6544,
      "step": 8560
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.6436,
      "step": 8570
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0002,
      "loss": 0.688,
      "step": 8580
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.6432,
      "step": 8590
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0002,
      "loss": 0.6629,
      "step": 8600
    },
    {
      "epoch": 1.81,
      "eval_loss": 0.7169492244720459,
      "eval_runtime": 68.6026,
      "eval_samples_per_second": 14.577,
      "eval_steps_per_second": 7.288,
      "step": 8600
    },
    {
      "epoch": 1.81,
      "mmlu_eval_accuracy": 0.48553671569218915,
      "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": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "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": 0.4146341463414634,
      "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": 0.2727272727272727,
      "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": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
      "mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "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": 0.6395348837209303,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.33,
      "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
      "mmlu_eval_accuracy_philosophy": 0.6176470588235294,
      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.5362318840579711,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.8181818181818182,
      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
      "mmlu_eval_accuracy_virology": 0.6111111111111112,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.1197648563716494,
      "step": 8600
    }
  ],
  "max_steps": 10000,
  "num_train_epochs": 3,
  "total_flos": 1.3731136853106524e+18,
  "trial_name": null,
  "trial_params": null
}